Cytokine evaluation in individuals with low back
pain using discographic lavage
Jason M. Cuellar, MD, PhDa, S. Raymond Golish, MD, PhDb, Merrill W. Reuter, MD, PhDc,
Vanessa G. Cuellar, MDa, Martin S. Angst, MDd, Eugene J. Carragee, MDb,
David C. Yeomans, PhDd, Gaetano J. Scuderi, MDb,*
aNew York University Hospital for Joint Diseases, New York, NY 10003, USA
bDepartment of Orthopaedic Surgery, Stanford University, Redwood City, CA 94063, USA
cPrivate Practice, Lake Worth, FL 33460, USA
dDepartment of Anesthesia, Stanford University, Stanford, CA 94305, USA
Received 18 February 2009; revised 20 November 2009; accepted 14 December 2009
AbstractBACKGROUND CONTEXT: The pathophysiology underlying degenerative disc disease and its
implication in painful syndromes remain unclear. However, spine magnetic resonance imaging
(MRI) can demonstrate changes in disc water content and the annulus; provocative discography
purportedly identifies degenerate discs causing serious low back pain; and biochemical assays have
identified local inflammatory markers. No study to date has correlated pain on disc injection during
discography evaluation with relevant MRI findings and biochemical markers.
PURPOSE: The purpose of this study was to correlate concordant pain on during discography to
biochemical markers obtained by disc lavage and MRI findings.
STUDY DESIGN: This is a Phase 1 Diagnostic Test Assessment Cohort Study (Sackett and
PATIENT SAMPLE: The patient sample included 21 symptomatic patients with suspected
discogenic pain and three Phase 1 control subjects.
OUTCOME MEASURES: The outcome measures included discography pain scores, MRI degen-
erative grades, and immunoreactivity to various inflammatory cytokine concentrations present in
disc lavage samples.
METHODS: Twenty-one symptomatic patients with lumbar degenerative disc disease and three
control subjects underwent discography, MRI, and biochemical analysis of disc lavage fluid.
Lumbar MRI was scored for Pfirrmann grading of the lumbar discs, and annular disruption was
identified by nuclear disc lavage. Disc lavage samples were analyzed for biochemical markers
by high-sensitivity immunoassay.
RESULTS: Eighty-three discs from 24 patients were studied: 67 discs from 21 patients with axial
back pain (suspected discogenic pain group) and 16 discs from 3 scoliosis patients without back
pain (Phase 1 control subjects). Among the biochemical markers surveyed, interferon gamma
(IFN-g) immunoreactivity was most consistently identified in patients with axial back pain. Discs
FDA device/drug status: not applicable.
Author disclosures: JMC (stock ownership, including options and
warrants, Cytonics; scientific advisory board, Cytonics; research: staff/ma-
terials, Cytonics); SRG (stock ownership, including options and warrants,
Cytonics Inc; scientific advisory board, Cytonics); MWR (stock owner-
ship, including options and warrants, Cytonics; private investments, in-
cluding venture capital, start-ups, Cytonics); VGC (stock ownership,
including options and warrants, Cytonics Corp; scientific advisory board,
Cytonics Corp); MSA (royalties, Stanford University; stock ownership, in-
cluding options and warrants, Cytonics; consulting, Trigemina Inc, Cortex
Inc, Vertex Inc; scientific advisory board, Cytonics Inc; research: investi-
gator salary, Scientific Imaginetics; research: staff/materials, Scientific
Imaginetics); EJC (consulting, Medtronic; scientific advisory board, Intrin-
sic Orthopedics, Cytonics; other office, Bioassetts; grants, AO Foundation;
fellowship support, DePuy Spine); DCY (stock ownership, including op-
tions and warrants, Cytonics Inc; scientific advisory board, Cytonics,
Inc); GJS (stock ownership, including options and warrants, Cytonics
Corp; private investments, including venture capital, start-ups, K2 Medi-
cal; board of directors, Cytonics; scientific advisory board, Cytonics; other
* Corresponding author. Department of Orthopaedic Surgery, Stanford
University, Redwood City, CA, 94063.
E-mail address: email@example.com (G.J. Scuderi)
1529-9430/10/$ – see front matter ? 2010 Elsevier Inc. All rights reserved.
