Bartus, RT, Baumann, TL, Brown, L, Kruegel, BR, Ostrove, JM and Herzog, CD. Advancing neurotrophic factors as treatments for age-related neurodegenerative diseases: developing and demonstrating “clinical proof-of-concept” for AAV-neurturin (CERE-120) in Parkinson’s disease. Neurobiol Aging 34: 35-61

Ceregene, Inc., San Diego, CA, USA.
Neurobiology of aging (Impact Factor: 5.01). 08/2012; 34(1). DOI: 10.1016/j.neurobiolaging.2012.07.018
Source: PubMed


Neurotrophic factors have long shown promise as potential therapies for age-related neurodegenerative diseases. However, 20 years of largely disappointing clinical results have underscored the difficulties involved with safely and effectively delivering these proteins to targeted sites within the central nervous system. Recent progress establishes that gene transfer can now likely overcome the delivery issues plaguing the translation of neurotrophic factors. This may be best exemplified by adeno-associated virus serotype-2-neurturin (CERE-120), a viral-vector construct designed to deliver the neurotrophic factor, neurturin to degenerating nigrostriatal neurons in Parkinson's disease. Eighty Parkinson's subjects have been dosed with CERE-120 (some 7+ years ago), with long-term, targeted neurturin expression confirmed and no serious safety issues identified. A double-blind, controlled Phase 2a trial established clinical "proof-of-concept" via 19 of the 24 prescribed efficacy end points favoring CERE-120 at the 12-month protocol-prescribed time point and all but one favoring CERE-120 at the 18-month secondary time point (p = 0.007 and 0.001, respectively). Moreover, clinically meaningful benefit was seen with CERE-120 on several specific protocol-prescribed, pairwise, blinded, motor, and quality-of-life end points at 12 months, and an even greater number of end points at 18 months. Because the trial failed to meet the primary end point (Unified Parkinson's Disease Rating Scale motor-off, measured at 12 months), a revised multicenter Phase 1/2b protocol was designed to enhance the neurotrophic effects of CERE-120, using insight gained from the Phase 2a trial. This review summarizes the development of CERE-120 from its inception through establishing "clinical proof-of-concept" and beyond. The translational obstacles and issues confronted, and the strategies applied, are reviewed. This information should be informative to investigators interested in translational research and development for age-related and other neurodegenerative diseases.


Available from: Raymond T Bartus
Advancing neurotrophic factors as treatments for age-related
neurodegenerative diseases: developing and demonstrating “clinical
proof-of-concept” for AAV-neurturin (CERE-120) in
Parkinson’s disease
Raymond T. Bartus*, Tiffany L. Baumann, Lamar Brown, Brian R. Kruegel,
Jeffrey M. Ostrove, Christopher D. Herzog
Ceregene, Inc., San Diego, CA, USA
Received 23 June 2012; received in revised form 26 July 2012; accepted 29 July 2012
Neurotrophic factors have long shown promise as potential therapies for age-related neurodegenerative diseases. However, 20 years of
largely disappointing clinical results have underscored the difficulties involved with safely and effectively delivering these proteins to
targeted sites within the central nervous system. Recent progress establishes that gene transfer can now likely overcome the delivery issues
plaguing the translation of neurotrophic factors. This may be best exemplified by adeno-associated virus serotype-2-neurturin (CERE-120),
a viral-vector construct designed to deliver the neurotrophic factor, neurturin to degenerating nigrostriatal neurons in Parkinson’s disease.
Eighty Parkinson’s subjects have been dosed with CERE-120 (some 7 years ago), with long-term, targeted neurturin expression confirmed
and no serious safety issues identified. A double-blind, controlled Phase 2a trial established clinical “proof-of-concept” via 19 of the 24
prescribed efficacy end points favoring CERE-120 at the 12-month protocol-prescribed time point and all but one favoring CERE-120 at
the 18-month secondary time point (p 0.007 and 0.001, respectively). Moreover, clinically meaningful benefit was seen with CERE-120
on several specific protocol-prescribed, pairwise, blinded, motor, and quality-of-life end points at 12 months, and an even greater number
of end points at 18 months. Because the trial failed to meet the primary end point (Unified Parkinson’s Disease Rating Scale motor-off,
measured at 12 months), a revised multicenter Phase 1/2b protocol was designed to enhance the neurotrophic effects of CERE-120, using
insight gained from the Phase 2a trial. This review summarizes the development of CERE-120 from its inception through establishing
“clinical proof-of-concept” and beyond. The translational obstacles and issues confronted, and the strategies applied, are reviewed. This
information should be informative to investigators interested in translational research and development for age-related and other neurode-
generative diseases.
© 2012 Published by Elsevier Inc.
Keywords: Parkinson’s disease; Translational research; Neurodegenerative diseases; Neurotrophic factors; Neurturin; CERE-120; Therapeutic development;
Clinical ‘proof-of-concept’; Gene transfer; Gene therapy; Viral vectors; Protein delivery
1. Introduction
Neurotrophic factors offer one of the most compelling
opportunities to significantly improve the treatment of seri-
ous age-related, neurological diseases such as Alzheimer’s
and Parkinson’s, as well as Huntington’s and amyotrophic
lateral sclerosis. The therapeutic potential of neurotrophic
factors to alleviate the symptoms and slow or even halt
disease progression in neurodegenerative diseases, includ-
ing Parkinson’s disease (PD), is widely acknowledged (Ap-
fel et al., 2000; Eriksdotter Jönhagen et al., 1998; Mufson et
al., 1999; Seiger et al., 1993) and has been independently
* Corresponding author at: Ceregene, Inc., 9381 Judicial Drive, Suite
130, San Diego, CA 92121. Tel.: 1 858 458 8834.
E-mail address: (R.T. Bartus).
Neurobiology of Aging xx (2012) xxx
0197-4580/$ see front matter © 2012 Published by Elsevier Inc.
Page 1
supported by research conducted by numerous laboratories
around the world. A major translational advantage of neu-
rotrophic factors is that they offer the opportunity to treat
both the symptoms of a disease (thus improving clinical
status) as well as its pathogenesis (thus delaying disease
progression) without any prerequisite, deep insight into the
etiology or specific pathogenic variables driving the disease
process. An editorial written more than two decades ago,
entitled “Neurotrophic factors: can the degenerating brain
be induced to heal itself,” helps illustrate the enthusiasm
many of us have felt for a long time, excerpted here: “When
one considers the history of neurology, the idea that one
might be able to treat patients so that their brain cells might
either withstand deadly perturbations or regenerate to a
healthier, more functional state is truly revolutionary. Never
before in the history of medical science could we imagine
the means to induce damaged parts of the brain to heal”
(Bartus, 1989a). While it was clearly too early to know
whether neurotrophic factors might eventually live up to
those early expectations, it would have been even more
difficult for anyone to have known that after more than two
decades of animal research and many clinical trials attempt-
ing to show efficacy in humans, their ability to treat human
neurodegenerative diseases would continue to remain un-
fulfilled this long.
Neurotrophic factors are endogenous proteins that have
consistently demonstrated that under conditions of neuro-
degeneration they are able to activate neuronal repair genes
when supra-physiological (i.e., biopharmaceutical) levels
are achieved. Induction of these repair genes routinely pro-
duces morphological and functional restoration of the de-
generating neurons, significantly slowing further neurode-
generation and even protecting against cell death (Hefti et
al., 1989). Thus, decades of research using numerous animal
models argues that neurotrophic factors provide the oppor-
tunity to substantially improve neuronal vitality and func-
tion in human neurodegenerative diseases (thus potentially
improving symptoms and extending the value of current
pharmacotherapies), as well as to protect against further
neurodegeneration (possibly slowing, halting, or even re-
versing disease progression).
An extremely important point for translational purposes
is that neurotrophic factors appear to provide functional and
morphological benefit to their responsive neurons, no matter
how the neurons are damaged or impaired. Investigators
have consistently shown benefit of neurotrophic factors
against cutting and/or crushing axons, exposure to neuro-
toxins, free radical donors, inflammatory agents and other
cytotoxic agents, genetic mutations, protein processing de-
fects, and the effects of age. Thus, neurotrophic factors
seem to represent a final common therapeutic pathway to
achieve neuronal restoration and protection, likely provid-
ing potential benefit independent of which of many possible
pathogenic cascade(s) are truly responsible for the disease
and thus free of theoretical insight, assumptions, or uncer-
tainties surrounding those issues. The potential therapeutic
effects of neurotrophic factors seem to be “pathogenic neu-
tral,” which offers a major translational advantage, given
the apparent complexity of most chronic neurodegenerative
diseases as well as the uncertainty and controversy regard-
ing which pathogenic variables are most important. There-
fore, if one is able to identify a neuronal population whose
degeneration and/or loss of function has been linked to the
symptoms or pathogenesis of a disease, then the appropriate
neurotrophic factor can likely provide restorative effects
independent of a clear understanding of the pathogenesis
involved. This rather unique characteristic of neurotrophic
factors provides a significant, perhaps unprecedented oppor-
tunity to reduce risk in the development of “first in class”
therapeutics for serious, unmet needs. This approach to treat
neurodegenerative diseases leverages decades of cross-dis-
ciplinary research that collectively establishes “nonclinical
proof-of-concept” for the potential benefit of neurotrophic
factors when degeneration of a specific neuronal population
is known to represent a key feature of a disease.
This scenario makes neurotrophic factors a compelling
target for translational research and development (R&D).
Moreover, the complex but powerful biology of neu-
rotrophic factors suggests that if a significant reduction in
clinical symptoms can be achieved, then a slowing of dis-
ease progression should also occur, simply because the
same repair genes activated by the neurotrophic factor to
improve symptoms should also produce healthier neurons
that are better able to withstand the pathogenic variables
responsible for disease progression. As many have noted in
the past, this possibility of reversing and slowing disease
progression represents the “Holy Grail” for neurological
diseases and neurotrophic factors arguably provide the best
opportunity to accomplish this in the foreseeable future.
The major reason neurotrophic factors have not lived up
to their early promise centers around the long-standing
translational obstacles that impeded safe and effective de-
livery. While an editorial written decades ago titled “Deliv-
ery to the brain: the problem lurking behind the problem”
forewarned that a major translational stumbling block for
neurotrophic factors might involve successful delivery to
the brain (Bartus, 1989b), that problem has proven to be far
more difficult than we had reason to believe at the time.
Similarly, while no one can be certain that solving delivery
issues will necessarily produce the anticipated clinical ben-
efit, it has become increasingly accepted that unless the
delivery problems are solved, reliable and meaningful clin-
ical benefit will likely not be achieved.
Numerous clinical trials, testing many different neu-
rotrophic factors in several different neurodegenerative dis-
eases, have been conducted over the past 20 years (Apfel,
2002; Apfel et al., 1998, 2000; Eriksdotter Jönhagen et al.,
1998; Gill et al., 2003; Lang et al., 2006; Marks et al., 2008;
Miller et al., 1996; Nutt et al., 2003; Penn et al., 1997;
Slevin et al., 2005; Tuszynski et al., 2005; Wellmer et al.,
2 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 2
2001), with mixed results. Three relatively recent, open
label studies, in particular, fueled further enthusiasm for the
possible therapeutic benefits of neurotrophic factors. The
first involved infusion of recombinant glial cell line-derived
neurotrophic factor (GDNF) in 5 PD subjects (Gill et al.,
2003); the next involved ex vivo gene transfer of nerve
growth factor (NGF) in 8 Alzheimer’s disease subjects, 6 of
whom were evaluable (Tuszynski et al., 2005); the third
involved in vivo gene transfer of NRTN (neurturin) in 12
PD subjects, distinguished also by using clear, predefined
primary and secondary efficacy end points (Marks et al.,
2008). While all 3 studies reported preliminary evidence of
clinical benefit, the uncontrolled, open-label nature of the
studies rendered all these results preliminary and merely
suggestive (e.g., see Alterman et al., 2011). In fact, later
studies failed to repeat the motor improvements reported for
GDNF (Lang et al., 2006), the cognitive improvements
reported for NGF (Arvanitakis et al., 2006), or the very
early emergence (i.e., within initial postdosing months) of
motor improvements seen with NRTN (Marks et al., 2010).
More importantly, no controlled trial has yet established
that neurotrophic factors can dramatically improve the clin-
ical symptoms in any neurodegenerative disease, let alone
delay disease progression.
The majority of the recent effort with neurotrophic fac-
tors has been focused on PD, primarily testing GDNF.
Nonclinical studies supported the possible use of GDNF,
infusing the protein into the brain (Gash et al., 1995, 1996;
Kirik et al., 2000a; Maswood et al., 2002), as well as
delivering it via gene therapy (Choi-Lundberg et al., 1998;
Kirik et al., 2000b; Kordower et al., 2000). While several
subsequent clinical studies tested GDNF in moderately ad-
vanced PD patients, all of them have infused the protein
using a chronically indwelling pump into an intracerebral
cannula, thus providing a single “point source” of protein in
the targeted areas of the large human brain. Collectively, the
results of these efforts have been mixed at best (Gill et al.,
2003; Lang et al., 2006; Slevin et al., 2005), with a single,
controlled study showing no evidence for any real benefit
(Lang et al., 2006). Many investigators have agreed that the
point source of delivery employed produced poor distribu-
tion of the protein throughout the putamen, contributing to
the negative results (Morrison et al., 2007; Salvatore et al.,
2006; Sherer et al., 2006).
The cumulative insight gained in the past two decades of
clinical trials attempting to translate the therapeutic poten-
tial of neurotrophic factors provided crucial information
about the delivery requirements that must be met for neu-
rotrophic factors to succeed as human therapeutics. For
example, to treat chronic neurodegenerative diseases, ade-
quate levels of neurotrophic factors must be maintained for
very long periods of time (i.e., years), for when the proteins
return to basal levels, their benefit is typically lost (Fischer
et al., 1987; Hefti et al., 1989; Snider and Johnson, 1989;
Sofroniew et al., 2001). Similarly, it is important that an
appreciable proportion of the degenerating cell population
be exposed to the neurotrophic factor in order to produce
sufficient restoration of neuronal function and thus achieve
measurable clinical improvement (though the exact propor-
tion required for therapeutic benefit is currently not clear,
and may differ between diseases). Because serious side
effects have been observed when delivery of neurotrophic
factors has inadvertently resulted in exposure to nontargeted
brain sites (e.g., periventricular tissue) the importance of
accurately predicting, controlling, and restricting protein
delivery specifically to the intended target has become ap-
parent (Day-Lollini et al., 1997; Eriksdotter Jönhagen et al.,
1998; Kordower et al., 1999; Nauta et al., 1999; Nutt et al.,
2003; Penn et al., 1997). Further complicating the transla-
tion of neurotrophic factors is the fact that because they are
proteins, chronic delivery can be notoriously difficult be-
cause of aggregation, misfolding, and development of neu-
tralizing antibodies. Moreover, as proteins, they cannot be
taken orally, do not cross the blood-brain barrier (BBB)
naturally and cannot typically be administered systemically,
even if linked to a BBB carrier, because of side effects
often induced by systemic exposure to organs and tissues
(McMahon, 1996; Pezet and McMahon, 2006). These is-
sues, individually and collectively, render safe and effective
delivery of neurotrophic factors extremely challenging. It is
for this reason that a consensus opinion has emerged among
investigators that the successful translation of neurotrophic
factors to the human clinic will first require that these
crucial delivery issues be solved (Bartus, 1989a; Bartus et
al., 2007; Kordower et al., 1999; Lang et al., 2006; Nutt et
al., 2003; Salvatore et al., 2006; Sherer et al., 2006).
