65. New York Times, 2 September 1987, p. B4.
66. J. Am. Med. Assoc. 258, 2097 (1987).
67. A staff dentist in a Texas prison who demanded the option to refiuse to treat
prisoners with diagnosed or suspected AIDS has been dismissed (WashingtonPost,
Health Section, 25 August 1987, p. 5).
68. J. Mann, Law, Med. Health Carc 14, 290 (1986).
69. World Health Organization, Special Program on AIDS, Report ofthe Consultation
on Internatinal Travel and H1V Infction (World Health Organization, Geneva,
70. S. S. Fluss, inAlDS in Chbdren,Adokscents, andHeterosexualAdults, R. F. Schinazi
and A. J. Nahmias, Eds. (Elsevier-North Holiand, New York, in press).
71. World Health Organization (WHO), WeeklyEpidemiol. Rec., 9 December 1987; P.
Piot et al., Scienc 239, 573 (1988).
72. Int. Dig. HealthLegis. 37, 533 (1986).
73. Ibid. 38, 478 (1987).
74. Ibid., p. 489.
75. Ibid. 38, in press.
76. Diario de CentroAmirica (No. 98) (18 June 1986), p. 2483.
77. BioLaw 2, U:446 (1987).
78. Ibid., p. U:596.
The Brain in AIDS: Central Nervous System
HIV-1 Infection and AIDS Dementia Complex
RICHARD W. PRICE,* BRUCE BREw, JOHN SIDTIS, MARC ROSENBLUM,
ADRIENNE C. SCHECK, PAUL CLEARY
Infection with human immunodeficency virus type 1
(HIV-1) is frequently complicated in its late stages by the
AIDS dementia complex, a neurological syndrome char-
acterized by abnormalities in cognition, motor perform-
ance, and behavior. This dementia is due pardally or
wholly to a direct effect of the virus on the brain rather
than to opportunistic infection, but its pathogenesis is not
well understood. Productive HIV-1 brain infection is
detected only in a subset of patients and is confined
largely or exclusively to macrophages, microglia, and
derivative multinucleated cells that are formed by virus-
induced cell fusion. Absence of cytolytic infection of
neurons, ohigodentrocytes, and astrocytes has focused
attention on the possible role of indirect mechanisms of
brain dysfunction related to either virus or cell-coded
toxins. Delayed development ofthe AIDS dementia com-
plex, despite both early exposure ofthe nervous system to
HIV-1 and chronic leptomeningeal infection, indicates
that although this virus is "neurotropic," it is relatively
nonpathogenic for the brain in the absence of immuno-
suppression. Within the context ofthe permissive effectof
immunosuppression, genetic changes in HIV-1 may un-
derlie the neuropathological heterogeneity of the AIDS
dementia complex and its relatively independent course in
relation to the systemic manifestations ofAIDS noted in
IT IS NOW CLEAR THAT INFECTION WITH HUMAN IMMUNODE-
ficiency virus type 1 (HIV-1) is complicated by a dementing
neurological disorder, the AIDS dementia complex, which is
both a common and an important cause ofmorbidity in patients in
advanced stages of infection (1). It was not long after the recogni-
tion ofAIDS in 1981 that reports began to appear of an unusual
encephalopathy in affected patients (2). Initial efforts to identify and
classify this neurological syndrome were directed toward identifying
an underlying opportunistic infection (3), but misgivings with this
approach arose as more detailed clinical-pathological studies were
performed (1, 4) and as a parallel disorder was observed in children,
who are less prone to opportunistic brain infections (5). Identifica-
tion of the retroviral etiology ofAIDS allowed introduction ofthe
hypothesis that HIV-1 itself might infect the brain and directly
cause dementia. This hypothesis, accounting for the frequency and
unique character ofboth the clinical syndrome and its neuropatholo-
gy, also found support in precedents ofretrovirus brain infections of
animals that had been studied as models of neurodegenerative
disorders. In particular, comparisons were made with visna virus,
the prototype lentivirus, which shares considerable biological simi-
larity and some genetic homology with HIV-1 (6). This rapidly led
to identification of HIV-1 in brains of demented patients, first by
Southem blot analysis and in situ hybridization (7) and subsequent-
ly by other techniques (8-16).
Although considerable progress has been made in characterizing
and understanding this new neurological disorder, many questions
remain regarding both its clinical and biological features (17). In
this article we review the clinical, epidemiological, and pathological
aspects of the AIDS dementia complex and discuss some of the
principal unresolved issues regarding its viral pathogenesis.
Clinkal Features ofAIDS Dementia Complex
Patients with the AIDS dementia complex present with a variable,
yet characteristic, constellation ofabnormalities in cognitive, motor,
and behavioral function (1). Perhaps the salient aspects of the
R. W. Price, B. Brew, J. Sidtis, M. Rosenblum, and A. C. Scheck are in the Departnent
of Neurology, Memorial Hospital, 1275 York Avenue, New York, NY 10021. P.
Cleary is at Harvard Medical School, Boston, MA 02115.
*To whom correspondence should be addressed.
SCIENCE, VOL. 239
on July 7, 2015
on July 7, 2015
on July 7, 2015
on July 7, 2015
on July 7, 2015
on July 7, 2015
on July 7, 2015
disorder are the slowing and loss ofprecision in both mentation and
motor control. Early in the illness, patients frequently report that
they must keep lists in order to carry out their normal activities and
that complex but formerly routine mental tasks take longer and need
to be consciously broken down into component steps. Concomi-
tantly, these patients often lose interest in their work as well as in
their social and recreational activities. It is this growing apathy along
with mental slowing that is frequently mistaken for depression by
their associates, yet dysphoria is often absent. Motor symptoms
usually lag somewhat behind, but exaggerated tremor or mild gait
unsteadiness may be among early complaints, while on examination,
slowing of rapid alternating eye and extremity movements and
abnormal "release" reflexes are common.