Unfortunately, the issues described preclude employing
most traditional pharmaceutical formulations and delivery
approaches as viable options for administering neurotrophic
factors to patients’ brains. A number of more innovative
methods have therefore been devised in an attempt to ef-
fectively deliver these proteins to the central nervous system
(CNS), but all have suffered serious limitations. For exam-
ple, various efforts to transport neurotrophic factors across
the BBB following systemic administration have been
tested. One method that showed promise exploited endog-
enous transport receptor-mediated systems located on the
abluminal surface of cerebral capillaries (e.g., transferrin
transport receptors). While early nonclinical conceptual
success was achieved (Bäckman et al., 1996, 1997; Bartus,
1999; Charles et al., 1996), the approach ultimately proved
impractical because of serious peripheral side effects in-
duced via exposing nontargeted tissue outside the brain to
the neurotrophic factor following intravenous injections
(McMahon, 1996; Pezet and McMahon, 2006), coupled
with only modest benefit, compared with infusion of pro-
teins directly into the brain. Infusions of neurotrophic factor
proteins directly into the ventricles of the brain, which
required a more invasive, neurosurgical approach, showed
early promise in animal studies of neurodegeneration. How-
3R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 3
ever, significant side effects in humans occurred in the
periventricular tissue exposed to high concentrations of the
neurotrophic factor (Eriksdotter Jönhagen et al., 1998; Kor-
dower et al., 1999; Nauta et al., 1999; Nutt et al., 2003; Penn
et al., 1997). While subsequent infusions of the proteins
directly into the degenerating parenchyma using chronically
indwelling pumps and cannula reduced some, but not all
safety issues (Hovland et al., 2007; Lang et al., 2006), poor
diffusion of the protein from a single point source severely
limited exposure of the protein to the larger area of degen-
erating brain tissue, thus likely limiting clinical benefit
(Lang et al., 2006; Salvatore et al., 2006; Sherer et al.,
2006). Similarly, complications involving indwelling hard-
ware posed serious safety risks (e.g., the formation of neu-
tralizing antibodies to GDNF and degeneration of distant
cerebellar neurons due to protein leaking uncontrollably
along paths of least resistance). In other words, while the
scientific foundation for neurotrophic factors is considered
well-established, attempts to translate the therapeutic poten-
tial to the clinical arena have been largely disappointing
because the technology required to deliver these complex
proteins in a safe, controlled, and sustained fashion to spe-
cific, targeted areas of the brain has been grossly inade-
2. CERE-120 (AAV2-NRTN), PD, and the use of gene
transfer to solve the delivery obstacles posed by
neurotrophic factors
Parkinson’s disease is a chronic, debilitating disease,
whose major symptoms involve loss of motor ability, in-
cluding bradykinesia, tremors, and problems with gait and
balance. It is widely recognized that these major symptoms
result from the progressive loss of function and eventual
death of the nigrostriatal dopamine neurons, linked at least
in part to the accumulation of misfolded
-synuclein aggre-
gates (McNaught and Olanow, 2006). While available phar-
maceutical agents for PD are generally effective during the
earlier stages of the disease, the benefit of existing treat-
ments eventually wanes as the disease progresses, due in
large part to an increasingly narrow therapeutic index. This
is manifested by significant portions of each day when
patients can no longer initiate or adequately control
movement (i.e., “wearing off” phenomenon), as well as
the emergence of disabling, peak-dose dyskinesias due to
increased drug sensitivity. While deep brain stimulation
(DBS), which involves a neurosurgical procedure and
implantation of chronically indwelling hardware, offers
temporary relief of certain motor symptoms, it also often
occurs with significant complications (Deuschl et al.,
2006) and does not provide satisfactory benefit to many
PD patients. Moreover, no current treatment, including
DBS, is able to slow disease progression. Thus, superior
treatments that might reduce symptoms in the more ad-
vanced stages of the disease, as well as delay further
progression, are clearly needed.
Neurturin, like GDNF discussed earlier, is a member of
the glial cell line-derived family of ligands (GFLs). Both
proteins provide robust trophic support for the midbrain
dopamine nigrostriatal neurons whose degeneration is piv-
otal to PD (Hoane et al., 1999; Horger et al., 1998; Kotz-
bauer et al., 1996; Lin et al., 1993; Oiwa et al., 2002;
Rosenblad et al., 1999; Tseng et al., 1998). Adeno-associ-
ated virus serotype-2 (AAV2)-NRTN (CERE-120) was de-
signed and developed to overcome all the delivery issues
posed by neurotrophic factors (discussed earlier) by circum-
venting the specific obstacles involved with delivering sus-
tained quantities of biologically active proteins into selec-
tively targeted brain tissue. It is designed to deliver the gene
for NRTN to targeted neurons, and subsequently program
these neurons to provide continuous, long-term, predictable
NRTN expression in selective, stereotactically-targeted re-
gions of the brain. CERE-120 consists of a genetically
engineered AAV2 vector that lacks all of the viral protein
coding sequences, encoding only human NRTN cDNA, thus
inducing expression of only human NRTN in the transduced
cells (Gasmi et al., 2007a). Thus, rather than attempting to
exogenously deliver the large, 3-dimensionally complex
protein directly to the targeted site, CERE-120 delivers the
gene for the protein to the targeted site, thereby inducing
local cells to manufacture and secrete the protein through
their endogenous cellular machinery.
An important aspect related to the safety of CERE-120 is
that only relatively low doses of vector are required to
provide relatively widespread coverage of the targeted ni-
grostriatal system with NRTN protein, avoiding significant
systemic exposure as well as exposure to other nontargeted
neuronal sites. Finally, because the CNS is relatively im-
munologically privileged and AAV2 is relatively nonin-
flammatory, the probability of eliciting an immune or in-
flammatory reaction is further reduced. At the same time, the
opportunity for robust, long-lasting therapeutic effects in
chronic neurodegenerative diseases following a single CERE-
120 administration seemed plausible with adeno-associated
virus serotype-2 in combination with the transcriptionally ac-
tive CAG promoter, for research from other laboratories had
shown this combination provided continuous, persistent ex-
pression of the transgene protein over several years (Bankie-
wicz et al., 2003, 2006) (an observation since replicated with
CERE-120; Herzog et al., 2012). Thus, CERE-120 is intended
to provide a lifetime of neurotrophic factor support via NRTN
following a single administration, offering additional safety
advantages, while eliminating numerous inconveniences com-
monly associated with indwelling hardware.
While the need for standard stereotactic surgical tech-
niques does require a “paradigm shift” from the way human
diseases have traditionally been treated, the magnitude of
the problems posed by neurodegenerative diseases demand
that truly innovative approaches be given serious consider-
4 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 4
ation. The progress made by employing DBS in PD, both in
terms of improved patient status as well as patient and
caregiver acceptance, offers compelling evidence that neu-
rosurgical approaches can be both effective and practical.
The need for indwelling hardware and requirements to make
postoperative adjustments, arguably makes DBS a more
complicated and invasive approach than gene transfer,
which based on current data will likely involve a single,
once-per-lifetime treatment. Moreover, despite the clinical
benefits of DBS and its deserved, growing acceptance, it has
no impact on the underlying pathogenesis and thus does
nothing to slow disease progression, falling well short of
what a neurotrophic factor/gene transfer approach might
provide. Thus, the collective characteristics of safe, very
long-term, controlled protein expression that can be targeted
to specific sites or systems supports the idea that gene
transfer may have finally solved the delivery problems re-
quired for neurotrophic factors.
3. The initial nonclinical program supporting
CERE-120 for PD
3.1. Introduction
Like virtually all translational R&D programs, the main
intent of the CERE-120 preclinical program was to accu-
rately determine whether CERE-120 was sufficiently safe
and effective to justify moving into Parkinson’s subjects for
further evaluation of safety and efficacy. A number of im-
portant and reasonably clear questions come to mind when
considering this general goal. One of the first and perhaps
most obvious characteristic was whether the NRTN ex-
pressed following CERE-120 administration can provide
robust neurotrophic support for nigrostriatal dopamine neu-
rons, improving their function and protecting against degen-
eration. While this type of question is one that most often
comes up when colleagues in academia discuss “transla-
tional” research, it is far from the final issue that must be
addressed and very often is similarly far from the most
important. In reality, the scientific literature is filled with
thousands of observations suggesting some new chemical
entity or new pathway might have significant therapeutic
value for some important human disease, though few ever
advance to clinical testing and the vast majority of those that
are tested in humans soon fail to advance much further.
Thus, while important to confirm that NRTN could provide
robust neurotrophic effects in degenerating neurons in ap-
propriate animal models, in many ways this was the easiest
and most straightforward (and therefore the least challeng-
ing and interesting) question that required study. Yet, a
number of other important translational questions required
answers for which little or no data yet existed, whether
considering NRTN specifically, or neurotrophic factors gen-
erally, and whether delivered via viral vectors or any other
means. Perhaps the most important was whether CERE-120
could be delivered safely, and whether NRTN could be
expressed for sufficiently long periods of time and with a
sufficiently wide therapeutic index to justify advancing the
program to clinical testing in PD patients. This issue was
greatly complicated by the heightened sensitivity to the
possible safety risks associated with injecting viral vectors
into humans, due to some of the early, unrelated safety
issues experienced in the gene therapy field (Kohn et al.,
2003; Raper et al., 2003). Because CERE-120 would be
injected directly into the brain, and once administered,
NRTN expression could not be terminated, it was necessary
to have even greater confidence that NRTN expression
would be safe and well tolerated. Other questions for which
very little data existed in the field, but were equally impor-
tant for translational purposes, involved whether protein
expression followed a predictable and orderly dose–re-
sponse and thus whether the amount of protein, as well as its
location and pattern of expression, could be adequately
controlled. Finally, it was also important to establish that the
NRTN expression would remain reasonably consistent over
long periods of time and would not migrate from the tar-
geted site.
These questions and the more general issues of confirm-
ing robust bioactivity of NRTN expressed via a viral vector
and the consequences of that expression, served as the
framework from which the CERE-120 nonclinical program
was developed, implemented, and executed. Additionally,
during the execution of the program, we were confronted
with a number of unexpected, and at times difficult, trans-
lational issues that required resolution in order for the pro-
gram to advance (see later section 3.5). Collectively, these
ultimately led to the initiation of 20 different nonclinical
studies prior to the first tests in humans.
Because the prevailing viewpoint at the time this pro-
gram was initiated (i.e., ca 2003) was that it was not nec-
essary (and possibly counterproductive) to deliver neu-
rotrophic factors directly to the degenerating cell bodies
located deep in the substantia nigra (SN), the initial ap-
proach employed for CERE-120 (and expression of NRTN)
was to target the degenerating terminal fields (i.e., striatum)
only, not directly targeting the degenerating cell bodies in
the SN. A great many studies had previously shown that
administering the neurotrophic factor to the terminal fields
in the striatum was both necessary and sufficient (for de-
tailed discussion of this issue, see Bartus et al., 2011a).
Thus, the initial nonclinical program adopted that approach,
administering CERE-120 to the striatum to evaluate several
translational characteristics essential to support and guide
eventual clinical testing in Parkinson’s patients using the
same striatal dosing approach.
In essence, three general questions were addressed by the
initial nonclinical R&D program: (1) can CERE-120 per-
form in a controlled and predictable manner?; (2) will
CERE-120 (and subsequent NRTN expression) prove un-
usually safe when tested in animals, even at excessively
high doses and over long time periods of time?; and (3) will
5R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 5
the NRTN expressed via CERE-120 remain potently and
persistently bioactive for long periods in a variety of model
3.2. CERE-120 performs in a controlled and predictable
Determining how well one can control and predict the
effects of any new therapeutic is clearly one of the most
important issues that must be addressed well before plan-
ning “first-in-human” testing. Without the ability to control
or predict a product’s effects, it is difficult to move confi-
dently into human testing. Importantly, this information also
provides the empirical information required to select the
doses to be tested in humans, as well as the insight into the
types of responses (beneficial or otherwise) that might be
expected and will need to be carefully monitored. There-
fore, characterizing NRTN expression in brain following a
range of orderly doses and over varying intervals of time
was an essential early component of the nonclinical pro-
The primary objective of delivering CERE-120 is to
achieve expression of NRTN throughout the targeted nigro-
striatal system, while at the same time avoiding NRTN
exposure to nontargeted sites within the CNS which could
conceivably cause side effects. In particular, we felt it im-
portant to avoid exposing periventricular areas to NRTN
since direct infusion of other neurotrophic factors (e.g.,
GDNF, NGF) into the ventricles had been shown to produce
adverse effects in animals and humans (Day-Lollini et al.,
1997; Eriksdotter Jönhagen et al., 1998; Kordower et al.,
1999; Nutt et al., 2003). Thus, it was important that an
understanding be achieved for the variables that influence
the volumetric spread (location) of protein as well as the
amount (i.e., protein levels), so that the insight gained might
eventually be applied for dose selection for the even larger
putamen in humans. It was also important to understand the
kinetics of expression following CERE-120 administration
in order to rationally design efficacy experiments as well as
design and interpret toxicology/safety experiments (i.e.,
knowing when maximal protein expression occurred was
important for each).
Several experiments were conducted in rats and monkeys
to examine the time course (or kinetics) of NRTN expres-
sion following CERE-120 administration and the relation-
ship between CERE-120 dose and NRTN expression. From
the start of this program, we recognized the importance of
two key, interrelated points: (1) the need to develop a dosing
strategy that would help protect patient safety and (2) that
the greatest risk of side effects from neurotrophic factors
had been linked to mistargeted protein, particularly in or
near periventricular brain sites. Thus, for the first gene
therapy/neurotrophic factor program to begin translating
work in animals to the human clinic, we necessarily adopted
an appropriately conservative dosing strategy that employed
more modest convection currents (i.e., induced by infusion
rates of 2
L per minute, lasting several minutes each),
coupled with strategically distributed points of infusion
within the targeted brain sites. While the number of needle
tracts was greater than that required for convection-en-
hanced delivery (CED), we avoided the notorious issue of
uncontrolled, mistargeted protein spreading along paths of
least resistance (especially white matter tracts) known to
regularly occur with CED (e.g., Ding et al., 2010; Lampson,
2001; Linninger et al., 2008), which would greatly heighten
the risk to patients. Our approach has shown that NRTN
expression is indeed successfully limited to the targeted
brain sites in both animals and humans, while also inducing
appropriate neurotrophic responses (see later text).
NRTN expression was quantified on the basis of volume
of distribution (via immunohistochemistry, IHC) as well as
levels of protein (via ELISA or enzyme-linked immunosor-
bent assay). Collectively, these studies established that: (1)
the amount of NRTN expressed (measured via ELISA) and
volume of NRTN expressed in brain (measured via immu-
nohistochemistry) is tightly correlated, both as a function of
dose as well as kinetics of expression over time (Gasmi et
al., 2007a); (2) NRTN volume of expression and levels of
protein reach an asymptote at approximately 1 month after
CERE-120 administration (Gasmi et al., 2007a, 2007b); (3)
volume of NRTN expression remains stable throughout the
life-span of the rat and for at least a full year in monkeys
(Herzog et al., 2009, 2011); (4) the volume of distribution of
NRTN can be controlled by manipulating the dose of
CERE-120 in both rats (Gasmi et al., 2007a, 2007b) and
monkeys (Herzog et al., 2008, 2009, 2011); and (5) by
distributing several injections of CERE-120 within the pu-
tamen at predetermined dose levels, it is possible to express
NRTN throughout a significant portion of the putamen
and/or SN, while avoiding exposure to other (nontargeted)
brain regions (Bartus et al., 2011a; Gasmi et al., 2007a,
2007b; Herzog et al., 2007, 2008, 2009, 2011).
These experiments with CERE-120 administration dem-
onstrate stable NRTN protein expression that can be con-
trolled by modulating dose of CERE-120, thus helping to
accurately target expression to the nigrostriatal neurons and
avoid mistargeted protein to other areas where side effects
might be induced. The data generated made planning and
launching an efficient and effective safety/toxicology non-
clinical program that much easier, for knowing that maxi-
mum expression occurred within a month and that protein
expression did not change thereafter greatly simplified the
planning and interpretation of the safety/toxicity studies.
The data from these studies also provided the empirical
foundation for dose selection for nonclinical efficacy and
safety studies.
3.3. CERE-120 appears unusually safe, even at high
doses and over long time periods
Another important component of any nonclinical devel-
opment program involves the extensive testing for safety
6 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 6
and potential toxicity of a product prior to planning tests in
humans. Not only is it important to define the safety limits
of a product (typically in terms of maximum dose that can
be tolerated) but it is also important to attempt to define the
types of side effects or toxicity that might be induced (to
help make rational and informed “risk:benefit” judgments).
Additionally, when the major safety risks are defined, as
well as the doses that induce them, it is possible to compare
those doses with those required for efficacy (which obvi-
ously needs to be significantly lower). This then allows one
to define a “therapeutic index,” or range of doses that
provide the desired effect while avoiding any undesirable
Constructing and executing an unusually thorough and
conservative nonclinical safety program for CERE-120,
prior to moving into human testing, was appropriate for
several related reasons. First, the history of efforts to trans-
late the promise of gene therapy and neurotrophic factors
into clinical trials has been fraught with many unanticipated
and sometimes serious side effects. Secondly, intervention
in the human brain necessarily raises even greater concerns
when considering the possible side effects that might occur
with a novel experimental intervention. Finally, CERE-120
was expected to produce relatively permanent (i.e., irrevers-
ible) NRTN expression, necessitating that the burden of
proof for safety be set higher than normal to establish that
no harm could reasonably be expected from administration
of CERE-120 or long-term expression of NRTN.
In response to these considerations, we tested large num-
bers of rats and monkeys, at doses hundreds of times higher
than those required for bioactivity and efficacy in the animal
models, using many different end points to search for pos-
sible side effects (Table 1). It was hoped such a large array
of measurements (many not even involving the CNS) and
use of unusually high doses would help identify potential
side effects that might occur from entirely unanticipated
sources or in unexpected ways. Six different safety/toxicity
studies in rats and 4 studies in monkeys were conducted.
Because age is a major risk factor in PD, we included
studies with aged rats and monkeys. The data from these
studies established a wide safety margin for CERE-120. In
fact, surprisingly, no CERE-120-related (or NRTN-related)
toxicity or side effects were observed at any dose or time
point, on any measurement, in any of the studies conducted.
For this reason, CERE-120 enjoyed the rare circumstance
that no “maximum tolerated dose” could be identified in
either rats or monkeys. Even the highest doses that could
physically be administered, which to our knowledge are far
higher than anyone else has previously tested or reported
(4 10
vector genomes [vg]/brain to rats and 3.6 10
vg/brain to monkeys) and were substantially higher than
required to produce the desired neurotrophic responses (Ta-
bles 2 and 3), produced no side effects. Thus, a pristine
safety profile was established over a range of doses (many
intentionally excessive), up to 1.5 years post-CERE-120 in
rats and a year in monkeys, with persistent and stable NRTN
expression confirmed over these long time points.
Finally, a specific safety issue that must be addressed for
all gene therapy translational programs is identifying where
Table 1
Safety/toxicology outcomes in monkeys and rats (3 through 12 months,
post CERE-120)
No histopathological abnormalities in brain, including the targeted
nigrostriatal system and cerebellum
No histopathology abnormalities in organs, including heart, lungs, liver,
No evidence of brain inflammatory or immune reactions
No abnormalities on neurological/behavioral examinations
No weight loss
No Schwann cell hyperplasia
No evidence of pain
No NRTN detected in CSF
No systemic immune response to NRTN
Dose-related increase in antibody titers to adeno-associated virus
serotype-2, but reaction not associated with any other outcome
Data derived from Gasmi et al. (2007b) and Herzog et al. (2008, 2009,
Key: CSF, cerebrospinal fluid; NRTN, neurturin.
Safety-toxicity dose-equivalents for putamen in nonhuman primates
were approximately 95-fold higher than the initial low Phase 1 dose,
approximately 21-fold higher than the high Phase 1 and Phase 2a dose,
and 6-fold higher than the Phase 2b (i.e., ongoing study) dose; see
Table 3.
Specific reactions shown by others to be linked to neurotrophic factor
exposure to periventricular regions.
Table 2
Range of preclinical models and variety of relevant end points
demonstrating neurotrophic activity with CERE-120
6-OHDA rat model of PD (Gasmi et al., 2007a, 2007b)
Protection of nigral cells at multiple time points (up to 7 mo) and
range of doses
Functional (behavioral) benefit
Aging rats (Herzog et al., 2011)
Persistent expression of biologically active NRTN up to 20 mo, post
Elevated pERK (phosphorylated extracellular regulated kinase) in
substantia nigra neurons persisted for same duration
Aged rats (Herzog et al., 2011)
“Classic” neurotrophic-induced hypertrophy of nigral neurons
Enhanced number of pERK positive nigral neurons
Young, healthy monkeys (Herzog et al., 2008, 2009)
Enhanced nigrostriatal TH
Activation of pERK signaling
“Classic” neurotrophic-induced hypertrophy of nigral neurons
Aged monkeys (Herzog et al., 2007)
Enhanced F-Dopa PET uptake in striatum
Enhanced TH in striatum and nigra
Increased numbers of TH-positive nigral neurons
MPTP monkey model of PD (Kordower et al., 2006)
Long-lasting improvement in motor performance
Preservation/protection of nigral neurons
Enhanced TH in nigra and pERK activation
Enhanced TH in terminals in striatum
Key: 6-OHDA, 6-hydroxydopamine; PD, Parkinson’s disease; PET, posi-
tron emission tomography; TH, tyrosine hydroxylase; NRTN, neurturin;
pERK, phosphorylated extracellular signal-regulated kinase; MPTP,
7R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 7
the vector distributes itself following administration. Re-
ferred to as “biodistribution” studies, quantitative polymer-
ase chain reaction (PCR) methods are employed to quantify
the amount of vector in the targeted tissue, as well as
determine whether it is present in numerous other parts of
the body, over varying periods of time. Our biodistribution
studies demonstrated that the vast majority of CERE-120
genome remained in the targeted striatum. Outside the
brain, low levels were detected only in the cervical lymph
nodes, spleen, and liver and these resolved shortly after
dosing. Importantly, no CERE-120 mRNA expression was
detected (by reverse transcription-polymerase chain reac-
tion) in these tissues, indicating that measurable transduc-
tion of cells outside the brain did not occur. Consistent with
lack of transduction outside the brain, no evidence for
NRTN outside the brain (or even in the cerebrospinal fluid;
CSF) has ever been seen.
In sum, the results from the CERE-120 safety/toxicology
studies (Bartus et al., 2007; Gasmi et al., 2007b; Herzog et
al., 2007, 2008, 2009, 2011) not only supported advancing
into clinical trials in humans, but made establishing doses
for humans, as well as selecting doses for animal efficacy
studies prior to that much easier. Because excessive CERE-
120 doses, or NRTN expression did not cause any side
effect or toxicity (even at the microscopic, cellular level),
we could focus on selecting doses that would adequately
cover the targeted brain sites, while assuring significant
protein did not exceed the targeted boundaries, for we still
could not yet be certain that mistargeted protein may not
cause some of the side effects reported by others infusing
protein, or allowing protein to leak into or near the ventri-
cles (Eriksdotter Jönhagen et al., 1998; Hovland et al., 2007;
Kordower et al., 1999; Lang et al., 2006; Nauta et al., 1999;
Nutt et al., 2003; Penn et al., 1997).
3.4. NRTN expressed via CERE-120 is potently and
persistently bioactive
We believe that no animal model can truly capture all the
important elements of any human neurodegenerative dis-
ease and therefore it is important to study a wide range of
possibilities (for discussion of these issues, see reviews;
Bartus 1988; Bartus et al., 1983). For this reason, we elected
to test CERE-120 in multiple and varying animal models,
recognizing that as a complex disease, PD likely involves
many different etiologic and pathogenic variables that drive
the degeneration and death of the nigrostriatal dopamine
neurons. Our operating assumption is, if one could demon-
strate that the NRTN expressed following CERE-120 ad-
ministration can overcome a wide variety of nigrostriatal
perturbations in animals, it should then provide trophic
support to the same degenerating nigrostriatal neurons in
PD patients.
We therefore conducted a series of experiments using
both rats and monkeys to test whether the NRTN expressed
following CERE-120 administration is capable of producing
the desired neurotrophic response in nigral dopamine neu-
rons, using a variety of measurements (see Table 2). Col-
lectively, these studies consistently showed that NRTN,
when covering a reasonable but by no means complete
volume of the striatum, can induce a neurotrophic response
in nigrostriatal neurons following CERE-120 administra-
tion, thus enhancing their function, improving their status,
and protecting them from degeneration and death (Bartus et
al., 2011a; Gasmi et al., 2007a, 2007b; Herzog et al., 2007,
2008, 2009, 2011; Kordower et al., 2006). Together, these
data establish that the NRTN protein expressed following
CERE-120 administration is robustly and persistently bio-
active in a wide variety of model systems. These data
therefore provided further justification for advancing
CERE-120 into human testing in PD subjects.
3.5. Additional translational issues and obstacles
In addition to the more traditional translational issues
reviewed in the prior section, during the planning and exe-
cution of the CERE-120 program, we were confronted with
a number of more unique translational issues that required
resolution. Some of these were anticipated while others
were unexpected and emerged during the collection of data
from other studies. However, all required resolution in order
to gain sufficient information to determine whether the
program should be suspended or advanced. Most required
additional, previously unplanned studies to be conducted to
better understand the issues at hand and thus ascertain the
risks involved, or define a solution for resolution. Some of
these were fairly general to gene transfer and/or chronic
Table 3
Cross-species dose equivalents (based on relative volume of targeted brain sites), enabling rational, species-to-species dose comparisons
Species Volume of
target (mm
Efficacy considerations Safety considerations
Doses tested per
(vg 10
Dose equivalents
(vg 10
Highest dose tested
per hemisphere
(vg 10
Highest dose
equivalent tested
(vg 10
Rat (striatum) 25 0.16–4.0 0.006–0.16 20 0.8
Monkey (striatum) 1200 150 0.13 1800 1.5
Human Phase 1: first in humans (putamen) 4000 65–270 0.016–0.07 270 0.07
Human Phase 2a (putamen) 4000 270 0.07 270 0.07
Human revised Phase 1/2b (putamen) 4000 1000 0.25 1000 0.25
Key: vg, vector genomes.
8 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 8
protein delivery, whereas others were more specific to
NRTN and/or GFLs. The main issues are summarized in
Table 4.
CERE-120 is comprised of an adeno-associated virus
(AAV) vector, and because AAV occurs naturally, a signif-
icant portion of the population has been exposed to it. While
AAV appears generally innocuous (e.g., no clinical symp-
toms or disease have ever been linked to AAV exposure), a
significant proportion of the human population (30%– 60%)
nonetheless is reported to be positive for AAV neutralizing
antibodies (Calcedo et al., 2009). While it is not clear
whether these neutralizing antibodies to AAV might negate
the viral vector, or possibly impair the transduced cells, this
possibility was one we were not willing to accept and
therefore sought some means to gain empirical data to help
ascertain the degree of risk involved with dosing future
subjects who might have pre-existing antibodies to AAV.
Though we initially were frustrated to learn that no vali-
dated artificially immunized models of AAV existed to
address this issue experimentally, we later discovered that
many nonhuman primates have naturally been exposed to
AAV and developed neutralizing antibodies to the virus.
This provided us with a reasonable approach to empirically
address this issue, testing CERE-120 in monkeys with and
without pre-existing neutralizing antibodies to AAV. These
studies demonstrated, at least using the brain-directed dos-
ing procedures employed for CERE-120, that prior-existing
antibodies to AAV have no impact on NRTN expression or
its bioactivity, nor do they induce an inflammatory response
following AAV-mediated gene transfer to the brain. To
Table 4
Major, nontraditional translational issues confronted and resolved during the development of CERE-120 (adeno-associated virus serotype-2-NRTN) for
Parkinson’s disease
Translational issue confronted Approach Outcome/decision
Concern regarding pre-
existing Ab to AAV and
impact on safety and/or
NRTN expression
Identify primates that are positive for AAV Ab
and administer CERE-120 to test for possible
Pre-existing antibodies do not affect CNS expression, bioactivity,
or toxicity, at least with 1-time administration of CERE-120 to
Concern for whether a vector
allowing regulation of
gene (i.e., ability to turn
expression on and off) is
required to assure safety
Evaluate available regulatable vectors; assess risks
with using “nonregulatable” vectors, including
testing high doses of CERE-120 over extended
time periods, using many different end points to
help assess potential risks of unregulated protein
None of available regulatable vectors were deemed ready for
clinical use; CERE-120 (producing high, persistent NRTN
expression levels) found to be very safe; therefore, use of a
regulatable vector neither feasible or required
Issue of initial poor
expression/secretion of
NRTN with natural NRTN
pre-pro sequence
Understand phenomenon, identify source of
problem and find/implement solution
Pre-pro sequence for human NGF substituted for natural pre-pro
NRTN sequence, correcting problem
Loss of vector due to binding
to dosing hardware during
CERE-120 administration,
thus reducing deliverable
dose in nonlinear fashion
Understand phenomenon; elucidate vector binding
characteristics and develop means to prevent
binding, therefore preventing loss of dose
Short-term solution (to maintain clinical time lines): develop and
implement method to “prime” injection needle with vector to
saturate binding sites prior to dosing.
Long-term solution (for eventual multicenter studies and beyond):
develop and implement new dosing needle with inner liner that
does not bind vector and proves safe for human use
Concerns regarding dose-
related reduction in TH
(observed in rats)
Develop clear understanding of phenomenon via
series of explicit experiments; make judgment as
to implications
Reduction in TH IHC shown to be mere, transient, species-
specific (rats only) compensatory response to the enhanced
dopamine tone; deemed to be a nontoxic event; also, does not
occur in primates
General concern regarding
potential cerebellar toxicity
(based on reports from
GDNF toxicity with protein
infusions in monkeys)
Understand phenomenon by critically evaluating
available GDNF cerebellar toxicity data; conduct
detailed evaluation of cerebellum following
NRTN expression
No evidence of cerebellar toxicity seen with NRTN expressed via
gene transfer or any of the events that were linked to GDNF
toxicity; concluded evidence with GDNF pointed to deficiencies
in specific infusion methods used to deliver recombinant GDNF
Issue regarding how to select
safe and effective doses to
test in humans, based on
the available data, primarily
in animals
Collect dose–response safety and efficacy data in
animals; combine with information regarding
interspecies differences in target volume and use
to “scale” doses accordingly
Dose scaling approach employed; later quantification of NRTN
expression from postmortem brains from human PD subjects that
received CERE-120 confirm utility of “scaling” methods
Concerns regarding potential
safety issues of
stereotactically dosing SN
with CERE-120
Carefully evaluate all published literature relevant
to issue; conduct explicit experiments to address
specific questions remaining unanswered
Safety of targeting SN with AAV-NRTN established; wide safety
margin demonstrated with targeted CERE-120 to SN
Key: AAV, adeno-associated virus; Ab, antibodies; CNS, central nervous system; GDNF, glial cell line-derived neurotrophic factor; IHC, immunohisto-
chemistry; NGF, nerve growth factor; PD, Parkinson’s disease; SN, substantia nigra; TH, tyrosine hydroxylase; NRTN, neurturin.
9R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 9
date, the clinical data appear entirely consistent with pre-
dictions and expectations from these nonhuman primate
Another early issue we had to consider was whether to
attempt to employ one of the so-called “regulatable vectors”
still under development, or alternatively use a far simpler
and better characterized vector that would produce constant
protein expression, but with no means of modulating ex-
pression, let alone turning it off. Without a “regulatable
vector,” if any serious side effect were to occur following
NRTN expression, it would not be possible to halt protein
expression and we would be forced to deal with the side
effect using more traditional medical means, such as med-
ications. In the end, we decided that none of the available
regulatable vectors were yet ready for clinical use, and
worried that all of them posed more translational issues
and/or had as many inherent safety issues as did the use of
an unregulatable vector for NRTN expression (for more
detailed discussion, see Bartus, 2012). Fortunately, the
broad safety margins (and lack of any measurable toxicity)
argue that the use of a regulatable vector was/is not re-
quired, a point further supported by the lack of any subse-
quent effects to chronic NRTN expression in animals or
One of the more surprising (and initially disappointing)
observations noted in our initial efforts to characterize
AAV2-NRTN expression and bioactivity, was the observa-
tion of poor NRTN expression in the conditioned medium
from cells transduced with the NRTN gene 48 hours earlier
(Ceregene, Inc., San Diego, CA, USA, unpublished). We
confirmed and extended this confusing result by injecting
the same viral vector into the striatum of rats, which again
produced a relatively poor NRTN immunohistochemical
signal, particularly in comparison with a positive control, a
viral vector expressing GDNF (Fig. 1). More detailed ex-
amination of the histology slides clearly suggested the ab-
solute level of protein was low and that what little NRTN
was expressed was not efficiently secreted from the trans-
duced cells. Because we recognized that protein expression
is regulated via the pre-pro sequence, we surmised that for
unknown reasons, the natural NRTN pre-pro sequence
might limit the degree or efficiency of protein secretion,
especially when supraphysiological levels are sought, and
we therefore constructed, tested, and confirmed the effec-
tiveness of a substituted human NGF pre-pro sequence (Fig.
1). Replacing the natural NRTN pre-pro sequence with that
of NGF solved the expression/secretion deficiencies. Once
we confirmed that the protein secreted indeed was the full-
length human NRTN amino acid sequence, we established
the safety and efficacy of the protein in a series of subse-
quent experiments and have continued the CERE-120 pro-
gram with this bioengineered construct ever since.
Yet another general translational issue we encountered
was the loss of vector because of binding that occurred
when passing CERE-120 through our dosing hardware. We
performed a series of bench experiments with the vector and
dosing hardware to understand the phenomenon as well as
possible, learning among other things, that the binding was
saturable. This also explained why, with less purified (i.e.,
nonclinical grade) formulations than those we prepared for
Fig. 1. Photomicrographs comparing neurturin (NRTN) and glial cell line-derived neurotrophic factor (GDNF) expression achieved with different viral vector
constructs (immunohistochemical inserts) and extent of nigrostriatal dopamine neuronal protection against 6-hydroxydopamine (6-OHDA) toxicity with each
of these constructs (histograms). (A) Low and high power (inserted box) magnified views of robust GDNF expression following injection of vector expressing
the GDNF gene using the natural GDNF pre-pro sequence, versus weak NRTN expression following injection of vector expressing the NRTN gene using
the natural NRTN pre-pro sequence (note high power view suggests most protein remains intracellular). Note that GDNF vector provided significant
neuroprotection against 6-OHDA whereas the NRTN vector with its natural prepro sequence did not. (B) Low and high power magnification views of robust
NRTN expression following injection of vector expressing NRTN, with the substitution of the human NGF pre-pro sequence for natural NRTN pre-pro
sequence, as well as robust neuroprotection achieved with this vector against 6-OHDA toxicity. Note the substantially more robust NRTN immunohisto-
chemical signal, as well as evidence for ample extracellular secretion of NRTN (high magnification) and substantial neuroprotection against 6-OHDA,
compared with that achieved with NRTN vector in (A). GDNF data represent a separate replication of expression and neuroprotection using a similar GDNF
vector construct shown in (A). Data in (B) were derived from results previously published by Gasmi et al., 2007a and are shown here for pertinent comparison.
10 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 10
future human tests, many other investigators using AAV
had not witnessed a similar loss of vector in their laboratory
studies. Exploiting the observation that the AAV binding
sites inside the dosing cannula can be saturated, we devel-
oped a method to saturate all of those sites immediately
prior to each surgery. In order to stay on track for filing an
investigational new drug (IND) application, this required
assurance and formal validation that this method would
work under a range of conditions that might be encountered
in the operating room. Eventually we developed a proprie-
tary nonstainless steel lining for the inner wall of the injec-
tion needle to which the vector did not bind prior to our
Phase 2a study, and this approach continues to be used for
all subsequent human studies.
One of the more worrisome observations made early in
the CERE-120 program involved a dose-related reduction in
tyrosine hydroxylase (TH) expression in the nigrostriatal
neurons of rats. The loss of TH immunohistochemical stain-
ing raised the concern that high levels of NRTN might be
toxic to nigrostriatal neurons. We launched an investigation
to better understand the phenomenon and determine
whether it represented a liability that precluded the contin-
ued development of AAV2-NRTN (CERE-120). The integ-
rity of the dopamine neurons was further evaluated by using
a different immunohistochemical marker for dopamine neu-
rons (vesicular monoamine transporter type 2; VMAT-2).