With time, intellectual impairment becomes more pervasive,
broadly affecting nearly all aspects of cognition, with further
slowing and inaccuracy of performance. Increasing apathy, slowing
of speech, and mental impoverishment ultimately may progress to
near or absolute mutism and severe dementia. In parallel, gait
unsteadiness gives way to frank weakness, general hypokinesia, and
incontinence. Variants of the syndrome occur in which particular
aspects predominate. Thus, in some patients an agitated mental state
with mania or other forms of organic psychosis may occur; in
others, weakness with progressive paraparesis dominates the course.
Extrapyramidal dysfunction with bradykinesia and postural instabil-
ity may resemble Parkinsonism, but without the resting tremor.
Formal neuropsychological studies are useful in documenting and
serially following disease severity and in understanding the patho-
physiological basis ofthe symptoms and signs (18-20). The charac-
teristic abnormalities include difficulty with complex sequencing,
impaired fine and rapid motor movement, and reduced verbal
fluency, while other verbal abilities, including vocabulary and object
naming, tend to be maintained even when the disease is relatively
advanced. Also notable in many patients is a discrepancy between
their complaints of frequent forgetfulness and their relatively pre-
served performance on formal memory testing. In general, the
neuropsychological impairments become most prominent when
some or all of the following demands are placed on the patient:
performance under time pressure, problem solving, visual scanning,
visual-motor integration, and alternation between two or more
performance rules or stimulus sets (18, 20). Although depression
may coexist with, mimic, or potentially result from the AIDS
dementia complex, clinical evaluations as well as analyses of data
from self-report questionnaires indicate that depression does not
account for impaired performance on neuropsychological tests that
are most sensitive to the AIDS dementia complex. The pattern of
clinical and neuropsychological abnormalities conforms to what has
been termed a "subcortical dementia," previously applied to patients
with progressive supranuclear palsy, Huntington's disease, and
Parkinson's disease (21).
Computed tomographic scanning and magnetic resonance imag-
ing reveal the common early brain atrophy and, lest frequently,
abnormalities of the white matter in patients with this disorder (1,
22). A study conducted with positron emission tomography made it
possible to distinguish patients with AIDS dementia complex from
normal subjects on the basis of patterns of regional brain glucose
metabolism (23). The metabolic pathology of the disorder was
characterized bytwo patterns ofregional covariation, one influenced
primarily by subcortical metabolism and correlated with impairment
offine motor control, and the other influenced by cortical metabo-
lism and correlated with impairments in verbal fluency and problem
solving. With respect to disease progression, basal ganglion and
thalamic glucose hypermetabolism appeared to be an early compo-
nent of the disorder that was followed later by perturbation of
subcortical-cortical relationships and general hypometabolism.
The Epidemiology ofHIV-1 Infection ofthe
Central Nervous System
The AIDS dementia complex must be considered in the context of
the overall course ofHIV-1 infection and the still imprecise picture
ofother central nervous system (CNS) disorders known or suspect-
ed to be related to primary infection ofthe brain with HIV-1 (Fig.
1). The AIDS dementia complex is the most common ofthese, but
emerging evidence of earlier symptomatic and asymptomatic CNS
HIV-1 infection provides an important broader view ofthe interac-
tion of this retrovirus and the human nervous system.
Early symptomatic infeciown ofthe CNS. Although described only as
individual cases (15, 24), the CNS may be affected early in the
course ofHIV-1 infection. Either within the context ofthe serocon-
version-related, mononucleosis-like illness accompanying primary
HIV-1 infection, or somewhat later during the "latent" phase,
headache, as well as encephalitis, aseptic meningitis, ataxia, and
myelopathy have been described. These disorders are monophasic
with most resolving in weeks. Although their incidence is probably
low, these acute disorders may be underappreciated since they are
Fig. 1. Stages ofHIV-1 infection in relation to the development ofthe AIDS dementia complex
and other neurological complications. In this model, adapted from that proposed for systemic
disease by Redfield (51), we have divided systemic disease into four phases: an acute phase
following initial virus exposure; a latent phase without systemic symptoms or signs; and two
later phases, an early-late phase associated with "minor" opportunistic conditions and roughly
corresponding to the AIDS-related complex but incuding also the period when Kaposi's
sarcoma may develop, and a final late phase in which major, AIDS-defining, opportunistic
infections occur. Using different but conceptually parallel terminology, Redfield relates these
conditions to changing virus load and immunity to HIV-1. Thus, acute viremia may be
associated with high viral titers in the blood, but this is curtailed as appropriate host immune
responses suppress virus replication as in other acute viral syndromes. However, the virus
persists and over the course ofthe latent period, by virtue ofits effect on CD4+ lymphocytes and
perhaps macrophages, slowly erodes the defenses that protect against opportunistic pathogens as
well as HIV-L itself. At a certain threshold, the balance between the immune response and HIV-
1 replication is tipped so that the virus load increases and the rate of immunosuppression
accelerates. Present epidemiological and clinical observations allow the CNS complications to be
tentatively seen within this pathogenetic framework. Acute encephalitis and aseptic meningitis
accompany or soon follow initial viremia and exhibit a similar monophasic time frame,
terminating as CNS virus is limited by effective immune defenses. However, asymptomatic
infection may persist but without meaningfuil invasion ofthe parenchyma or brain dysfunction.
Only when immune defenses wane is there substantial invasion and HIV-1 replication in the
brain. Hence, the AIDS dementia complex is a relatively late sequela and requires the permissive
effect of immunosuppression.
Central nervous system events
- Asymptomatic infection
AIDS dementia complex
CNS opportunistic infections
clinically indistinguishable from other acute viral or postinfectious
encephalitides, most ofwhich never achieve specific diagnosis.