We observed no change following any CERE-120 dose or
level of NRTN expression, establishing that the dopamine
neurons were still alive but simply produced less TH. A
neurochemical analysis of striatal dopamine and dopamine
metabolites was then performed (using fresh tissue), which
revealed a small but significant increase (37%) in the ratio
of the dopamine metabolites (HVA (homovanillic acid) and
DOPAC (3,4-dihydroxyphenylacetic acid)) to dopamine
levels in the CERE-120-treated rats, as well as a small but
statistically significant decrease (21%) in absolute levels of
dopamine, compared with formulation buffer control rats.
These data suggested a modest increase in the rate of do-
pamine turnover and enhanced dopaminergic activity in the
nigrostriatal system of these CERE-120 treated rats. Simi-
larly, performance on a number of behavioral tasks that
have been shown to be dependent on the status of the
nigrostriatal system (e.g., adjusted forelimb stepping in re-
sponse to movement, vibrissae-elicited forelimb placing,
and plank and rod walk tasks) was examined in rats 6
months after CERE-120 administration to the striatum. We
observed no differences between CERE-120 treated and
formulation buffer control treated rats in any of these be-
haviors, despite confirming reductions in TH-immunoreac-
tivity in their nigrostriatal neurons. These data comple-
mented the histological findings, further confirming the
functional integrity of the nigrostriatal system, despite re-
duced TH immunoreactivity following delivery of high dose
CERE-120 for 6 months in rats (Ceregene, Inc., unpub-
lished results).
Finally, the results in rats contradicted observations in
monkeys, where several studies demonstrated enhanced in-
tensity of TH staining following CERE-120 administration,
both in the nigra as well as the striatum of nonhuman
primates. This enhanced TH signal was observed well
within the temporal window of when a decreased signal is
seen in rats (i.e., 3 months) and occurred consistently across
a 60-fold range of CERE-120 doses (Herzog et al., 2007,
2008, 2009). Thus, in contrast to the observations following
a high-dose of CERE-120 in rats, in no case was a decrease
in TH staining ever observed following CERE-120 in non-
human primates, suggesting this was a strange but relatively
unimportant species-specific phenomenon.
At the time we were conducting our studies, three pub-
lications emerged that directly addressed the same phenom-
enon in an extremely informative manner (Georgievska et
al., 2004a, 2004b; Rosenblad et al., 2003). These studies,
using GDNF (as opposed to ours, using NRTN), reported
results very similar to our unpublished findings, with even
more compelling evidence that the phenomenon was a tran-
sient (fully reversible) (Georgievska et al., 2004a), compen-
satory response to increased dopamine activity following
delivery of the neurotrophic factor. These beautifully con-
ceived studies persuasively argued that delivery of neu-
rotrophic factors, such as GDNF (and by association,
NRTN), to young healthy rats causes a transient increase in
dopamine activity, requiring a compensatory decrease in
TH activity in order to maintain normal levels of dopamine
activity. Thus, the reduction in TH staining is far from a
toxic reaction but rather a normal, species-specific response
to enhanced dopamine tone, providing welcomed support
that the downregulation we had unexpectedly observed with
our high dose of CERE-120 in rats was not an issue that
posed any significant risk to PD patients.
We were also confronted with concerns regarding pos-
sible cerebellar toxicity but not due to anything we observed
with CERE-120 and NRTN, but rather related to general
concerns about GFLs being neurotoxic (NRTN and GDNF
are both GFLs). Around the time we were preparing docu-
mentation for the Food and Drug Administration (FDA)
prior to filing our initial CERE-120 IND, we became aware
of reports of Purkinje cell loss and associated astrocytosis in
the cerebellum in several monkeys receiving high dose
infusions of GDNF (infused as purified protein into the
putamen via indwelling cannula). The study was sponsored
by Amgen, who suspended their program in the light of
these observations, and the generation of neutralizing anti-
bodies to GDNF in both monkeys and PD subjects (Hovland
et al., 2007; Lang et al., 2006; Slevin et al., 2005), as well
as disappointing Phase 2 efficacy data (Lang et al., 2006).
Clearly, the observations of possible toxicity following
chronic infusions of GDNF warn that similar toxicity could
occur with chronic expression of NRTN and we therefore
closely examined the GDNF reports, as well as our own
high dose CERE-120 pathology slides to determine whether
11R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 11
a similar safety liability existed for NRTN expressed via
administration of CERE-120.
It soon became clear that the most likely explanation for
both the cerebellar pathology and the development of anti-
bodies was leakage of the GDNF protein into the CSF or
blood from the infusion pump implanted under the skin
and/or the cannula implanted in the striatum. This conclu-
sion was supported by the fact that recombinant GDNF
leaked into the CSF in a dose-related manner and a signif-
icant number of monkeys in the high dose group even had
detectable recombinant GDNF in their plasma. Equally im-
portant, dose-related signs for mistargeted trophic factor in
periventricular sites was also observed, including weight
loss as well as meningeal thickening in the medulla oblon-
gata and segments of the spinal cord. We observed none of
these reactions following high dose CERE-120 administra-
tion and subsequent chronic NRTN expression, including
no NRTN ever detected in serum or CSF, even at excessive
doses of CERE-120 or after a full year of continuous NRTN
expression. No antibodies to NRTN have ever been detected
with any dose of CERE-120 or at any time point, and none
of the classic signs of nontargeted neurotrophic factor ex-
posure to periventricular sites have been observed. Most
importantly, no evidence of any histological changes was
noted in the cerebellum or in any other brain region or body
system for that matter. While it is still not clear what
mechanism(s) might be responsible for the constellation of
safety issues caused by the high dose of GDNF in monkeys,
we concluded that the safety issues could collectively and
parsimoniously be linked to deficiencies in protein delivery
using indwelling hardware, leading to leakage of the protein
to untargeted and undesirable sites (e.g., periventricular
regions and possibly systemic exposure). This leakage of
putaminally-targeted GDNF ironically induced the same
unwanted neurotrophic response in the periventricular sites
previously reported when neurotrophic factors were inten-
tionally infused into the lateral ventricles (Eriksdotter Jön-
hagen et al., 1998; Nutt et al., 2003). Our conclusion was
independently supported by another laboratory reporting
that similar infusions of radiolabeled GDNF into the mon-
key putamen produced leakage of GDNF to superficial
layers of the occipital cortex and cerebellum (Salvatore et
al., 2006).
In conclusion, while it is possible that GDNF possesses
a much less desirable safety profile than NRTN, a more
likely explanation for all the data are that gene transfer (or
gene therapy), at least as described and used here, provides
a relatively safer and more effective means of delivering
proteins than does chronic infusions of protein via mechan-
ical hardware. Clearly, a large part of the successful safety
profile enjoyed by CERE-120 relates to the delivery ap-
proach that relies on use of modest convection currents,
coupled with strategically placed infusion sites within the
targeted brain sites (thus limiting uncontrolled spread of
protein along white matter tracts well-established for more
aggressive CED approaches; e.g., Ding et al., 2010, Lamp-
son 2001, Linninger et al., 2008).
4. Creating a paradigm for selecting initial CERE-120
doses to test in PD subjects
As reviewed in prior sections of this article, we had
successfully addressed all the conventional translational is-
sues, as well as many unique and sometimes unexpected
issues required to move a novel treatment into human test-
ing. However, an important issue that still remained to be
tackled before we could actually proceed with testing in
humans involved selection of the CERE-120 doses to be
tested. This question presented a particularly onerous chal-
lenge because CERE-120 is a far different therapeutic agent
than the vast majority of treatments, whether already ap-
proved or still in development. Because of these differences,
there was really no clear guidance or “blueprint” to direct
our thinking as we defined the initial “first-in-human”
More specifically, most novel treatments are either in-
gested (orally) or administered parenterally (e.g., subcuta-
neously, intramuscularly, or intravenously). These routes of
administration permit investigators to use conventional
pharmacokinetic (PK) methods to track the performance of
a novel therapeutic agent, measuring when peak plasma
levels are reached following administration, when and what
other “compartments” (or organ systems) the agent tends to
migrate to and/or accumulate in, how long it persists, and
when and how it is cleared (e.g., breakdown by liver, fil-
tered by kidneys and excreted in urine, etc.). Such PK
information can be obtained in animal tests and used to
define initial (starting) doses for testing in human volun-
teers. When human testing begins, further PK information is
obtained, and by comparing with prior data from animals,
adjustments to doses can be rationally and easily made. By
linking this information to surrogates for efficacy in human
volunteers, investigators can even project doses likely to be
therapeutic. In any case, doses can be “ramped up,” begin-
ning with intentionally low (even subtherapeutic) doses to
assure no serious safety problems are induced and gradually
increase the dose to achieve likely therapeutic levels.
This situation with CERE-120 was far more difficult and
complicated. Because we are specifically targeting CERE-
120 to a select site within the human brain, it is not possible
to monitor dose levels via PK methods, as conventionally
done. Moreover, no surrogate is even available to determine
the distribution of AAV vector following injection or to
confirm NRTN expression in the living human brain. Thus,
monitoring results for the purpose of “dose escalation” are
far more difficult and less effective with gene therapy, as
compared with conventional, systemically administered
agents. Another major difference is that contrary to nearly
all “first in human” tests of novel products, gene transfer
studies do not allow the opportunity to withdraw the biop-
12 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 12
harmaceutical agent, for once the vector is administered, it
will likely express the transgene protein for the life of the
transduced cells. Clearly, the stakes for doing unexpected
harm are much greater. Another important difference is that
contrary to the vast majority of all new agents being eval-
uated, gene therapy moves directly from tests in animals to
tests in human patients (in this case, moderately advanced
PD patients who are recruited to volunteer as a subject in a
research clinical trial by an investigator affiliated with one
of the university hospitals serving as a site for the study).
Initiating human dosing in patients (even if technically
defined as “subjects”) creates additional issues to consider,
which in turn are further complicated by the fact that ad-
ministration of CERE-120 requires an invasive procedure
(i.e., stereotactic brain surgery). Placing patient volunteers
at the risk of injury from the surgical procedure raises
additional ethical issues (and regulatory guidance) that ex-
ceeds that when nonpatient volunteers are involved in rel-
atively noninvasive trials. For example, one cannot initially
test intentionally very low doses (to help assure no initial
safety issues) and then gradually increase the dose to ap-
proach therapeutically relevant levels, as is typically done
for small molecules. Rather, the risk of surgery to the
subject requires that all doses administered have at least a
rational chance of providing some efficacy. Thus, the need
to test doses sufficiently high to project some likely benefit
runs completely at odds with the competing desire to begin
doses sufficiently low to avoid any unexpected (and possi-
bly irreversible) side effects or toxicity.
One characteristic of CERE-120 that provided a major
advantage to selecting doses for humans involved the fact
that the nonclinical testing consistently showed that CERE-
120 was very safe, so much so that we could not produce
any side effect or evidence of any toxicity at any dose or
time point, even when neurons and glia were examined
under high power microscopy directly at the point of ad-
ministration, where NRTN levels were estimated to be ex-
tremely high. Thus, these data argued that we did not need
to concern ourselves with expressing too much NRTN at
any given location, or be concerned with a “therapeutic
window” for CERE-120 (see Fig. 2a). Rather, we merely
needed to focus on selecting doses that might adequately
cover the targeted putamen, without causing NRTN expres-
sion to spread outside its borders. The latter point (i.e.,
limiting expression to the targeted putamen) seemed to be
especially important, because prior studies infusing neu-
rotrophic factors in both humans and primates warned that
serious side effects could be induced if high concentrations
of the protein were exposed to periventricular tissue. It was
for this reason (as the first gene therapy/neurotrophic factor
program to begin translating work in animals to the human
clinic), that we necessarily adopted an appropriately con-
servative dosing strategy. This strategy employed more
modest convection currents (e.g., infusion rates of 2
L per
minute), coupled with strategically distributed infusion
points within the targeted brain sites. While the number of
needle tracts was greater than that required for more aggres-
sive CED, we avoided the notorious issue of uncontrolled,
mistargeted protein spreading along paths of least resistance
(especially white matter tracts) known to regularly occur
with CED (e.g., Ding et al., 2010; Lampson, 2001; Lin-
ninger et al., 2008). A clear advantage of our dosing strategy
and its translation to the clinic was that tests in rats and
monkeys established that we could reasonably predict and
control the volume (and amount) of protein expressed by
varying the absolute number of vector genomes (i.e., copies
of NRTN gene) of CERE-120.
We therefore developed a unique approach to dose se-
lection, using extensive dose-range testing in animals to
project the volume of NRTN expression likely to occur in
the human brain. We then used human magnetic resonance
images from PD patients and human histological sections to
construct a 3-D model of the human putamen within the
human brain with the aid of computer software (Fig. 2b,
left). This computer model then allowed us to visualize how
to maximally cover the putamen by adjusting dose level and
location of injection sites (while attempting to not exceed its
boundaries) simply by distributing varying volumes of
NRTN (linked to varying hypothetical doses of CERE-120)
within the putamen (Fig. 2b, right). Our goal was simply to
cover as much of the putamen as possible with as few
needle tracts as possible. In other words, rather than at-
tempting to establish the “correct” or most desired concen-
tration or level of NRTN in the targeted tissue, or worrying
about whether too much protein might be expressed within
a subregion of the targeted tissue, we established a dosing
paradigm to reflect the pharmacological properties of
CERE-120, focusing on the volume of NRTN expression
and its location.
The information that permitted us to estimate volumes of
NRTN expression that should occur in humans, per hypo-
thetical CERE-120 dose, was derived from another method
we applied to dose selection: “interspecies” scaling (Bartus
et al., 2011a). This method: (1) used the dose response data
we generated in monkeys (and also to some extent, rats),
which (2) informed us as to what volume of NRTN should
be achieved per dose injected, which (3) when combined
with knowledge of differences in the volumes of the tar-
geted region for each species (i.e., monkeys, rats, and hu-
mans), allowed us to (4) calculate the total dose required to
maximize coverage of the targeted site for that species,
while avoiding substantial NRTN expression outside the
targeted site (Fig. 2c and Table 3). Of course, given that we
had not yet collected any evidence that vector spread and
protein expression in human brain would precisely follow
that established for monkeys (i.e., we needed to consider the
possibility that it might spread further), we selected 2 rea-
sonably conservative doses for our initial Phase 1 trial,
estimated (based on animal expression data) to likely cover
between 25% and 40% of the putamen (note that 100% of
13R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 13
Fig. 2. Unique paradigm applied to selecting doses for human testing. (A) Experiments using CERE-120 do not provide the conventional pharmacokinetic
(PK) data most often used to help select doses for humans, while results with CERE-120 demonstrated that, when delivered to the brain, it did not follow
conventional dose–response patterns seen with systemically administered drugs (e.g., no maximum tolerated dose could be defined); therefore, a new
paradigm was required for selecting doses for human testing. Importantly, because we could find no evidence that neurturin (NRTN), when properly targeted within
the nigrostriatal system, had any deleterious or toxic effect at any dose, we developed confidence that excessive NRTN expression within the targeted putamen
was not likely a major safety concern. However, several accounts in the published literature indicated that neurotrophic factor protein expression exceeding
14 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 14
the putamen cannot be targeted in any case, because por-
tions of it become too narrow, particularly its tail). Later
quantification of NRTN expression from CERE-120-treated
human autopsy cases conservatively estimated putaminal
coverage to be approximately 15% (Bartus et al., 2011b).
However, it is not known whether AAV and/or NRTN
diffuse less widely in human (compared with nonhuman
primate) brain or alternatively whether the relatively poor
condition of the immersion-fixed, postmortem human au-
topsy tissue might impact the ability to visualize NRTN.
This issue is likely further exacerbated by the tissue damage
necessitated by the harsh antigen retrieval immunohisto-
chemical staining techniques required to visualize NRTN
(i.e., heating the tissue), coupled with our dependence on a
relatively poor anti-NRTN antibody. Therefore, for all these
reasons, it remains unclear what the true quantitative NRTN
coverage in the human brain might be, though there are
logically sound and empirically supported reasons to be-
lieve it is very likely greater than the 15% currently con-
servatively estimated. While a more definitive answer
awaits the development of better antibodies to NRTN and
other methodological advances, it is nonetheless the case
that significant improvement in the function of degenerating
nigrostriatal systems and motor behavior can be achieved in
animal models with less than 15% striatal coverage of
neurotrophic factors (see Table 2), though it remains to be
seen what coverage might be required in PD patients. In any
case, the current CERE-120 clinical program has since in-
creased the putaminal dose by nearly 4-fold (based on
additional safety data collected since the launch of the initial
clinical study in 2005) as well as added CERE-120 delivery
directly to the degenerating dopamine neurons in the SN.
Based on recent data in animals (Ceregene, Inc., unpub-
lished), we estimate we should nearly cover all targetable
portions of both the putamen and the SN with NRTN,
assuming reasonable anterograde and retrograde transport is
gradually restored in the PD patients following the expres-
sion of NRTN in both components of these neurons (see
later text).