Asymptomatic infection. In contrast to the apparent rarity of the
acute syndromes, early asymptomatic HIV-1 infection ofthe CNS,
or at least ofthe leptomeninges, is relatively common and may even
be the rule. Studies ofthe cerebrospinal fluid (CSF) have provided a
valuable "window" into the biology of HIV-1 infection of the
nervous system. Evidence ofearly infection and local host reactions
in the CSF of asymptomatic seropositive patients include:
increase in the number ofinflammatory cells or protein content and
the presence of "oligoclonal" immunoglobulin bands of undeter-
mined specificity; (ii) local, "intrablood-brain barrier" synthesis of
antibody to HIV-1; and (iii) isolation ofvirus or detection of viral
antigens (25, 26). Although data are limited, even asymptomatic
patients may experience an acute phase ofinfection in which there is
detectable virus and host cell reaction in the CSF, as well as a
subsequent phase in which virus is reduced or absent but the local
production of HIV-1 antibodies continues.
Aseptic meningitis. Aseptic meningitis occurs not only in the
setting of seroconversion, but, even more commonly, later in the
course ofHIV-1 infection, usually as the latent phase progresses to
the AIDS-related complex (ARC) or full-blown AIDS (2,27). Both
acute and chronic forms ofthis disorder have headache as their most
prominent feature. The CSF again shows a mononuclear pleocyto-
sis, usually with normal glucose and mildly elevated protein, and
HIV- 1 is readily isolated. This syndrome itselfis benign, although it
carries an overall poor prognosis and is a harbinger ofother AIDS-
related complications. Whether or not these patients have a higher
incidence or pursue a different course with respect to the AIDS
dementia complex is uncertain.
AlDS dementia complex. The epidemiology and course of the
AIDS dementia complex are imprecisely defined with estimates
being derived principally from clinical and pathological series rather
than more representative population-based or prospective analyses.
Hence the frequency of this syndrome at each stage of systemic
HIV- 1 infection remains uncertain. However, ifthe early monopha-
sic encephalitides are excluded, the AIDS dementia complex appears
to be confined to the later phases of systemic infection (Fig. 1).
Most cases occur in the setting of systemic AIDS (1) as defined by
the Centers for Disease Control (CDC) criteria (28) relating to life-
threatening opportunistic infections. In our own selected clinical
experience we have estimated that at the time of AIDS diagnosis
perhaps one-third ofpatients exhibit overt and one-quarter subclini-
cal AIDS dementia complex; these numbers progress to an eventual
preterminal prevalence of perhaps two-thirds of patients with
clinically significant forms of the disorder and one-quarter with
(1, 29). Clearly, however, the disorder may
develop before the diagnosis of systemic AIDS, and a substantial,
though indeterminant, number of patients with "minor" systemic
disease (ARC) develop the syndrome (29, 30).
The AIDS dementia complex may also develop in the absence of
any systemic symptoms, but this is unusual; most, ifnot all, ofthese
patients have laboratory evidence of immunosuppression. It is
important to emphasize that while some reports have described
neuropsychological impairment in HIV-1 seropositive individuals
withoutARC or AIDS, these individuals were not immunologically
normal and were probably on the threshold of the later stages of
HIV-1 infection (30). Furthermore, the definition ofimpairment in
the testing procedures is not firmly established. Thus, the neuro-
psychological abnormalities reported in these small series should not
be extrapolated to the large existing population of seropositive
individuals and do not now justify excluding seropositive individ-
uals from employment.
Overall, however, the AIDS dementia complex is one ofthe most
important complications associated with HIV-1 infection. Whether
appearing early or late, whether dominating the course of infection
or compounding systemic disease, this neurological condition grad-
ually impairs function in work, in daily life, and eventually in self
care. The progressive disability ofpatients with this disorder also has
broad societal implications that relate to lost productivity as well as
to the costs of long-term assistance and institutional care.
Pathology of the AIDS Dementia Complex
and CNS HIV-1 Infection
The pathological abnormalities in patients with the AIDS demen-
tia complex are variable, and it is not yet clear that the clinical
syndrome corresponds to a single etiopathogenetic entity. The
principal histopathological abnormalities are most prominent in the
subcortical structures, notably in the central white matter, deep gray
structures including the basal ganglia and thalamus, the brain stem,
and the spinal cord (4, 31). There is relative sparing of the cortex.
One can divide the histopathological abnormalities into three
seemingly discontinuous, but frequently coexisting, sets: diffuse
pallor of the white matter, multinucleated cell encephalitis, and
vacuolar myelopathy (4). The most common ofthese, diffuse white
matter pallor accompanied by astrocytic reaction, involves particu-
larly the central and periventricular white matter and generally
parallels in severity the neurological symptomatology (Table 1).
Diffiuse white matter pallor can also be identified in mild degree in
an appreciable number of patients with apparently normal or
subclinically altered neurological status, and indeed the presence of
atrophy and diffiuse pallor in individuals without overt neurological
Table 1. Subclassification ofAIDS patients and the dementia complex according to clinical, pathological, and virological findings. The listed frequencies are
based on estimates derived from our earlier studies (1, 4, 7, 13, 52) but should only be considered tentatively established since some ofthe observations
involved different sets ofpatients. AIDS patients are divided into three groups on the basis ofthe presence (groups 1 and 2) or absence (group 3) ofclinically
overt AIDS dementia complex and the pathological identification ofmultinucleated cell encephalitis (group 1). The degree ofdiffuse pallor tends to segregate
with the severity of clinical dementia although there is clear overlap among the three groups. The frequencies given for vacuolar myelopathy include only
findings previously categorized as grades 1 (severe) and 2 (moderate) and omits mild changes; other workers have remarked on a lower incidence ofthis ab-
nonnality. The HIV-1 detection methods included immunohistochemistry, Southern blot, and in situ hybridization; direct isolations ofvirus from brain or
CSF were not included in the classification.
of the AIDS
Moderate to severe
Mild to moderate
Moderate to severe
Mild to moderate
Absent to mild
SCIENCE, VOL. 239
disease provides an additional indication of the high frequency of
brain involvement in HIV-1-infected individuals. When pallor is
present without multinucleated cells, inflammatory changes are
characteristically scant, consisting ofa few perivascular lymphocytes
and brown-pigmented macrophages accompanying the astrocytosis.
Multinucleated cells are found in a subgroup of patients with
more severe clinical disease. In the brains of these patients reactive
infiltrates are more prominent and consist of perivascular and
parenchymal foamy macrophages, microglia, and lymphocytes,
along with the multinucleated cells (Fig. 2A). These infiltrates are
most often concentrated in the white matter and deep gray struc-
tures, including particularly the basal ganglia, thalamus, and pons.
In the white matter, they are often surrounded by focal rarefaction
of myelin. In both white and gray matter there is accompanying
reactive astrocytosis, but loss ofoligodendrocytesor neurons is not
characteristic. Additional, but far less common, white matter
changes include spongy vacuolation, which may be diffuse or focal.
In pediatric cases the pathology is similar except that it also includes
a calcific vasculopathy (5).
Inflammation with multinucleated cells may also be present in the
spinal cord, but in our own experience a vacuolar myelopathy is
more common (4, 32). The latter pathologically resembles subacute
combined degeneration resulting from vitamin B12 deficiency, but
serum B12 levels are normal. The vacuolation appears to result from
swelling within the layers ofthe myelin sheaths. Similar changes can
also extend into the brainstem ofsome patients. Although there is a
general correlation between the incidence of vacuolar myelopathy
and the other pathological abnormalities found in the brain, the
myelopathy can certainly occur in the absence ofthe multinucleated
cell-associated changes. Additionally, there is not always a one-to-
one correlation between the severity ofthis myelopathy and that of
the brain abnormalities, and no pediatric examples of vacuolar
myelopathy have been published. These discrepancies leave very
much open the question ofwhether vacuolar myelopathy is a variant
of the process causing the multinucleated cell formation and other
brain changes described above or whether it is pathogenetically and
etiologically independent. Indeed, the overall pathological diversity
underlying the AIDS dementia complex is one of its fundamental
puzzling aspects, and a reason to retain the latter term to describe a
syndrome until the underlying etiologies and pathogenetic processes
There is now firm evidence that the AIDS dementia complex, at
least when multinucleated cells are present histopathologically, is
associated with direct brain infection by HIV-1. Southem blot
analysis has shown both a high frequency (comparable to that of
lymphatic tissue) and high copy number of proviral DNA in the
brains of patients with this pathology, and both integrated and
nonintegrated forms ofthe genome have been identified (7). In situ
hybridization and immunohistochemical studies have also revealed
the presence of viral nucleic acid and antigens within these brains
(8-13), and HIV-1 has been cultured directly both from brain and
CSF ofdemented patients (15, 26). In addition, HIV-1 virions have
been detected by electron microscopy (14).
In our own experience, the distribution of productively infected
cells identified immunocytochemically parallels the neuropatholog-
ical findings; these cells are most prominent in subcortical struc-
tures, although cortical involvement can also occur (13). Although
there remains some controversy conceming the cell types involved in
productive brain infection, there is emerging consensus that macro-
phages and multinucleated cells derived from macrophages are
principal participants (Fig. 2B) (8, 9, 12, 13, 16). Although not an
invariant finding in HIV-1-infected brains, the multinucleated cells
are histological markers of productive HIV-1 brain infection and
indeed the multinucleation almost certainly results from direct virus-
5 FEBRUARY I988
induced cell fusion in situ; these cells are thus the in vivo counter-
parts of the syncytial cytopathology characteristic of HIV-1 in cell
culture (33). Viral antigens have also been detected in other cells
with cellular processes (13). Many ofthese cells have been identified
morphologically and histochemically as microglia (Fig. 2, C and D)
(12, 34). Microglia additionally appear to participate in the forma-
tion of multinucleated forms (32). Whether other cell types in the
brain, includingastrocytes, oligodendrocytes, neurons, or vascular
endothelium are also infected is less clear (10, 13, 14). Cell culture
studies have demonstrated low-level infection of astrocytic tumor
cell lines (35), and similar restricted or latent infection ofthese and
other native brain elements has not been ruled out. As with visna
virus (36), such infection may be pathogenetically important.
Viral Pathogenesis ofAIDS Dementia
Unresolved issues of the pathogenesis of the AIDS dementia
complex and CNS HIV-1 infection can be considered in the context
of three fundamental questions.
1) What determines progression ofHIV-1 CNS infection and its
various dinicalmanifrstationsat each stage? There may be genetically
determined host differences that influence initial susceptibility to
HIV-1 infection and perhaps the speed of disease progression.
Likewise, other infections, past or present, may also modulate CNS
infection. However, the evolution of CNS infection probably
depends principally on: (i) changes in the host immune response to
the virus and (ii) changes in the infecting virus.