The dose scaling method also provided an objective
means for assuring that the human dose would be within the
range shown to be biologically effective in animal models,
and still be well below the highest safety/toxicity dose
tested in animals (see Table 3). This method provided a
somewhat independent means of confirming that the doses
selected using the 3-D model (described, above) were ap-
propriate, based not on projections of protein expression,
but rather on empirical safety/toxicity and efficacy dose–
response data. Thus, as shown in Table 3, the 2 initial Phase
1 human doses selected were 11 to 21 times lower than dose
equivalents shown to be safe in rats and monkeys, respec-
tively. At the same time, both doses were well within the
dose range shown to be efficacious in 6-OHDA (6-hydroxy-
dopamine) rats, while the higher of the 2 doses approached
the single dose tested and shown to be effective in MPTP
(1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) monkeys (i.e.,
it was 54% of the MPTP monkey dose, by volume of
target). However, as discussed below, with the cumulative,
long-term safety data generated in monkeys since the initi-
ation of the CERE-120 clinical program, and the additional
safety data generated in the initial Phase 1 and Phase 2a
clinical trials, we have since further increased the CERE-
120 dose to the putamen so that in the current Phase 2b trial,
the human dose is nearly twice the comparable MPTP dose
and 64% greater than the highest 6-OHDA comparable
5. Tests of CERE-120 in Parkinson’s subjects
establishes first evidence for “clinical proof-of-
concept” for neurotrophic factors for age-related
neurodegenerative diseases
5.1. Initial open-label, Phase 1 trial
Following the successful execution of the nonclinical
program (previous sections), an initial Phase 1 safety trial in
moderately advanced PD patients was launched. Two dose
levels of CERE-120 were tested (1.3 10
vg, or copies of
the NRTN gene and 5.4 10
vg, total) for both hemi
spheres. Six subjects were enrolled into each cohort with
doses delivered bilaterally into the terminal fields of the
degenerating nigrostriatal neurons in the putamen. Formal
safety (and preliminary efficacy) evaluations were con-
ducted every 3 months for 12 months, after which all sub-
the boundaries of the targeted brain site could produce side effects and therefore did pose a point for concern. Thus, rather than focusing on conventional
dose–response issues of maximum tolerated dose (based on levels of protein expression), we focused on the location of protein expression, helping to assure
NRTN was exposed to the targeted site, while avoiding NRTN expression in other, nontargeted brain sites. (B) Using a computer graphics program
(Neurolucida, MicroBrightField, Inc.;, along with magnetic resonance imaging images and histological slides of
human brain, we constructed a 3-D model to visualize the human putamen (left) and help determine the amount and location of NRTN expression that would
provide the greatest coverage, with the fewest needle tracts, while avoiding protein expression outside its boundaries (right, and panel C, left). (C) Comparison
of 2 different dosing schemes used in CERE-120 clinical trials. Left: 4 separate needle tracts (2 deposits per tract) were used to distribute CERE-120 into
the putamen (only) of each hemisphere in the initial Phase 1 and Phase 2a studies. Right: 3 separate needle tracts (single deposit per tract) were used to
distribute a 4-times higher dose into the putamen of each hemisphere, as well as an additional needle tract delivering CERE-120 directly to the substantia
nigra of each hemisphere in the revised Phase 1 and Phase 2b studies. In all cases, all dosing per each hemisphere was done via a single burr hole in the
skull. (D) Examples of magnetic resonance images of the human nigrostriatal system illustrating stereotactic targeting of CERE-120 to the putamen (left
panel: 1 of 3 targets in the right hemisphere) and substantia nigra (right panel: site 1 of 2 deposits along a single needle track in the right hemisphere), applying
the information reflected in (A) and (B) to establish stereotactic coordinates and projection paths for dosing CERE-120, as depicted in (c).
15R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 15
jects were enrolled into a long-term follow up protocol for
continued monitoring for 5 years, post CERE-120 adminis-
tration. Two movement disorder centers participated in the
trial (recruiting, enrolling, and screening subjects, perform-
ing the stereotactic surgery required for dosing CERE-120,
and making the routine safety and efficacy assessments:
University of California, San Francisco, and Rush Univer-
sity Medical Center in Chicago; for key personnel, see
Acknowledgements). No serious adverse events were noted
during the entire duration of the trial, providing initial evi-
dence for the safety of CERE-120 in humans, corroborating
the safety record previously established in animal studies.
Preliminary evidence for clinical improvement at 12 months
post dosing was also observed (p 0.001) on a number of
clinically relevant motor and quality-of-life end points
(Marks et al., 2008), including the prespecified primary end
point (Unified Parkinson’s Disease Rating Scale [UPDRS]
motor-off) and several hierarchical, secondary end points
(e.g., self-report diary scores). Though this represented the
largest open label study with a neurotrophic factor tested in
an age-related neurodegenerative disease to suggest clinical
benefit, we recognized the tentative nature of these uncon-
trolled, open-label data, and the still-relatively small num-
ber of subjects tested. We therefore moved quickly but
responsibly to initiate a randomized, double-blind Phase 2a
trial to more effectively establish the safety and more rig-
orously test for possible efficacy.
5.2. Multicenter, double-blind, sham surgery-controlled
Phase 2a trial
The randomized, sham surgery-controlled, double-blind
Phase 2a trial tested similarly moderately advanced PD
subjects (n 58; mean [ SD] age 59.1 [ 7.86]; 74.1%
males; H&Y (Hoehn and Yahr scale) off 3.03 [ 0.63];
H&Y on 2.02 [ 0.57]), as recruited for the initial Phase
1 trial. The higher of the two Phase 1 doses (i.e., 5.4
vg) was tested in all subjects in the treatment group
(treated to sham 2:1 distribution). Nine leading move-
ment disorder sites in the United States participated (see
This initial controlled trial with CERE-120 had three
overlapping but clearly distinguishable and escalating goals.
The first goal involved expanding and enhancing the safety
database. While the initial, 12-subject, “first-in-human”
Phase 1 trial with CERE-120 revealed no safety issues a
year after CERE-120 dosing, it was now important to test
even greater numbers of subjects using a greater variety of
investigators and neurosurgeons over a wider and more
geographically and culturally diverse area.
A second goal of this trial involved expanding the pre-
liminary evidence of efficacy generated in the Phase 1 trial.
This involved generating data, under double-blind, con-
trolled conditions that might establish initial “clinical proof-
of-concept” that CERE-120 can provide objectively-de-
fined, reliable, and meaningful benefit to Parkinson’s
patients. While definitions for “clinical proof-of-concept”
vary among published studies, the vast majority of pub-
lished accounts with which we are familiar include evidence
of clear clinical benefit often (but not always) involving a
randomized, controlled clinical design. The latter require-
ment seems particularly important for CNS indications and
essential for surgical-based therapies (e.g., see arguments
raised by Alterman et al., 2011), including PD. Moreover,
given the subjective nature of most of the validated mea-
sures used to assess severity of PD symptoms, it also seems
important that improvement be seen on a number of differ-
ent, clinically relevant, motor and health-related, quality-of-
life assessments that provide a greater level of confidence
via mutually corroborating evidence. Because no controlled
clinical trial had yet demonstrated clinical benefit for any
neurotrophic factor for any age-related neurodegenerative
disease, this goal represented an important and formidable
feat for this first, controlled trial in the CERE-120 program.
Generating efficacy data for registration purposes repre-
sented the third and final major goal for this study. The
efficacy data required for establishing “clinical proof-of-
concept” can be clearly distinguished in the literature from
so-called “pivotal trials” intended to generate efficacy data
specifically to be used for registration purposes with the
FDA. Establishing efficacy for FDA registration purposes
requires a “positive” (i.e., statistically significant) outcome
for a prespecified primary measurement and time point,
using a statistical analysis defined in advance of breaking
the blind. Other clinically relevant motor and quality-of-life
end points (and time points) are relegated to “secondary”
status. Importantly, this regulatory requirement explicitly
prevents investigators from engaging in post-hoc “data min-
ing” at the conclusion of pivotal trials, where only those
measurements, end points, and statistical analyses produc-
ing the desired results are acknowledged, while the more
disappointing outcomes are ignored. Clearly, such selective,
post hoc focus on favored end points is inappropriate,
whether for purposes of supporting “clinical proof-of-con-
cept,” or for registration purposes.
Thus, in the event the efficacy results from our first
controlled trial with CERE-120 were truly spectacular, we
added elements to the protocol that would allow the data to
be used for registration purposes with the FDA (i.e., for
eventual inclusion in an application requesting the FDA to
approve CERE-120 for use as a new treatment for PD).
Clearly, this goal was much bolder and more difficult than
the others, particularly because this represented the first-
ever controlled trial with CERE-120. As stated, evidence for
efficacy for registration purposes required that the trial
achieve a sufficiently robust and statistically significant ben-
efit on a single, predefined “primary endpoint” at a prede-
termined, single point in time and we selected the motor
component (Part 3) of the UPDRS in the practically defined
off state (i.e., no Parkinsonian medications for at least 12
hours), measured at 12 months post-treatment. For the spe-
16 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 16
cific purposes of generating data to file with the FDA, all the
other measurements and time points were merely used to
help characterize the clinical benefit, in the event the pri-
mary end point showed a statistically significant effect,
while also providing some important confirmation of gen-
eral clinical improvement. Additionally, the data from these
secondary measures can sometimes also be used later for
developing and supporting the language of the label with the
FDA, in the event a statistically significant effect on the
primary end point was achieved. Thus, the primary end
point is the most important for purposes of generating reg-
istration data for the FDA.
5.3. “Clinical proof-of-concept” for therapeutic benefit of
CERE-120 in PD: results from a controlled Phase 2a
The first controlled trial of CERE-120, testing a neu-
rotrophic factor against age-related PD, generated several
important new findings. First, the safety of the stereotactic
dosing procedure, the use of NRTN gene transfer and the
continuous NRTN exposure to the brains of Parkinson’s
patients were all clearly supported, extending the prelimi-
nary safety data collected in the initial Phase 1 trial. Com-
pared with the Phase 1 trial, many more subjects were added
to the safety database, involving far more investigators,
neurosurgeons, and clinical sites, with no unexpected tox-
icity or side effects observed following CERE-120 admin-
While the “primary endpoint,” defined for purposes of
FDA registration (i.e., UPDRS Part 3; motor-off at the
12-month time point) showed no statistically significant
difference between the improvement that occurred in both
the CERE-120 and sham groups, many other end points did
favor CERE-120 with little evidence for a similar effect of
the sham group. Indeed, clear evidence for “clinical proof-
of-concept” was seen following CERE-120 administration,
based on consistently superior performance of CERE-120
relative to the sham group. For example, at the 12-month
protocol-specified evaluation period, when all 24 prespeci-
fied efficacy end points used in this trial were considered, 19
end points favored CERE-120 (compared with the sham
control group). If all end points are considered independent
from each other, the probability of this outcome occurring
by chance is extremely low (p 0.007, 2-sided exact sign
test). While it is unlikely that all the end points are literally
completely independent of each other, they clearly are not
entirely “dependent” or related to each other either, for
many can deteriorate differentially in different patients as
the disease progresses; thus, the p value may be difficult to
define precisely, but the probability of this event occurring
by chance is nevertheless quite low. Moreover, the 5 of 24
measures that did not favor CERE-120 (as well as a few of
those that did) represented clinically trivial differences, well
within the noise of those measurements and therefore truly
reflected no difference between groups. Thus, this analysis
indicates a positive clinical effect of CERE-120 that is
difficult to attribute to chance. We also prespecified a select
number of end points for a more focused pairwise statistical
analysis based on guidance from results from the initial
open-label Phase 1 trial, along with clinical judgment as to
which end points were more clinically relevant to PD status
and likely to be improved by NRTN expression targeted to
the nigrostriatal system (predefined in the Statistical Anal-
ysis Plan (SAP)). These focused end points included the
UPDRS, diary scores, PDQ-39 (Parkinson’s disease ques-
tionnaire-39), and seven meter, timed walking test-off. We
focused on only 2 time points: 12 months (the protocol-
specified time point for the primary end point) and 18
months (the last data point for which we have blinded data
in a reasonable number of subjects). When these individual,
a priori-defined “focused” end points were tested, many
showed a statistically significant benefit from CERE-120
when compared with the sham group. Importantly, no mea-
sure showed a similar statistical advantage for the sham
surgery group (Table 5). The fact that the vast majority of
end points favored CERE-120 with the overall analysis, and
many focused pairwise comparisons suggested statistically
significant benefit from CERE-120 (and none for sham),
provided initial evidence for “clinical proof-of-concept”
that NRTN can improve the symptoms of Parkinson’s pa-
Additional evidence supporting “clinical proof-of-con-
cept” for CERE-120 was obtained from additional, pre-
defined analyses of data from the subset of patients who
remained blinded beyond 12 months (roughly the initial
50% of the subjects randomized for the protocol). Impor-
tantly, these analyses were also prescribed in the formal
SAP (statistical analysis plan) well in advance to complet-
ing the study and therefore long before breaking the blind.
An analysis of these longer-term treated subjects revealed
an increasingly greater clinical benefit from CERE-120.
First, 14 out of all 15 of the prespecified end points favored
CERE-120 (note: fewer end points were assessed beyond 12
months), a distribution of effects favoring CERE-120 that,
once again, was very unlikely to occur by chance (p
0.001, two-sided exact sign test). Furthermore, when the a
priori-defined “focused” end points (see above) were tested,
even more now suggested statistical significance, including
the protocol-specified primary efficacy measure (UPDRS
motor-off) when assessed at 18 months, postdosing (p
0.023) (Marks et al., 2010)(Table 5 and Fig. 3a). As was the
case for the 12-month prespecified analysis time point, no
end point showed a similar statistical advantage for the
sham surgery group at the 18-month, prespecified time
point. It might also be worth noting that other prespecified
end points also showed statistically significant benefit of
CERE-120 (e.g., Clinical Global Impressions, subject and
investigator rated scales; timed walking-on), but because
they were not defined in advance as among the select end
points to be included in our “focused statistical analyses”
17R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 17
(see above), these data are not shown. Certainly, some
caution should be exercised regarding specific interpretation
of fine details, for it is not possible to ascertain the extent to
which each motor and quality-of-life efficacy measurement
tested in this protocol is independent from each other and
thus the precise p value should not be taken literally. Similar
caution should be exercised when considering any of the
specific pair-wise comparisons and those p values should
not be overinterpreted for it was not possible to correct for
multiple comparisons. Nonetheless and more importantly,
the weight of all these findings, and the very low probability
that all these measurements could favor CERE-120 over
sham by chance (with no evidence for similar benefit of the
sham group), inevitably forces the conclusion that CERE-
120 indeed provided some reliable clinical benefit (albeit
incompletely defined) over the entire course of blinded
assessments, with the likely scenario that this benefit was
even greater as more time elapsed following treatment with
In conclusion, though this first controlled study with
CERE-120 did not show a treatment benefit on the primary
end point at 12 months (and therefore the data from this
study were not immediately useful for FDA registration
purposes), almost all of the end points favored CERE-120
over sham and many clinically important, predefined end
points did suggest a statistically reliable benefit of CERE-
120, including the same measurement used for the primary
end point, when evaluated after the longer, post CERE-120
interval of 18 months. These data therefore demonstrate
“clinical proof-of-concept” for CERE-120, based on the
number of blinded end points showing a benefit of CERE-
120 (with no end point showing sham surgery was mean-
ingfully better) and the increasing benefit seen with the
analysis of the 18-month data that remained blinded. At the
same time, these data more generally provide the first ever,
properly controlled, “proof-of-concept” evidence that neu-
rotrophic factors can improve the clinical status of an age-
related neurodegenerative disease. Of note is an autono-
mous editorial written by a PD pioneer in the development
of DBS which accompanied the publication of these data.
This editorial clearly and independently supported the con-
clusion that the data from this Phase 2 trial establish “proof-
of-concept,” stating: “The findings of Marks and colleagues
provide the first clinical evidence of a clinical benefit of
gene therapy in PD; these results will serve as a starting
reference that it is hoped will be exceeded in future trials”
(Benabid, 2010). While we appreciated this independent
support for establishing “proof-of-concept,” we also agreed
that these results should be viewed as a mere starting point
for the clinical development of this concept, because the
magnitude of the clinical benefit achieved, as well as the
delayed time course to see even greater benefit, suggested
that the neurotrophic effect had not yet been optimized.
Fortunately, autopsied brains of 2 subjects who were ad-
ministered CERE-120 as part of this trial, and later died of
unrelated causes, provided essential insight for guiding our
attempts to further enhance the neurotrophic effects of
NRTN and therefore the magnitude of the clinical benefit.
5.4. Analysis of autopsy tissue from Parkinson’s subjects
previously administered CERE-120 provided important
and unique insight
Two subjects treated with AAV2-NRTN during this trial
died of unrelated causes 1.5 and 3 months later. An analysis
of the brains they and their families donated for further
study not only provided further “proof-of-concept” evi-
dence for NRTN’s benefit in PD, but just as importantly,
provided unique insight essential for optimizing the target-
ing and dosing of CERE-120 in moderately advanced PD
Table 5
For purposes of testing clinical proof of concept, an overall analysis involving all 24 efficacy end points employed in the CERE-120 double-blind,
controlled Phase 2a trial established that 19 favored CERE-120 at 12 months (p 0.007) and all but 1 favored CERE-120 at 18 months (p 0.001)
12-mo evaluation 15- to 18-mo evaluation
Sham surgery: change
from baseline
CERE-120: change
from baseline
p value at
12 mo
Sham surgery: change
from baseline
CERE-120: change
from baseline
p value at
18 mo
UPDRS I “off” 0.95 0.32 0.01
1.27 0.26 0.02
UPDRS II “off” 2.25 3.35 0.4 0.82 3.32 0.07
UPDRS II “on” 1.6 0.89 0.03
Not tested
UPDRS III “off” 6.95 7.19 0.9 5.64 11.21 0.02
PD diary “off” 0.23 h 1.00 h 0.07
0.52 h 1.48 h 0.09
PD diary “on without troubling
0.80 h 1.00 h 0.3 0.55 h 2.25 h 0.05
Timed walking “off” 3.00 s 2.65 s 0.6 0.55 s 8.11 s 0.02
PDQ-39 1.20 2.83 0.03
Not tested
Certain “motor-related” and “quality-of-life” (QOL) end points were defined in advance to allow for select or “focused” statistical pairwise comparisons; these
motor and QOL measure end points were limited to the UPDRS, self-report diaries, 7-meter timed walking test (off only), and PDQ-39. Pairwise comparisons
on these selective, predefined end points suggesting a benefit of CERE-120 at 12 months and even more at 18 months. For trend and statistical significance,
not a single measure suggested a similar benefit of sham. See text for more details. Note: top line data adapted from Marks et al, 2010.