Progressive immunosuppression is, of course, the hallmark of
HIV-1 infection. In addition to conferring vulnerability to opportu-
nistic infections, this immunosuppression also affects the host's
defense against HIV-1 (37). Waning defenses against HIV-1 and a
resultant increase in the "virus load" probably also influence HIV-1
infection ofthe CNS and provide a framework for interpreting the
neurological manifestations at each phase of the evolving infection
(Fig. 1). The acute CNS disorders occurring at or near the time of
initial exposure to HIV-1 and seroconversion can be viewed as
pathogenetically similar to other acute encephalitides and aseptic
meningitides caused by organisms that are of relatively low neuro-
virulence. In most patients, immunological responses effectively
reduce or eliminate the virus from the brain, and, mechanistically,
CNS dysfunction may relate as much to the inflammatory changes
and "bystander effects" as to the virus itself; hence the good
The subsequent course of asymptomatic CNS infection on the
one hand indicates that the virus is "neurotropic" by virtue of its
predilection to infect the leptomeninges early and frequently, andon
the other hand provides evidence ofthe low neurovirulence ofHIV-
1 despite its persistence, at least as long as immune defenses restrict
its replication and spread. Neuropathology is mild or absent, and
our own studies thus far indicate that productive brain infection is
not detected by immunohistochemistry. Invasion and replication of
virus within the brain and secondary injury are therefore probably
minimal or absent during this phase.
As the degree ofimmunosuppression increases and the transition
from the latent period to the later phases of disease occurs, HIV-1
replication, including that in the CNS, may escape immune control.
This probably explains why the AIDS dementia complex begins to
appear at this point. In this respect, HIV-1 may be similar to JC
virus, which causes progressive multifocal leukoencephalopathy only
in the absence ofan effective host antiviral response (38). In the case
of HIV-11, however,
"opportunity" by virtue ofits effect on the immune system and then
it is the same organism that first creates
takes advantage of this opportunity to infect the brain.
Although immunosuppression appears to have a necessay permis-
sive influence on the development ofthe AIDS dementia complex, it
may not account for discrepancies in the coevolution ofsystemic and
CNS disease noted in some patients. Thus, some patients develop
multiple opportunistic infections and have minimal circulating
CD4+ T lymphocytes yet remain neurologically intact, while others
show progressive neurological deterioration over many months to
years but develop few or no major opportunistic infections. Such
differences provide a clinical basis for examining whether there are
neurotropic variants of HIV-1 that more readily infect brain
macrophages, microglia, or other neural cell types. Likewise, differ-
ences in viral strains may contribute to the heterogeneity of disease
pattems and pathologies. One can speculate, for example, that some
viral strains may be more likely to cause aseptic meningitis than the
AIDS dementia complex, or that some will induce vacuolar change
in the spinal cord while others will cause only multinucleated cell
encephalitis. Genetic drift with genesis ofvariants ofHIV-1 may, in
fact, be more important in causing these different pathologies than
in determining overall vulnerability of the nervous system. Virus
strain differences may also explain apparent geographic differences
in pathology. In attempting to establish the relevance ofneurotropic
variants of HIV-1, therefore, one needs to consider not only the
presence of neurological disease but also its pathological substrate.
The virus is well known to have a high mutation rate and to show
marked polymorphism (39). Not only do isolates differ from one
patient to another, but several isolates from a single patient may
differ from each other. Putative neurotropic variants have, in fact,
been reported, with differing cell tropisms, cytopathology, and
pattems of replication in cell culture (16, 40). However, in the
absence of more direct model test systems, the biological and
Fig. 2. HIV-1 infection of brain
(A) A section ofdeep,white matter
showing spongy vacuolation, as-
trogliosis,and several multinucleat-
ed giant cells. Perivascular mono-
nuclear cells are also present (hema-
toxylin eosin, x425). (B) Infection
of perivascular multinucleated cells
with HIV revealedbyan immuno- Ii|
peroxidase technique (13) with the
use of a monoclonal antibody to
core protein, p25, (avidin-biotin
same protein, p25,
pleomorphic mticroglia with elon-
gated cytoplasmic processes (avi-
din-biotin imm-unoperoxidase with
hematoxylin counterstain, x375).
(D) In situ hybridization with a
35S-labeled RNA probe (8) to de-
tect HIV-1 RNA which is coloca-
lized in cells binding the lectin Ri-
cinis comminis agglutinin I (34), a
marker of macrophages and mi-
croglia (lectin histochemical prepa-
ration with hematoxylin counter-
, xC) The
is present mn
pathogenetic importance of these changes remains speculative.
2) Why (and how)
particular interest because productive HIV-1 brain infection appears
to be confined principally to macrophages, that is, cells extrinsic to
the brain, and to a lesser extent to microglia, their counterpart
resident in the brain. Why then the apparent selection ofthe brain as
a target? Perhaps there is less selectivity than we now recognize, and
CNS infection is, in fact, only part of a generalized infection,
attracting attention because of its functional importance. Another
altemative is that chronic leptomeningeal infection or nonproduc-
tive infection of astrocytes or other neural elements may initiate
involvement ofthe brain parenchyma. The brain has been shown to
contain messenger RNA corresponding to the CD4 molecule that
serves as the receptor for HIV-1 on lymphocytes (41). However, the
cells expressing this determinant in brain, and its role in brain
infection, remain to be clarified. If the permissiveness of macro-
phages and microglia is under immune control, the end stage of
CNS involvement with multinucleated cell encephalitis may simply
represent amplification by these elements of latent or indolent
(astrocytes, oligodendrocytes, neurons) or extraneural (leptomen-
It has also been suggested that monocyte-derived macrophages
become infected peripherally prior to migrating into the CNS: the
'Trojan horse" hypothesis (36). The large number of infected
macrophages in some brains imply that (i) infected monocytes
selectively migrate into the brain, (ii) latent infection of monocyte/
macrophages is selectively activated in the brain, or (iii) there is
secondary spread of HIV-1 to these permissive cells within the
brain. Although each ofthese mechanisms may operate, we favor the
third as the dominant one. Similarly, microglia are probably infected
in situ, although there is still some controversy about the timing of
is the brain infected? This question is of
SCIENCE, VOL. 239
their ontogeny (42). Thus, many questions remain. Some of the
most important include: Does cell-free virus in blood infect the
brain by way ofendothelial cells? Does the CSF become infected via
the choroid plexus, as with visna virus, with the virus spreading
secondarily to the brain parenchyma? What determines the distribu-
tion of virus and tissue injury within the brain so that productive
infection involves most prominently the deep gray structures and
white matter? Is this related to target cell vulnerabilities or to the
route of entry?
3) How is the CNS injured by infection? The mechanisms underly-
ing the neurological dysfunction ofthe AIDS dementia complex are
far from clear. The paucity or absence ofproductive infection or lysis
ofneurons, oligodendrocytes, or astrocytes indicates processes other
than simple destruction of these essential cells. In the subset of
patients with multinucleated cells in the brain (group 1 in Table 1),
the subcortical distribution of HIV-1-infected cells conforms to
that of the neuropathology. Microfoci of infected macrophages,
microglia, and derivative multinucleated cells are frequently accom-
panied by local edema and, uncommonly, frank demyelination. This
pattern suggests that indirect mechanisms oftissue damage predom-
inate and relate to the release of toxic substances by infected cells.
These may be cell-coded products, for example, cytokines such as
tumor necrosis factor or enzymes with bystander effects on sur-
rounding tissue (43); and they may be released either as part ofthe
physiological cellular reaction to infection or as a result of virus-
induced alteration of cell metabolism. Alternatively, products of
viral genes may be toxic to surrounding uninfected tissue. Recently,
a model of this type of indirect effect was delineated that involved
the HIV-1 envelope glycoprotein gpl20 interfering with the activi-
ty of a neurotrophic factor, neuroleukin, with which it has partial
sequence identity (44).
Thus far we have detected productive infection immunohisto-
chemically in only a minority ofpatients without mulinucleated cells
and with pathology restricted to diffuse white matter pallor (group 2
in Table 1). The pathogenesis ofinfection and brain injury in these
patients, therefore, is even less clear than in group 1 patients.
Perhaps the insensitivity of current methods prevents detection of
underlying latent or indolent infection in which there is limited
transcription ofviral genes yet impaired cell function. Although the
cellular substrate of the nearly ubiquitous brain atrophy in these
patients is not yet defined, indolent or intermittent brain HIV-1
infection, or even toxins released into the CSF by leptomeningeal
infection or into the blood by systemic infection, might reduce brain
myelin by an effect on oligodendrocyte metabolism. The diffuseness
of the white matter pallor in these patients suggests a generalized
toxic process. The functional integrity of neurons or astrocytes
might be similarly affected by toxins. Wiley and colleagues have
suggested that HIV-1 infection of endothelial cells might result in
altered vascular permeability (10).
The vacuolar myelopthy has also not yet been related to direct
HIV-1 infection in the regions of the spinal cord showing myelin
changes. In particular, oligodendrocytes that elaborate and maintain
the vacuolated myelin have not been shown to be productively
infected. A murine retrovirus in which neuronal vacuolation occurs
in the absence ofapparent virus infection ofthe neurons themselves
provides a precedent for secondary pathogenic effects of infected
cells on neighboring cells or of nonproductive infection (45).
Therapeutic Rationale and Prospects
The evidence that the AIDS dementia complex is caused either
partially or wholly by direct HIV-1 brain infection and that the virus
frequently invades the CNS early in the course ofsystemic infection
even in the absence ofsymptoms provides the rationale for seeking
antiviral drugs that can penetrate the blood-brain and blood-CSF
barriers (46). The CNS may be considered a possible reservoir for
infection, much as it provides a sanctuary for leukemic cells.
However, the clinically quiescent or latent phase of infection that
appears to continue as long as host antiviral defenses are intact
suggests that therapies capable ofrestoring immune function might
also be important in preventing or reversing CNS symptomatology.
What can be expected of antiviral chemotherapy in the AIDS
dementia complex? The relatively bland neuropathology in some
patients, the prominent affliction ofwhite rather than gray matter,
and the major involvement of cells other than neurons, astrocytes,
and oligodendrocytes in productive infection all provide a theoreti-
cal background for the hope that neurological injury may be
reversible. We have observed spontaneous substantive improvement
in some patients, and preliminary data reported by others suggest
that the antiviral nucleoside zidovudine (AZT, azidothymidine) may
be capable of ameliorating the AIDS dementia complex (47).
Further studies are now needed to confirm these findings and to
establish the overall impact ofsuch therapy, including its long-term
Additional Implications ofthe AIDS Dementia
The role ofHIV-1 in the AIDS dementia complex has rekindled
interest in the involvement ofother viruses, particularly retroviruses,
in neurological disease. This has already resulted in the discovery
that another retrovirus, human T-lymphotropic virus type
(HTLV-1), is associated with a progressive myelopathy referred to
in certain equatorial regions as tropical spastic paraparesis (TSP) and
in Japan as HTLV-1-associated myelopathy (48). One report also
suggested an association ofHTLV-1 or a related virus with multiple
sclerosis (49), but this has not been confirmed by others (50).
Nevertheless, HIV-1 infection of the CNS and the AIDS dementia
complex provide a novel view ofhuman viral pathophysiology and
strong justification for considering retroviruses in other neurologi-
cal and psychiatric disorders ofunknown etiology.
REFERENCES AND NOTES
1. B. A. Navia, B. D. Jordan, R. W. Price,Ann. Neurol. 19, 517 (1986).
2. S. L. Horowitz et al., ibid. 12, 80 (1982); G. Gopinathan etal.,Neurology 33, 105
(1983); W. D. Snider et al., Ann. Neurol. 14, 403 (1983); R. L. Levy, D. E.
Bredesen M. L. Rosenblum,J. Neurosurg, 63, 475 (1985).
3. S. L. Nielsen, C. K. Petito, C. D. Urmacher, J. B. Posner,Am. J. Clin. Pathol. 82,
678 (1984); C. B. Britton and J. R. Miller, Neurol. Clin. 2, 315 (1984).