Key: PD, Parkinson’s disease; UPDRS, Unified Parkinson’s Disease Rating Scale, PDQ-39, Parkinson’s Disease Questionnaire-39.
Statistical significance; not corrected for multiple comparisons.
Statistical trend; not corrected for multiple comparisons.
18 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 18
patients (Bartus et al., 2011b). A number of new and im-
portant findings were noted from these brains. First, expres-
sion of NRTN in the targeted PD putamen following CERE-
120 delivery was confirmed, demonstrating that CERE-120
performed precisely as it was designed to do (Fig. 3b).
Secondly, a clear but modest increase in TH signal was also
observed in the PD putamen, which was topographically
linked to targeted NRTN expression. This observation of-
fered the first clear evidence that the status of degenerating
dopamine neurons can be enhanced in the moderately ad-
vanced PD brain (via the accepted functional surrogate of
TH immunohistochemistry; Fig. 3b). They therefore also
provided empirical evidence in human brain for the more
general, broader conclusion that neurotrophic factors can
indeed produce positive effects on degenerating neurons, as
predicted by decades of research in animals. Finally, they
Fig. 3. (A) Comparison of performance, over time, in CERE-120 treated subjects versus sham-surgery control subjects (all data blinded). Note emergence
of significant benefit of CERE-120 over time on the predefined “primary endpoint” by 18 months post dosing. Note that while this “primary endpoint” did
not demonstrate a clinical benefit from CERE-120 at 12 months, a protocol-prescribed analysis of longer-term, blinded data showed significant benefit at 18
months. These data, along with several other end points showing benefit of CERE-120 at the blinded 12- and 18-month time points (with no end point favoring
the sham control group) (Table 5), provide the first data generated in a controlled clinical trial that a neurotrophic factor can improve the clinical status of
an age-related neurodegenerative disease (data adapted from Marks et al., 2010). (B) Photomicrographs showing neurturin expression (NRTN; left panel) and
tyrosine hydroxylase induction (TH; right panel) in a Parkinson’s subject’s putamen following CERE-120 administration. Similar data exist for patients
treated between 6 weeks to 4.3 years post dosing prior to death from unrelated causes. These data provided the first evidence that vector-mediated in vivo
gene transfer can be used to express a neurotrophic factor in human brain (left panel). Moreover, expression of this neurotrophic factor was able to improve
status of degenerating dopamine neurons, as evidenced by enhanced TH fiber staining in neurons projecting from the substantia nigra (right panel). However,
the response observed was relatively weak and inconsistent (compared with that achieved in animal models), suggesting the neurotrophic response, while
important, was nonetheless suboptimal. Further evaluation of autopsy brains revealed relatively small levels of NRTN in the soma of degenerating neurons
following expression in the terminal fields, providing parsimonious explanation for suboptimal TH response and modest and delayed clinical improvement.
All these data support the growing recognition that Parkinson’s disease (PD) (like many age-related neurodegenerative diseases) is an axonopathy and that
serious deficiencies in axonal transport in PD occur long before the death of the neurons (data adapted from Bartus et al., 2011b).
19R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 19
provided further “proof-of-concept” support for CERE-
120, establishing that AAV-based gene transfer can pro-
duce targeted expression of NRTN in human brain and
that the NRTN expressed can then enhance the status of
degenerating dopamine terminals. Important corrobora-
tion of these interpretations were received via an inde-
pendent guest editorial that accompanied the publication
of these data, stating that this study provided “proof that
gene therapy with AAV2-NRTN (aka CERE-120) results
in functional transgene expression in target cells in hu-
mans” (Lewis and Standaert, 2011).
The analysis of these brains provided crucial, unique
information that gave us a better understanding of the status
of degenerating dopamine neurons in moderately advanced
PD. It also gave us the insight to devise a strategy to deal
with the deficiencies noted in the degenerating neurons in
order to (hopefully) maximize the neurotrophic response
from NRTN. More specifically, despite the confirmation of
NRTN expression in the targeted putamen, and the evidence
for enhanced TH in the terminal fields of the putamen,
another observation that was particularly striking was that
NRTN staining in the 1.5- and 3-month post CERE-120
brains was markedly sparse in the SN pars compacta (SNc),
where the degenerating cell bodies originate. This observa-
tion ran counter to results obtained from decades of inde-
pendent research with neurotrophic factors in animal mod-
els around the globe, including many more recent studies
with CERE-120 in nonhuman primates (see Bartus et al.,
2011b). Thus, the difference noted in those two early, post
CERE-120 PD brains was entirely unexpected and clearly
ran counter to well accepted dogma based firmly on the
volumes of data collected in numerous animal models. The
fact that NRTN expression in the targeted terminal fields
(i.e., the putamen) is not mirrored by corresponding NRTN
in the neuronal cell bodies (in the SNc) reveals a funda-
mental difference in the function of degenerating nigrostri-
atal neurons in moderately advanced PD versus that com-
monly seen in conventional animal models of PD. In the
latter instances, when a neurotrophic factor is administered
to, or expressed in the terminal fields, it is transported back
to the perikarya, eventually signaling cellular repair genes in
the nucleus to initiate a cascade of responses that help
strengthen and repair the cell. The lack of corresponding
NRTN in the soma of the PD dopamine neurons of the SNc
reveals an important difference that argues for a serious
deficiency in retrograde axonal transport in the degenerat-
ing, slowly dying neurons in moderately advanced PD that
is not currently captured in available animal models of the
disease. Indeed, if one were to design a test for the integrity
of retrograde transport in neurons of living PD patients,
expressing a neurotrophic factor in the terminal fields of
those neurons and later checking for histological evidence
of the protein being present in the cell bodies of those same
neurons might be one of the more conclusive (if not im-
practical) tests possible. While it seems likely that similar
deficiencies in anterograde transport also exist in these same
degenerating axons, which also would impact neuronal
function as well as the ability of treatments to improve that
function, we have not generated any data that directly ad-
dresses that possibility. In any case, the conclusion that
NRTN and/or CERE-120 is not adequately transported from
the dopaminergic terminals to the cell bodies is consistent
with a growing literature involving many age-related, neu-
rodegenerative diseases in humans (including PD) that ar-
gue that these diseases are actually axonopathies. That is,
the major pathogenesis begins with the long axons dying
back from terminal end toward soma, and in the course of
the disease, suffer serious deficiencies in axonal transport as
an early pathogenic event, long before the neuron dies
(Bartus et al., 2011b; Braak et al., 1999; De Vos et al., 2008;
Morfini et al., 2009; Raff et al., 2002; Roy et al., 2005).
Based on the insight gained from these histological anal-
yses, we concluded that delivery of CERE-120 directly to
the degenerating cell bodies (in the SN), as well as the
terminal fields in the putamen, would be required to cir-
cumvent the transport deficiencies suffered by these neurons
and thus maximize the neurotrophic responses to NRTN.
Moreover, in light of the emerging literature regarding ax-
onopathies, it seems very likely that our observations and
conclusions are not limited to NRTN or the nigrostriatal
system. Thus, this same analysis will most likely prove to be
true for any other neurotrophic factor (or vector expressing
a neurotrophic factor) intended to restore the function and
status of degenerating neurons in the majority of human
neurodegenerative diseases.
The concept of impaired retrograde transport limiting the
extent to which NRTN (and NRTN signaling) is able to
extend from the terminals of the degenerating neurons to
their cell bodies is entirely consistent with 2 other observa-
tions from our controlled Phase 2a trial. First, though
CERE-120 was able to produce clear evidence of im-
proved clinical status in the treated PD subjects, the
magnitude of the response was less than expected, but did
increase over time; this is precisely what would be ex-
pected in the face of suboptimal retrograde transport of
the neurotrophic factor delivered to the terminal fields of
degenerating neurons. Similarly, though we observed in-
duction of TH in the terminal fields of the putamen of the
PD autopsy cases, it was far less extensive and reliable
than that observed in prior primate studies with CERE-
120 following equivalent NRTN expression in the puta-
men (see Bartus et al., 2011b for extensive discussion of
this issue). This observation is consistent with the conclusion
that higher levels of NRTN are required in the SN cell bodies
to induce the desired, optimal neurotrophic response and thus
produce a satisfactory clinical response.
The loss of axonal transport efficiency and the inter-
pretation for its impact on CERE-120 efficacy recently
received independent confirmation with an analysis of
two additional, longer-term post CERE-120 autopsy
20 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 20
brains (Herzog et al., 2012) from patients treated with
CERE-120 4 and 4.3 years prior to dying from unrelated
causes. This analysis showed NRTN expression in the pu-
tamen similar to that previously seen in the 1.5- and
3-month post CERE-120 cases, but now also revealed clear
NRTN-positive neurons in the SN (while still less than
3%–5%, this is far greater than the 0.1% estimated for the
shorter-term expression cases). Even more intriguing is that
the area of the TH-induction footprint in the longer term
cases now more closely matches that of the corresponding
NRTN footprint (as opposed to the earlier cases where it
was much smaller). Collectively, these longer-term autopsy
data provide further, independent support for a transport
deficiency in moderately advanced PD patients while also
providing a corroborating biological correlate for the in-
creased and delayed clinical benefit seen following CERE-
120 treatment (i.e., Table 5 and Fig. 3). In addition to
providing independent corroborative evidence supporting
‘clinical proof-of-concept’, they provide further support for
the need to target the SN directly with neurotrophic factors
or vectors delivering the genes for these proteins. For all
these reasons, target and dose-optimization of CERE-120
was implemented for the subsequent new Phase 1/2b stud-
ies, specifically targeting the cell bodies in the SN with
6. Back to the laboratory as a prerequisite to
enhance the dosing strategy by targeting the SN with
Before we would seriously consider implementing the
novel dosing approach of targeting both the SN and puta-
men, we felt it prudent to carefully evaluate its potential
risks, relative to the intended benefits. The SN is a relatively
small structure that lies deep within the midbrain and there-
fore it presents a potentially more challenging target for
stereotactic surgery. Additionally, because it lies in close
proximity to other anatomical structures that mediate a
diverse number of functions, targeting the SN requires spe-
cial consideration and caution. Moreover, several investiga-
tors had recently warned that serious weight loss is likely to
occur if one attempts to target the SN with neurotrophic
factors such as GDNF and NRTN (Manfredsson et al.,
2009a, 2009b; Su et al., 2009). Others had also suggested
that neurotrophic influence on the ventral tegmental area,
which lies close the the SN and is also comprised of dopa-
minergic neurons, could lead to psychiatric symptoms, such
as addiction or psychosis (Lu et al., 2009).
Therefore, prior to advancing the CERE-120 program
forward, we paused to conduct a risk evaluation and miti-
gation strategy (REMS), which identified and then consid-
ered several potential risks associated with stereotactically
targeting the SN in humans with a viral vector intended to
chronically express the neurotrophic factor NRTN. The
main potential risks considered were the stereotactic surgery
involved with placing an infusion needle into the SN, the
exposure of SN neurons (and their afferents and efferents)
to continuous NRTN protein, and the possible exposure of
neurons far outside the targeted SN to NRTN, due to pos-
sible mistargeted injections or unexpected spread of vector
or NRTN from the targeted SN. This analysis was per-
formed by senior scientific and medical staff at Ceregene,
Inc., including Ceregene’s Scientific Advisory Board, along
with a number of key opinion leaders and consultants,
involving several neurosurgeons, neurologists, neuroscien-
tists, and psychiatrists familiar with midbrain dopamine
function (see Acknowledgements). The conclusions of the
REMS included an overwhelming consensus that targeting
the SN was appropriate and scientifically justified, that any
of the potential risks identified were hypothetical at most
(with no clear empirical data to support the concerns) and
that all were of very low probability. Moreover, in the very
unlikely scenario that one of the risk events were to occur,
it could likely be mitigated with standard medical treatment
(e.g., counseling, pharmaceutical intervention, etc.).
Nonetheless, to help assure the safety of PD subjects who
would be subjected to our revised dosing strategy, we per-
formed a series of additional studies in rats to further assess
the safety, feasibility, and effectiveness of CERE-120
(AAV2-NRTN) when stereotactically targeted to the SN,
thus inducing chronic NRTN expression in the midbrain.
These studies involved measuring a wide variety of end
points and testing a range of doses (including dose multiples
up to 200 times higher than required for the desired trophic
response) in order to help ascertain the consequences of
unintended protein expression far from the targeted SN
(Bartus et al., 2011a). These studies were also helpful in
providing more information to define an appropriate dose of
AAV2-NRTN that should safely and effectively cover the
SN in PD patients (using the same “dose scaling” methods
described earlier to translate animal doses for the human
putamen; see Table 3).
These studies confirmed that no CERE-120 dose directed
at the SNc produced any serious side effects or toxicity. A
dose-related increase in NRTN expression was seen, with
the lower doses successfully limiting NRTN to the targeted
peri-SN and the highest dose producing predictably mistar-
geted NRTN well outside the SN. We were also able to
replicate the reduction in weight gain reported by others
(but only with the highest, excessive dose) and demon-
strated that it can clearly be dissociated from NRTN in the
targeted SN and rather is linked to mistargeted NRTN in the
far remote diencephalon (and possibly the CSF) (Bartus et
al., 2011a). To further test for possible links between tar-
geting dopaminergic SN neurons with NRTN and the effect
on weight, we demonstrated that prior destruction of these
with 6-OHDA had no impact on the weight loss phenome-
non, arguing that exposing these neurons to the neu-
rotrophic factor was not the culprit for the weight changes,
while at the same time, further implicating mistargeted
21R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 21
protein. To help support part of the risk mitigation strategy,
we also showed that more than 50% of the effect on body
weight can be easily prevented with simple dietary supple-
mentation of preferred foods to rats. This provided further
confidence that this low hypothetical risk, if it were to occur
in humans, could very likely be mitigated with appropriate
instructions and dietary intervention. Finally, we demon-
strated that relatively low AAV2-NRTN doses provided
significant neuroprotection against 6-OHDA toxicity (com-
pared with the very high doses that had no toxicity), estab-
lishing a wide safety margin and therapeutic index for nigral
targeting with CERE-120.
In sum, the REMS we performed, along with the
generation of these new CERE-120 data specifically tar-
geting the SN in rats, provided clear support for the
safety and feasibility of targeting the SN with AAV2-
NRTN in PD patients. Based on this information, we
proceeded to advance this program with a revised clinical
protocol intended to optimize the targeting and dosing of
CERE-120 in PD subjects.
7. Launching a revised Phase 1/2b protocol to
optimize the targeting and neurotrophic effects of
Using the collective information and insight gained from
all our research with CERE-120, but especially the results of
the controlled Phase 2a trial and the autopsy cases, we
designed and implemented a new “target and dose-opti-
mized” Phase 1/2b protocol. A number of significant pro-
tocol improvements were incorporated in an attempt to
increase the rate and magnitude of the neurotrophic re-
sponse to CERE-120, while also improving our ability to
measure the clinical benefit from that response in a con-
trolled trial. Clearly, the most significant change (and the
one we evaluated most closely prior to implementing; see
prior sections) involved targeting the SN directly with
CERE-120. At the same time, in an effort to execute every
logical means to further enhance the neurotrophic response,
we also increased the dose to the putamen by 4-fold. This
increase in putaminal dose was now justified on the basis of
the additional clinical and nonclinical safety data generated
with CERE-120 since the inception of the clinical program
in 2005; see Fig. 2c and d. When we initiated the CERE-120
clinical program, the safety/toxicology studies supporting
human testing included 6-month endpoints in rats and 3
months in nonhuman primates. While the 12-month rat and
monkey safety studies were well underway at that time (and
we did submit “in-life” safety data supporting the IND), we
did not complete the histology/pathology end points until
much later. By that time, the clinical program was well on
its way and it would have been dangerously irrational to
change any variable as fundamentally important as dose
level without clear evidence for the need. Thus, when the
initial Phase 1 study was completed, with no safety issues
noted and preliminary evidence of possible efficacy at
both doses, we elected to use the higher of the Phase 1
doses for the Phase 2a study. However, now that we were
considering the nigral targeting approach, which would
necessarily require us to conduct another Phase 1 study
and armed with even more complete and robust rat and
monkey safety/toxicity data, as well as the accumulated
safety data from the Phase 1 and Phase 2a subjects, it now
made sense to consider elevating the putaminal dose. We
therefore elected to increase the dose to the putamen by
4-fold, while also reducing the number of needle tracts
from 4 to 3, decrease the number of CERE-120 bolus
injections in the putamen from 8 (2 per each of 4 needle
tracts) to 3 (see Fig. 2c) and increase the infusion rate (to
putamen, only) from 2 to 3
L per minute (based, in part,
on additional new data in nonhuman primates we col-
lected; Herzog et al., 2011).
Another protocol improvement (intended to maximize
the clinical response and help distinguish between the
CERE-120 group and the sham-surgery group) involved
employing a longer evaluation period (from 12 months vs.
at least 15 and up to 24 months) giving CERE-120 more
time to exert its effects on the degenerating neurons, as well
as more time for the placebo response to subside. Other
protocol changes intended to similarly increase our sensi-
tivity and power to detect a difference between CERE-120
and sham included changing the ratio of treated to sham
subjects from 2:1 to 1:1 (which increases the statistical
power and according to some accounts, might further reduce
the placebo response) and modifying the inclusion/exclu-
sion criteria to enroll somewhat earlier-stage (e.g., no min-
imum “off” time; highest age range lowered to 70 years, no
minimal disease duration or minimum UPDRS score) and
healthier subjects, all of whom nonetheless suffered signif-
icant motor fluctuations.
The Phase 1 safety component of this Phase 1/Phase 2b
protocol involved 2 dose cohorts of 3 PD subjects each (the
first using the same putaminal dose as in the prior Phase 2a
trial 5.4 10
vg, total for both hemispheres) while adding
the SN dose (4.0 10
vg, total for both hemispheres); the
second cohort retained the same SN dose, but increased the
putaminal dose by about 4-fold (2.0 10
vg, total for
both hemispheres). No serious or unexpected safety issues
were observed in the first several months postdosing, which
supported advancing to the Phase 2b double-blind, efficacy/
safety component of the protocol. We have now collected
data for 24 months postdosing and continue to note no
significant or unexpected safety issues (Baumann et al.,
The Phase 2b efficacy/safety component of the protocol
(n 51) is a multicenter, sham surgery controlled double-
blind study, powered at 90% to see a difference between
sham and CERE-120 on the motor-off (Part 3) UPDRS.
Eleven leading movement disorders medical centers across
22 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 22
the United States participated in the trial (see Acknowledge-
ments). The specific safety, motor and quality-of-life assess-
ments are very similar to those used in the initial Phase 2a
trial. The Phase 2b component completed all dosing in
November 2011 and top line data are therefore expected
well before mid-year 2013 (i.e., after all subjects have
completed their 15-month postdosing assessment, the data-
base will be cleaned, audited and locked, the blind broken,
and the prescribed statistical analyses performed).
8. Synopsis and conclusions
Following a long and rich history of research into the
powerful biological effects of neurotrophic factors and sev-
eral unsuccessful attempts to translate their therapeutic po-
tential to treat age-related neurodegenerative diseases, a
significant translational breakthrough appears to have oc-
curred. The emergence and further development of gene
transfer has provided the technological means to solve the
multiple delivery issues inherent with the need to selectively
target these complex proteins to specific sites in the human
brain for sustained periods of time, while retaining their
biological activity. The effort to develop AAV2-NRTN
(CERE-120) for PD produced several findings that allow a
number of important conclusions to be drawn. First, despite
the safety issues noted in several early trials infusing neu-
rotrophic factors into Alzheimer’s disease and PD brains
with mechanical hardware, it seems clear that serious
safety issues can be avoided with appropriately targeted
and delivered neurotrophic factors. To date, CERE-120
has been safely administered to approximately 80 PD
subjects (many of them more than 5–7 years ago), with
over 300 cumulative patient-years exposure to sustained
NRTN expression and no unexpected or serious toxicity
noted. This scenario is due in large part to the proper
application of gene transfer, which not only avoids the
complicated (and at times, unsafe) practices of infusing
neurotrophic factor proteins to the brain via mechanical
hardware, but also circumvents a host of protein delivery
issues and obstacles. This involved a one-time adminis-
tration of relatively small amounts of viral vector con-
taining the gene for NRTN to carefully targeted sites in
the brain and genetically programming neurons in the
vicinity to continuously express the protein, using their
endogenous, cellular machinery.
Similarly, gene transfer is proving to be an “enabling”
technology for the long-term expression of biologically ac-
tive proteins, avoiding the many interrelated pitfalls (e.g.,
formation of neutralizing antibodies, loss of bioactivity and
aggregation of the proteins) often experienced by past meth-
ods attempting to provide a constant source of protein to
treat chronic human diseases (see Herzog et al., 2011). This
cumulative data with CERE-120 establishes that long-term,
targeted, biologically active NRTN can be safely expressed
in both animals and humans inducing a robust and persistent
neurotrophic response. When considered together with the
prior literature related to numerous neurotrophic factors,
these data argue that previous, disappointing clinical results
with neurotrophic factors has less to do with errors in
interpretation of basic research or with the animal data not
accurately predicting potential responses in humans, and far
more to do with more mundane but extremely complicated
and difficult protein delivery issues. The “proof-of-concept”
data generated with CERE-120 help give credence to the
premise that if one can provide sustained, biologically ac-
tive neurotrophic factor to the major population of degen-
erating neurons, one can indeed expect the factor to induce
appropriate repair genes and thus improve neuronal status
and function. However, the experience with CERE-120 also
warned us that even this task is more difficult than initially
conceived, for degenerative changes in the targeted neuro-
nal population increase the difficulty of achieving adequate
coverage of neurotrophic factor throughout the degenerating
neurons, requiring adjustments to neural targeting that had
previously not been appreciated.
Recognizing that our trial in PD did not achieve an
optimal clinical and histological neurotrophic response to
NRTN, we attempted to build upon what had been learned
in the initial “first-in-humans” Phase 1 and Phase 2a trial
and developed a revised protocol intended to both optimize
the potential therapeutic effects of NRTN, while also im-
proving the ability to differentiate the CERE-120 group
from the sham-surgery control group in a randomized, con-
trolled clinical trial. As stated in an independent editorial
published in Movement Disorders: “As one of the few
therapies immediately available that may not only slow PD
progression, but also improve outcomes, we feel that the
potential benefits of a clinically successful CERE-120 treat-
ment cannot be ignored” (Lewis and Standaert, 2011). We
not only appreciate that independent endorsement, but also
recognize the impact the effort and results achieved with
CERE-120 likely have on attempts by others also working
to harness the therapeutic potential of neurotrophic factors.
We also recognize that the history of developing entirely
novel, innovative treatments is fraught with examples that
warn us that the outcome of any single clinical trial is
uncertain and that disappointing results can occur for a wide
variety of reasons, thus causing disappointment even when
the treatment was later shown to work. It is with these
thoughts in mind that we share the insight, experiences and
information we gained during the course of developing
CERE-120. In the worst case, the efforts and means we
employed to develop CERE-120 to establish the first “clin-
ical proof-of-concept” data for neurotrophic factors for age-
related neurodegeneration and more recently to further im-
prove upon those initial effects, may all prove helpful to
others who might continue the effort to develop neu-
rotrophic factors as possible treatments for age-related neu-
rodegenerative diseases.
23R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 23
The body of work reviewed in this article constitutes 10
years of strategic planning and research activity involving
the assistance and input of scores of individuals who are not
authors and we therefore gratefully acknowledge their in-
valuable contributions to the development of CERE-120.
First, the input, assistance and collaborative research of
Jeffrey Kordower was particularly valuable, as was the
guidance and input from Warren Olanow, Eugene Johnson,
and the rest of the Ceregene Scientific Advisory Board,
including Andres Lozano, Rusty Gage, Anthony Lang, In-
der Verma, and Jude Samulski. The invaluable guidance
and advice for statistical issues received from Charles S.
Davis, CSD Biostatistics, Inc., San Diego, CA, is also ap-
preciated. Additionally, the advice and participation of
many other neurologists and neurosurgeons who consulted
and/or participated in the CERE-120 clinical trials is grate-
fully acknowledged, particularly by Nicholas Boulis, Ron
Alterman, Paul Larson, Mark Stacy, Dennis Turner, Roy
Bakay, Joseph Jankovic, William Marks, Jill Ostrem, Herb
Meltzer, and David Barba, as well as Floyd Bloom, David
Weiner, Philip Starr, Ray Watts, Barton Guthrie, Stewart
Factor, Leo Verhagen, Richard Simpson, John Nutt, Mi-
chele Tagliati, Jerrold Vitek, Karl Kieburtz, Carol Tam-
minga, Kathleen Poston, Gordon Baltuch, Matthew Stern,
Jaimie Henderson, Christopher Goetz, Blair Ford, Lawrence
Severt, Catherine Cho, and Guy McKhann, as well as their
study coordinators. The specific contributions of Joao
Siffert, as Ceregene’s Chief Medical Officer during parts of
the Phase 2a and Phase 1/2b studies is gratefully acknowl-
edged, as is that of many other past and current Ceregene
employees, including Mehdi Gasmi, Dominick Vacante,
Dan Lee, Kathie Bishop, Alistair Wilson, Eli Ketchum,
Anthony Ramirez, Eugene Brandon, and Michael Steel,
including all scientific, quality and clinical personnel at
Ceregene who made substantial contributions to the CERE-
120 program, as often evidenced by their role as coauthors
in the numerous publications and regulatory documents
emanating from this program. Also, we thank Amanda San-
chez Losada for help and assistance in preparing the man-
uscript for this report. Finally, our heartfelt appreciation is
given to the Michael J. Fox Foundation for Parkinson’s
Research for their continued interest and support (including
grant funding) during the course of the several clinical trials
conducted with CERE-120, as well as the many courageous
people with PD who not only volunteer to participate in
experimental trials such as CERE-120 but often donate their
brains for study so that much more might be learned and
therefore others might benefit from their efforts.
Alterman, R.L., Tagliati, M., Olanow, C.W., 2011. Open-label surgical
trials for Parkinson Disease: time for reconsideration. Ann. Neurol. 70,
5– 8.
Apfel, S.C., 2002. Nerve growth factor for the treatment of diabetic
neuropathy: what went wrong, what went right, and what does the
future hold? Int. Rev. Neurobiol. 50, 393– 413.
Apfel, S.C., Schwartz, S., Adornato, B.T., Freeman, R., Biton, V., Rendell,
M., Vinik, A., Giuliani, M., Stevens, J.C., Barbano, R., Dyck, P.J.,
2000. Efficacy and safety of recombinant human nerve growth factor in
patients with diabetic polyneuropathy: a randomized controlled trial.
rhNGF Clinical Investigator Group. JAMA 284, 2215–2221.
Apfel, S.C., Kessler, J.A., Adornato, B.T., Litchy, W.J., Sanders, C., Rask,
C.A., 1998. Recombinant human nerve growth factor in the treatment of
diabetic polyneuropathy. NGF Study Group. Neurology 51, 695–702.
Arvanitakis, Z., Tuszynski, M.H., Bakay, R.A., Potkin, S.G., Bartus, R.T.,
Bennett, D., 2007. A phase 1 clinical trial of CERE-110 (AAV-NGF)
gene delivery in Alzheimer’s disease [abstract]. American Academy of
Neurology. Boston, MA.
Bäckman, C., Rose, G.M., Bartus, R.T., Hoffer, B.J., Mufson, E.J., Gran-
holm, A.C., 1997. Carrier mediated delivery of NGF: alterations in
basal forebrain neurons in aged rats revealed using antibodies against
low and high affinity NGF receptors. J. Comp. Neurol. 387, 1–11.
Bäckman, C., Rose, G.M., Hoffer, B.J., Henry, M.A., Bartus, R.T., Friden,
P., Granholm, A.C., 1996. Systemic administration of a nerve growth
factor conjugate reverses age-related cognitive dysfunction and pre-
vents cholinergic neuron atrophy. J. Neurosci. 16, 5437–5442.
Bankiewicz, K., Daadi, M., Pivirotto, P., Bringas, J., Sanchez-Pernaute, R.,
Herscovitch, P., Carson, R., Eckelman, W., Cunningham, J., Reutter,
B., VanBrocklin, H.F., Eberling, J.L., 2003. Long term evaluation of
AAV/AADC gene transfer in parkinsonian monkeys. Neuroscience
299, 14.
Bankiewicz, K.S., Forsayeth, J., Eberling, J.L., Sanchez-Pernaute, R., Pi-
virotto, P., Bringas, J., Herscovitch, P., Carson, R.E., Eckelman, W.,
Reutter, B., Cunningham, J., 2006. Long-term clinical improvement in
MPTP-lesioned primates after gene therapy with AAV-hAADC. Mol.
Ther. 14, 564 –570.
Bartus, R.T., 2012. Translating the therapeutic potential of neurotrophic
factors to clinical “proof of concept”: a personal saga achieving a
career-long quest. Neurobiol. Dis.
Bartus, R.T., Brown, L., Wilson, A., Kruegel, B., Siffert, J., Johnson, E.M.,
Jr., Kordower, J.H., Herzog, C.D., 2011a. Properly scaled and targeted
AAV2-NRTN (neurturin) to the substantia nigra is safe, effective and
causes no weight loss: support for nigral targeting in Parkinson’s
disease. Neurobiol. Dis. 44, 38 –52.
Bartus, R.T., Herzog, C.D., Chu, Y., Wilson, A., Brown, L., Siffert, J.,
Johnson, E.M., Jr., Olanow, C.W., Mufson, E.J., Kordower, J.H.,
2011b. Bioactivity of AAV-2-neurturin gene therapy (CERE-120):
differences between Parkinson’s disease and nonhuman primate brains.
Mov. Disord. 26, 27–36.
Bartus, R.T., Herzog, C.D., Bishop, K., Ostrove, J.M., Tuszynski, M., Kor-
dower, J.H., Gasmi, M., 2007. Issues regarding gene therapy products for
Parkinson’s disease: the development of CERE-120 (AAV-NTN) as one
reference point. Parkinsonism Relat. Disord. 13, S469 –S477.
Bartus, R.T., 1999. The blood–brain barrier as a target for pharmacological
modulation. Current opinion in drug discovery and development. Curr.
Opin. Drug Discov. Dev. 2, 152–167.
Bartus, R.T., 1989a. Drug delivery to the brain: the problem lurking behind
the problem? Neurobiol. Aging 10, 621.
Bartus, R.T., 1989b. Neurotrophic factors: can the degenerating brain be
induced to heal itself? Neurobiol. Aging 10, 513.
Bartus, R.T., 1988. The need for common perspectives in the development
and use of animal models for age-related cognitive and neurodegen-
erative disorders. Neurobiol. Aging 9, 445– 451.
Bartus, R.T., Dean, R.L., Flicker, C., Beer, B., 1983. Behavioral and
pharmacological studies using animal models of aging: implications for
studying and treating dementia of Alzheimer’s type. Banbury Report
1515: Biological aspects of Alzheimer’s disease. Cold Spring Harbor
Laboratory, Cold Spring Harbor, NY. 207–218.
24 R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 24
Baumann, T., Lang, A.E., Lozano, A.M., Olanow, C.W., Bartus, R.T.,
2012. AAV-2-neurturin (CERE-120) and Parkinson’s disease: the
safety and feasibility of combined substantia nigral and putaminal
stereotactic targeting via a Phase 1/2b clinical trial in advanced Par-
kinson’s disease. Abstract No. 9. Movement Disorders Society 16th
Intl. Congress, Dublin, Ireland.
Benabid, A.L., 2010. Gene therapy for Parkinson’s disease: do we have the
cure? Lancet Neurol. 9, 1142–1143.
Braak, H., Sandmann-Keil, D., Gai, W., Braak, E., 1999. Extensive axonal
Lewy neurites in Parkinson’s disease: a novel pathological feature
revealed by alpha-synuclein immunocytochemistry. Neurosci. Lett.
265, 67– 69.
Calcedo, R., Vandenberghe, L.H., Gao, G., Lin, J., Wilson, J.M., 2009.
Worldwide epidemiology of neutralizing antibodies to adeno-associ-
ated viruses. J. Infect. Dis. 199, 381–390.
Charles, V., Mufson, E.J., Friden, P.M., Bartus, R.T., Kordower, J.H.,
1996. Atrophy of cholinergic basal forebrain neurons following exci-
totoxic cortical lesions is reversed by intravenous administration of an
NGF conjugate. Brain Res. 728, 193–203.
Choi-Lundberg, D.L., Lin, Q., Schallert, T., Crippens, D., Davidson, B.L.,
Chang, Y.N., Chiang, Y.L., Qian, J., Bardwaj, L., Bohn, M.C., 1998.
Behavioral and cellular protection of rat dopaminergic neurons by an
adenoviral vector encoding glial cell line-derived neurotrophic factor.
Exp. Neurol. 154, 261–275.
Day-Lollini, P.A., Stewart, G.R., Taylor, M.J., Johnson, R.M., Chellman,
G.J., 1997. Hyperplastic changes within the leptomeninges of the rat
and monkey in response to chronic intracerebroventricular infusion of
nerve growth factor. Exp. Neurol. 145, 24 –37.
De Vos, K.J., Grierson, A.J., Ackerley, S., Miller, C.C., 2008. Role of
axonal transport in neurodegenerative diseases. Annu. Rev. Neurosci.
31, 151–173.
Deuschl, G., Herzog, J., Kleiner-Fisman, G., Kubu, C., Lozano, A.M.,
Lyons, K.E., Rodriguez-Oroz, M.C., Tamma, F., Tröster, A.I., Vitek,
J.L., Volkmann, J., Voon, V., 2006. Deep brain stimulation: postoper-
ative issues. Mov. Disord. 21 (Suppl 14), S219 –S237.