4. B. A. Navia, E. S. Cho, C. K. Petito, R. W. Price,Ann. Neurd. 19, 525 (1986).
5. L. G. Epstein etal., ibid. 17,488 (1985); A. L. Belman etal., ibid. 18,560 (1985).
6. N. Nathanson etal., Rev. Infect. Dis. 7, 75 (1986); M. A. Gonda et al., Scince 227,
7. G. M. Shaw et al., Science 227, 177 (1985).
8. S. Koenig et al., ibid. 233, 1089 (1986).
9. M. H. Stoler, T. A. Eskin, S. Benn, R. C. Angerer, L. M. Angerer, J. Am. Med.
Assoc. 256, 2360 (1986).
10. C. A. Wiley, R. D. Schrier, J. A. Nelson, P. W. Lampert, M. B. A. Oldstone, Proc.
Natl. Acad. Sci. U.S.A. 83, 7089 (1986).
11. D. H. Gabuzda et al., Ann. Neurol. 20, 289 (1986).
12. R. Vazeux et al., Am. J. Pathol. 126, 403 (1987).
13. T. Pumarola-Sune, B. A. Navia, C. Cordon-Cardo, E. S. Cho, R. W. Price,Ann.
Neurol. 21, 490 (1987).
14. L. G. Epstein et. al., AIDS Res. 1, 447 (1985); F. Gyorkey, J. L. Melnick, P.
Gyorkey, J. Infect. Dis. 155, 870 (1987).
15. D. D. Ho et al., N. Engi. J.Mcd. 313, 1493 (1985).
16. S. Gartner, P. Markovits, D. M. Markovits, R. F. Betts, M. Popovic,J. Am. Mcd.
Assc. 256, 2365 (1986).
17. R. W. Price, J. Sidtis, M. Rosenblum, Ann. Neurol., in press.
18. S. Tross et al., AIDS, in press.
19. Proceedings ofIIIIntematiwnalConftrenceonAIDS, Washington, D.C., 1 to 5 June
20. D. Ornitz et al., in (19), p. 189; J. J. Sidtis, in ibid., p. 55.
5 FEBRUARY I988
21. M. L. Albert, R. G. Feldman, A. L. Willis,J. Neurol.Nerosurg.Psychiaty 37, 12 Download full-text
(1974); J. L. Cummings and D. F. Benson,Arcb. Neurol. 41, 874 (1984).
22. J. Jarvik t al., in (19), p. 88.
23. D. A. Rottenberg at al., Ann. Neurl. 22, 700 (1987); J. R. Moelier, S. C.
Stroher, J. J. Sidtis, D. A. Rottenberg,J. Cereb. BloodFlowMab. 7,649 (1987).
24. C. A. Came at at., Lancet 1985-II, 1206 (1985); D. A. Denning, J. Anderson, P.
Rudgc, H. Smith, Br. Med. J. 294, 143 (1987); B. J. Brew, M. Perdices, D. A.
Cooper, in (19), p. 135.
25. L. Resnick et al., N.Exgl.J. Med. 313, 1498 (1985); J. Goudsmit et a., Lancet
1986-U, 177 (1986); I. Elovaara at al.,J. Neurl. Sci. 78, 331 (1987).
26. J. A. Levy, J. Shimabukuro, J. Hollander, J. Mills, L. Kaminsky, Lanct 1985-U,
586 (1985); H. Hollander and J. A. Levy, Ann. Int. Med. 106; 692 (1987); L.
Resnick, J. R. Berger, P. Shapshak, W. W. Tourtellotte,Neuroly,in press; M. E.
Handsfield, in ibid., p. 55; J. C. McArthur et al., in ibid., p. 21.
27. H. Hollander and S. Stingari,Am. J. Med. 83, 813 (1987).
28. CDC, Morbid. Mortal. Weekly Rep. 34, 373 (1985); ibid. 36, 1S (1987).
29. B. A. Navia and R. W. Price,Areb. Neurl. 44,65 (1987); R. W. Price, J. J. Sidtis,
B. A. Navia, T. Pumarola-Sunc, D. B. Ornitz, inAIDS and TbeNr'wos System, M.
L. Rosenblum, R M. Levy, D. E. Bredesen, Eds. (Raven, New York, 1988), pp.
30. A. J. Saykin
31. C. K. Petito et al., J.Neuropathol. Exp. Neural. 45, 635 (1986); H. Budka, Acta.
Neuropathol. (Berin) 69, 253 (1986); L. R. Sharer etal., Hum. Patbol. 17, 271
(1986); RI H. Rhodes, ibid. 18, 636 (1987); T. Kato, A. Hirano, J. F. LUena, H.
M. Dembitzcr,Acta. Neuropathol. 73, 287 (1987).
32. C. K. Petito et al., N.EngI.J. Med. 312, 874 (1985).
33. M. Popovic, M. G. Sarngadharan, E. Read, R. C. Galo, Scienc 224, 497 (1984).
34. J. Michaels, R W. Price, M. K. Rosenbhlum, in preparation.
35. C. Cheng-Mayereat., Proc. Nael.AcadSci. U.SA. 84, 3526 (1987); S. Dewhurst
tol., Fed. Eur. Biodcem. Soc. 213, 138 (1987); J. Weber, E. Robey, R. Axel, R.
Weiss, in (19), p. 110.