Ding, D., Kanaly, C.W., Bigner, D.D., Cummings, T.J., Herndon, J.E.,
Pastan, I., Raghavan, R., Sampson, J.H., 2010. Convection-enhanced
delivery of free gadolinium with the recombinant immunotoxin MR1-1
[erratum in 2010;98:9]. J. Neurooncol. 98, 1–7.
Eriksdotter Jönhagen, M., Nordberg, A., Amberla, K., Bäckman, L., Eben-
dal, T., Meyerson, B., Olson, L., Seiger, Shigeta, M., Theodorsson, E.,
Viitanen, M., Winblad, B., 1998. Intracerebroventricular infusion of
nerve growth factor in three patients with Alzheimer’s disease. De-
ment. Geriatr. Cogn. Disord. 9, 246 –257.
Fischer, W., Wictorin, K., Björklund, A., Williams, L.R., Varon, S., Gage,
F.H., 1987. Amelioration of cholinergic neuron atrophy and spatial
memory impairment in aged rats by nerve growth factor. Nature 329,
65– 68.
Gash, D.M., Zhang, Z., Cass, W.A., Ovadia, A., Simmerman, L., Martin,
D., Russell, D., Collins, F., Hoffer, B.J., Gerhardt, G.A., 1995. Mor-
phological and functional effects of intranigrally administered GDNF
in normal rhesus monkeys. J. Comp. Neurol. 363, 345–358.
Gash, D.M., Zhang, Z., Ovadia, A., Cass, W.A., Yi, A., Simmerman, L.,
Russell, D., Martin, D., Lapchak, P.A., Collins, F., Hoffer, B.J., Ger-
hardt, G.A., 1996. Functional recovery in parkinsonian monkeys
treated with GDNF. Nature 380, 252–255.
Gasmi, M., Herzog, C.D., Brandon, E.P., Cunningham, J.J., Ramirez, G.A.,
Ketchum, E.T., Bartus, R.T., 2007a. Striatal delivery of neurturin by
CERE-120, an AAV2 vector for the treatment of dopaminergic neuron
degeneration in Parkinson’s disease. Mol. Ther. 15, 62– 68.
Gasmi, M., Brandon, E.P., Herzog, C.D., Wilson, A., Bishop, K.M., Hofer,
E.K., Cunningham, J.J., Printz, M.A., Kordower, J.H., Bartus, R.T.,
2007b. AAV2-mediated delivery of human neurturin to the rat nigro-
striatal system: Long-term efficacy and tolerability of CERE-120 for
Parkinson’s disease. Neurobiol. Dis. 27, 67–76.
Georgievska, B., Jakobsson, J., Persson, E., Ericson, C., Kirik, D., Lund-
berg, C., 2004a. Regulated Delivery of Glial Cell Line-Derived Neu-
rotrophic Factor into Rat Striatum, Using a Tetracycline-Dependent
Lentiviral Vector. Hum. Gene Ther. 15, 934 –944.
Georgievska, B., Kirik, D., Björklund, A., 2004b. Overexpression of glial
cell line-derived neurotrophic factor using a lentiviral vector induces
time- and dose-dependent downregulation of tyrosine hydroxylase in
the intact nigrostriatal dopamine system. J. Neurosci. 24, 6437– 6445.
Gill, S.S., Patel, N.K., Hotton, G.R., O’Sullivan, K., McCarter, R., Bun-
nage, M., Brooks, D.J., Svendsen, C.N., Heywood, P., 2003. Direct
brain infusion of glial cell line-derived neurotrophic factor in Parkinson
disease. Nat. Med. 9, 589 –595.
Hefti, F., Hartikka, J., Knusel, B., 1989. Function of neurotrophic factors
in the adult and aging brain and their possible use in the treatment of
neurodegenerative diseases. Neurobiol. Aging 10, 515–533.
Herzog, C.D., Brown, L., Chu, Y., Baumann, T., Kordower, J.H., Bartus,
R.T., 2012. Robust, stable, targeted, long-term neurturin expression and
enhanced tyrosine hydroxylase labeling in Parkinson’s disease brain 4
years following delivery of CERE-120 (AAV2-neurturin) to the human
putamen. Program No. 755.01. Neuroscience Meeting Planner. Wash-
ington, D.C.: Society for Neuroscience. Online.
Herzog, C.D., Bishop, K., Brown, L., Wilson, A., Kordower, J.H., Bartus,
R.T., 2011. Gene transfer provides a practical means for safe, long-
term, targeted delivery of biologically active neurotrophic factor pro-
teins for neurodegenerative diseases. Drug Deliv. Transl. Res. 1, 361–
Herzog, C.D., Brown, L., Gammon, D., Kruegel, B., Lin, R., Wilson, A.,
Bolton, A., Printz, M., Gasmi, M., Bishop, K.M., Kordower, J.H.,
Bartus, R.T., 2009. Expression, bioactivity, and safety 1 year after
adeno-associated viral vector type 2-mediated delivery of neurturin to
the monkey nigrostriatal system support CERE-120 for Parkinson’s
disease. Neurosurgery 64, 602– 612.
Herzog, C.D., Dass, B., Gasmi, M., Bakay, R., Stansell, J.E., Tuszynski,
M., Bankiewicz, K., Chen, E.Y., Chu, Y., Bishop, K., Kordower, J.H.,
Bartus, R.T., 2008. Transgene expression, bioactivity, and safety of
CERE-120 (AAV2-neurturin) following delivery to the monkey stria-
tum. Mol. Ther. 16, 1737–1744.
Herzog, C.D., Dass, B., Holden, J.E., Stansell, J., 3rd, Gasmi, M., Tuszyn-
ski, M.H., Bartus, R.T., Kordower, J.H., 2007. Striatal delivery of
CERE-120, an AAV2 vector encoding human neurturin, enhances
activity of the dopaminergic nigrostriatal system in aged monkeys.
Mov. Disord. 22, 1124 –1132.
Hoane, M.R., Gulwadi, A.G., Morrison, S., Hovanesian, G., Lindner,
M.D., Tao, W., 1999. Differential in vivo effects of neurturin and glial
cell-line-derived neurotrophic factor. Exp. Neurol. 160, 235–243.
Horger, B.A., Nishimura, M.C., Armanini, M.P., Wang, L.C., Poulsen,
K.T., Rosenblad, C., Kirik, D., Moffat, B., Simmons, L., Johnson, E.,
Milbrandt, J., Rosenthal, A., Bjorklund, A., Vandlen, R.A., Hynes,
M.A., Phillips, H.S., 1998. Neurturin exerts potent actions on survival
and function of midbrain dopaminergic neurons. J. Neurosci. 18, 4929
Hovland, D.N., Boyd, R.B., Butt, M.T., Engelhardt, J.A., Moxness, M.S.,
Ma, M.H., Emery, M.G., Ernst, N.B., Reed, R.P., Zeller, J.R., Gash,
D.M., Masterman, D.M., Potter, B.M., Cosenza, M.E., Lightfoot, R.M.,
2007. Six-month continuous intraputamenal infusion toxicity study of
recombinant methionyl human glial cell line-derived neurotrophic fac-
tor (r-met HuGDNF) in rhesus monkeys. Tox. Pathologe 35, 1013–
Kirik, D., Rosenblad, C., Björklund, A., 2000a. Preservation of a functional
nigrostriatal dopamine pathway by GDNF in the intrastriatal 6-OHDA
lesion model depends on the site of administration of the trophic factor.
Eur. J. Neurosci. 12, 3871–3882.
Kirik, D., Rosenblad, C., Bjorklund, A., Mandel, R.J., 2000b. Long-term
rAAV-mediated gene transfer of GDNF in the rat Parkinson’s model:
intrastriatal but not intranigral transduction promotes functional regen-
25R.T. Bartus et al. / Neurobiology of Aging xx (2012) xxx
Page 25
eration in the lesioned nigrostriatal system. J. Neurosci. 20,
4686 4700.
Kohn, D.B., Sadelain, M., Glorioso, J.C., 2003. Occurrence of leukaemia
following gene therapy of X-linked SCID. Nat. Rev. Cancer 3, 477–
Kordower, J.H., Herzog, C.D., Dass, B., Bakay, R.A., Stansell, J., Gasmi,
M., Bartus, R.T., 2006. Delivery of neurturin by AAV2 (CERE-120)-
mediated gene transfer provides structural and functional neuroprotec-
tion and neurorestoration in MPTP-treated monkeys. Ann. Neurol. 60,
706 –715.
Kordower, J.H., Emborg, M.E., Bloch, J., Ma, S.Y., Chu, Y., Leventhal, L.,
McBride, J., Chen, E.Y., Palfi, S., Roitberg, B.Z., Brown, W.D.,
Holden, J.E., Pyzalski, R., Taylor, M.D., Carvey, P., Ling, Z., Trono,
D., Hantraye, P., Déglon, N., Aebischer, P., 2000. Neurodegeneration
prevented by lentiviral vector delivery of GDNF in primate models of
Parkinson’s disease. Science 290, 767–773.
Kordower, J.H., Palfi, S., Chen, E.Y., Ma, S.Y., Sendera, T., Cochran, E.J.,
Cochran, E.J., Mufson, E.J., Penn, R., Goetz, C.G., Comella, C.D.,
Comella, C.D., 1999. Clinicopathological findings following intraven-
tricular glial-derived neurotrophic factor treatment in a patient with
Parkinson’s disease. Ann. Neurol. 46, 419 424.
Kotzbauer, P.T., Lampe, P.A., Heuckeroth, R.O., Golden, J.P., Creedon,
D.J., Johnson, E.M., Milbrandt, J., 1996. Neurturin, a relative of glial-
cell-line-derived neurotrophic factor. Nature 384, 467– 470.
Lampson, L.A., 2001. New animal models to probe brain tumor biology,
therapy, and immunotherapy: advantage and remaining concerns.
J. Neuro Oncol. 53, 275–287.
Lang, A.E., Gill, S., Patel, N.K., Lozano, A., Nutt, J.G., Penn, R., Brooks,
D.J., Hotton, G., Moro, E., Heywood, P., Brodsky, M.A., Burchiel, K.,
Kelly, P., Dalvi, A., Scott, B., Stacy, M., Turner, D., Wooten, V.G.,
Elias, W.J., Laws, E.R., Dhawan, V., Stoessl, A.J., Matcham, J., Cof-
fey, R.J., Traub, M., 2006. Randomized controlled trial of intraputa-
menal glial cell line-derived neurotrophic factor infusion in Parkinson
disease. Ann. Neurol. 59, 459 466.
Lewis, T.B., Standaert, D.G., 2011. Parkinson’s disease, primates, and
gene therapy: vive la différence? Mov. Disord. 26, 2–3.
Lin, L.F., Doherty, D.H., Lile, J.D., Bektesh, S., Collins, F., 1993. GDNF:
a glial cell line-derived neurotrophic factor for midbrain dopaminergic
neurons. Science 260, 1130 –1132.
Linninger, A.A., Somayaji, M.R., Mekarski, M., Zhang, L., 2008. Predic-
tion of convection-enhanced drug delivery to the human brain. J. Theor.
Biol. 250, 125–138.
Lu, L., Wang, X., Wu, P., Xu, C., Zhao, M., Morales, M., Harvey, B.K.,
Hoffer, B.J., Shaham, Y., 2009. Role of ventral tegmental area glial cell
line-derived neurotrophic factor in incubation of cocaine craving. Biol.
Psychiatry 66, 137–145.
Manfredsson, F.P., Burger, C., Rising, A.C., Zuobi-Hasona, K., Sullivan,
L.F., Lewin, A.S., Huang, J., Piercefield, E., Muzyczka, N., Mandel,
R.J., 2009a. Tight long-term dynamic doxycycline responsive nigro-
striatal GDNF using a single rAAV vector. Mol. Ther. 17, 1857–1867.
Manfredsson, F.P., Tumer, N., Erdos, B., Landa, T., Broxson, C.S., Sulli-
van, L.F., Rising, A.C., Foust, K.D., Zhang, Y., Muzyczka, N., Gor-
batyuk, O.S., Scarpace, P.J., Mandel, R.J., 2009b. Nigrostriatal rAAV-
mediated GDNF overexpression induces robust weight loss in a rat
model of age-related obesity. Mol. Ther. 17, 980 –991.
Marks, W.J., Bartus, R.T., Siffert, J., Davis, C.S., Lozano, A., Boulis, N.,
Vitek, J., Stacy, M., Turner, D., Verhagen, L., Bakay, R., Watts, R.,
Guthrie, B., Jankovic, J., Simpson, R., Tagliati, M., Alterman,
R., Stern, M., Baltuch, G., Starr, P.A., Larson, P.S., Ostrem, J.L., Nutt,
J., Kieburtz, K., Kordower, J.H., Olanow, C.W., 2010. Gene delivery of
AAV2-neurturin for Parkinson’s disease: a double-blind, randomised,
controlled trial. Lancet Neurol. 9, 1164 –1172.
Marks, W.J., Ostrem, J.L., Verhagen, L., Starr, P.A., Larson, P.S., Bakay,
R.A., Taylor, R., Cahn-Weiner, D.A., Stoessl, A.J., Olanow, C.W.,
Bartus, R.T., 2008. Safety and tolerability of intraputaminal delivery of
CERE-120 (adeno-associated virus serotype 2-neurturin) to patients
with idiopathic Parkinson’s disease: an open-label, phase I trial. Lancet
Neurol. 7, 400 408.
Maswood, N., Grondin, R., Zhang, Z., Stanford, J.A., Surgener, S.P., Gash,
D.M., Gerhardt, G.A., 2002. Effects of chronic intraputamenal infusion
of glial cell line-derived neurotrophic factor (GDNF) in aged Rhesus
monkeys. Neurobiol. Aging 23, 881– 889.
McMahon, S.B., 1996. NGF as a mediator of inflammatory pain. Philos
Trans R Soc Lond B Biol Sci 351, 431– 440.
McNaught, K.S., Olanow, C.W., 2006. Protein aggregation in the pathogenesis of
familial and sporadic Parkinson’s disease. Neurobiol. Aging 27, 530 –545.
Miller, R.G., Petajan, J.H., Bryan, W.W., Armon, C., Barohn, R.J., Good-
pasture, J.C., Hoagland, R.J., Parry, G.J., Ross, M.A., Stromatt, S.C.,
1996. A placebo-controlled trial of recombinant human ciliary neu-
rotrophic (rhCNTF) factor in amyotrophic lateral sclerosis. rhCNTF
ALS Study Group. Ann. Neurol. 39, 256 –260.
Morfini, G.A., Burns, M., Binder, L.I., Kanaan, N.M., LaPointe, N., Bosco,
D.A., Brown, R.H., Brown, H., Tiwari, A., Hayward, L., Edgar, J.,
Nave, K.A., Garberrn, J., Atagi, Y., Song, Y., Pigino, G., Brady, S.T.,
2009. Axonal transport defects in neurodegenerative diseases. J. Neu-
rosci. 29, 12776 –12786.
Morrison, P.F., Lonser, R.R., Oldfield, E.H., 2007. Convective delivery of
glial cell line-derived neurotrophic factor in the human putamen.
J. Neurosurg. 107, 74 83.
Mufson, E.J., Kroin, J.S., Sendera, T.J., Sobreviela, T., 1999. Distribution
and retrograde transport of trophic factors in the central nervous sys-
tem: functional implications for the treatment of neurodegenerative
diseases. Prog. Neurobiol. 57, 451– 484.
Nauta, H.J., Wehman, J.C., Koliatsos, V.E., Terrell, M.A., Chung, K., 1999.
Intraventricular infusion of nerve growth factor as the cause of sympathetic
fiber sprouting in sensory ganglia. J. Neurosurg. 91, 447–453.
Nutt, J.G., Burchiel, K.J., Comella, C.L., Jankovic, J., Lang, A.E., Laws,
E.R., Lozano, A.M., Penn, R.D., Simpson, R.K., Stacy, M., Wooten,
G.F., ICV GDNF Study Group. Implanted intracerebroventricular glial
cell line-derived neurotrophic factor, 2003. Randomized, double-blind
trial of glial cell line-derived neurotrophic factor (GDNF) in PD.
Neurology 60, 69 –73.
Oiwa, Y., Yoshimura, R., Nakai, K., Itakura, T., 2002. Dopaminergic
neuroprotection and regeneration by neurturin assessed by using be-
havioral, biochemical and histochemical measurements in a model of
progressive Parkinson’s disease. Brain Res. 947, 271–283.
Penn, R.D., Kroin, J.S., York, M.M., Cedarbaum, J.M., 1997. Intrathecal
ciliary neurotrophic factor delivery for treatment of amyotrophic lateral
sclerosis (phase I trial). Neurosurgery 40, 94 –99.
Pezet, S., McMahon, S.B., 2006. Neurotrophins: mediators and modulators
of pain. Annu. Rev. Neurosci. 29, 507–538.
Raff, M.C., Whitmore, A.V., Finn, J.T., 2002. Axonal self-destruction and
neurodegeneration. Science 296, 868 871.
Raper, S.E., Chirmule, N., Lee, F.S., Wivel, N.A., Bagg, A., Gao, G.P.,
Wilson, J.M., Batshaw, M.L., 2003. Fatal systemic inflammatory re-
sponse syndrome in a ornithine transcarbamylase deficient patient fol-
lowing adenoviral gene transfer. Mol. Genet. Metab. 80, 148 –158.
Rosenblad, C., Georgievska, B., Kirik, D., 2003. Long-term striatal overex-