36. L. Stowring et al., Viro
141, 311 (1985); A. T. Haase, Nature (London) 322,
37. J. M. A. Lange, D. A. Paul, F. deWolf, R. A. Coutinho, J. Goudsmit,AIDS 1, 15
(1987); I. Wendler, U. Bienzle, G. Hunsmann, AIDS Res. Hum. Retroirwses 3,
aol., in (19), p. 136; A. C. Collier, R. W. Coombs, B. Nikora, H. H.
t al., Neurlogy 37, 374 (1987); I. Grant et al., Ann. IntMed. 107,
157 (1987); F. deWolftat., Br.Med.. 295, 569 (1987); C. Pedersen a a., ibd.,
38. D. L. Walkcer, in Handbook ofClinical Neurlgy, P. J. Vinken and G. W. Bruyn,
Eds. (Elsevier/North-Holiand, Dordrecht, Netherlands, 1978), part 2, vol. 34, pp.
307-329; E. W. Willoughbyet al., Neurlgy 30, 256 (1980).
39. F. Wong-Staal
('985); B. H. Hahn e al., Proc. Nad. Acad. Sc. U.S.A. 82, 4813 (1985); B. H.
Hahn at al., Senxce 232, 1548 (1986).
40. H. Rubsamen-Waigmann at al.,J. Med. Viro. 19, 335 (1986); Y. Koyanagi at al.,
Scinca 236, 819 (1987); R. Anand at al., Lancet 1987-U, 234 (1987).
41. P. M. Maddon K al., Ge 47, 333 (1986).
42. R. L. Schelper and E. K. Adrian,J. Neuropatho. Exp. Neurol. 45, 1 (1986).
43. L. Ramer, S. H. Polmar, N. Paul, N. Ruddie,AIDS Res. Hum. Retroirue 3, 147
(1987); G. H. W. Wong, J. Krowka, D. P. Stites, D. V. Goeddel, in (19), p. 54; A.
S. Lau, S. E. Read, B. R. G. Williams, in ibid., p. 130; M. E. Eyster, J. J. Goedert,
M. C. Poon, 0. T. Preble, N. Exgl.J. Med. 309, 583 (1983); H. Smedley, M.
Katrak, K. Sikora, T. Wheeler, Br. Med. J. 286, 262 (1983).
44. M. E. Gurney, S. P. Heinrich, M. R. Loe, H.-S. Yin, Sacinc 234, 566 (1986); M.
R. Lee, D. D. Ho, M. E. Gurney, ibid. 237, 1047 (1987); B. A. Apatoff, M. R
Lee, M. E. Gurney,Ann. Neurol. 22, 156 (1987).
45. J. R Swarz, B. R. Brooks, R T. Johnson, Neurpatbol. Appl. Neurobiol. 7, 365
(1981); J. A. Bilello, 0. M. Pits, P. M. Hoffman,J. Vird. 59, 234 (1986).
46. R. Yarchoan and S. Broder, N. Eogl.J. Med. 316, 557 (1987).
47. R. Yarchoan at al., Lancet 1987-I, 132 (1987).
48. G. C. Roman,Ar. Neurol. 44,659 (1987); A.
(1985); M. Osame at al.,Ann. Neurol. 21, 117 (1987); M. Newton et al., Lant
1987-I, 415 (1987); 0. Morgan at al., ibid. 1987-U, 403 (1987).
49. H. Koprowski at al., Nature (London) 318, 154 (1985).
50. F. Lolli, S. Fredrikson, S. Kam-Hansen, H. Link,Ann. Neurol. 22, 67 (1987); S.
L. Hauser at al., Nature (London) 322, 176 (1986); A. Karpas, V. Kampf, A.
Siden, M. Koch, S. Poser, ibid., p. 177.
51. R. Redfield, personal communication.
52. R. W. Price at al., in Procedings ofH Intemationl Conferemnc on AlDS (Paris,
France, 1986), p. 27.
53. Our own studies were supported by PHS grant NS-19048, New York State AIDS
Institute grant AR-074, the Life and Health Insurance Medical Research Fund,
and the Rudin Foundation.
t al., Scenwc 229, 759 (1985); S. Bcnn et al., ibid. 230, 949
essain tal.,Lanct 1985-U,407
Education to Prevent AIDS:
Prospects and Obstacles
HARVEY V. FINEBERG
A number of obstacles thwart effective education to
prevent AIDS in the United States. These include the
biological basis and social complexity of the behaviors
that must be changed, disagreement about the propriety
of educational messages to prevent AIDS, uncertainty
about the degree ofriskto themajorityofAmericans, and
dual messages ofreassurance and alarm from responsible
officials. Long-term protection of an individual from
infection requires extreme changes in risk-taking behav-
ior. Partial shifts toward safer practices may be epidemio-
logically important in retarding the rate and extent of
spread of infection. Though some striking changes in
behavior have occurred, especially in homosexual popula-
tions in areas with high prevalence ofAIDS, educational
efforts to date have succeeded more in raising awareness
and knowledge about AIDS than in producing sufficient
changes in behavior. The United States has yetto mount a
nationwide comprehensive, intensive, and targeted educa-
tion program to prevent AIDS.
ODAY, MOSTAMERICANS VIEWTHE ACQUIRED IMMUNODE-
ficiency syndrome (AIDS) as the most serious health threat
confronting the United States (1). Approximately 50,000
Americans have been diagnosed with the disease since 1981 (2).
More than halfare dead, and no one with AIDS has yet been cured.
While researchers seek more effective therapies and biological
preventives, education and behavior change have been repeatedly
and correctly cited as the only available means of curailing the
spread of the human immunodeficiency virus type 1 (HIV-1)
responsible for AIDS.
Extensive epidemiologic investigation has affirmed the principal
means by which HIV-1 is spread-blood, sex, and birth (3). In the
United States, the major groups who have developed AIDS are men
exposed by homosexual contact, intravenous drug users who use
contaminated needles, hemophiliacs and blood recipients prior to
institution of protective measures, and the offspring of infected
The virus can be transmitted sexually from men to women and
The author is dean of the Harvard School of Public Health, Boston, MA 02115.
SCIENCE, VOL. 239