The Spine Journal 10 (2010) 212–218
with annular disruption and concordant pain reproduction at a visual analog scale of 7 to 10/10 had
greater IFN-g immunoreactivity than those without this finding (p5.003); however, at least some
IFN-g immunoreactivity was found in all but one disc in the symptomatic group.
CONCLUSIONS: Among the potential inflammatory markers tested in this Phase 1 study, IFN-g
immunoreactivity was most commonly elevated in discogram ‘‘positive’’ discs but absent in asymp-
tomatic controls. However, this marker was also frequently elevated in degenerative but ‘‘negative’’
discography discs. From these findings, Phase 2 and Phase 3 validity studies are reasonable to pur-
sue. Phase 4 utility studies may be performed concurrently to assess this method’s predictive value
in outcome studies. ? 2010 Elsevier Inc. All rights reserved.
Keywords:Intervertebral disc; Cytokines; Interferon gamma; Discography; Low back pain
Chronic low back pain (LBP) remains a challenging
condition to treat. Medical expenditures by Americans with
axial back pain syndromes rose approximately 65% from
1997 to 2005 . Although imaging studies, such as mag-
netic resonance imaging (MRI), can be used to document
pathologic and age-related changes in the spine, studies
have shown a poor correlation between MRI findings and
clinical presentation. Rates of disc degeneration and herni-
ation increase with normal aging and do not consistently re-
sult in pain . Discography, which emerged as a potential
diagnostic test to determine if discs were a primary source
of axial pain, has demonstrated limited success . Alter-
native strategies have suggested that it may be the presence
of specific inflammatory mediators that determines if a disc
is clinically painful.
Although few previous studies have measured inflamma-
tory markers in discs in vivo , several in vitro studies
have postulated a role for various cytokines in disc-related
pain syndromes [5–14]. Tumor necrosis factor-a and inter-
leukin (IL)-8 have been identified in surgical disc speci-
mens removed for the treatment of pain [4,11]. Kang
et al.  identified matrix metalloproteinase, nitric oxide,
prostaglandin E2, and IL-6 in cultures of discs retrieved in
patients with a lumbar disc herniation and radiculopathy.
Similarly, Takahashi et al.  identified IL-1-beta, IL-6,
and tumor necrosis factor-a in prepared tissue specimens
from retrieved human herniated disc tissue . Burke et al.
 identified high levels of both IL-6 and IL-8 in patients
with symptomatic degenerative disc disease who underwent
spinal fusion. It has been postulated that even a small
amount of these factors may be sufficient enough to initiate
an inflammatory process after rupture of the nucleus pulpo-
sus because of their ability to recruit other cytokine-produc-
ing cells and stimulate the upregulation of genes coding for
proinflammatory mediators .
Newer methods for rapid and sensitive cytokine detec-
tion have since been developed. We hypothesized that
provocative discography and/or radiographic change on
MRI would correlate to specific inflammatory mediators
detected in the disc space by using a highly sensitive
protein immunoassay on samples obtained during disc
This study was performed from June 2006 to June 2008.
Institutional review board approval was obtained according
to Health and Human Services Guidelines, and patients
provided informed consent at the time of enrollment.
Symptomatic subjects’ screening
Twenty-one patients ranging in age from 21 to 75 years
with axial back pain of at least 6 months’ duration were
enrolled in this study (study group). The patients were iden-
tified among 119 consecutive patients offered study
enrollment for the evaluation of their chronic pain. This
cohort was drawn from a single board-certified orthopedic
surgeon trained in spine surgery (G.J.S.). Patients with
a history of oral or injected corticosteroid medication
within a 3-month period before discography and those with
chronic medical conditions associated with metabolic or in-
flammatory disorders (insulin-dependent diabetes mellitus,
severe coronary arterydisease,
autoimmune diseases) were excluded from the study.
gender, age, insurance, work status, and reported pain on
a visual analog scale (VAS) of 0 to 10 before the procedure.
Provocative discography, recording annular disruption, pain
intensity, and concordance for each disc injected, was per-
formed by an experienced discographer.
A blinded analysis of MRI was performed with each
injected disc level classified according to the Pfirrmann
grading scale  by a board certified radiologist.
Phase 1 control subjects
A Phase 1 Diagnostic Validity Control (ie, application of
test in subjects definitively known not to have the index dis-
ease) by Sackett and Haynes Criteria  was also per-
formed using disc lavage samples from asymptomatic
subjects and adolescents undergoing surgical treatment
for scoliosis deformity. Disc lavage, without subsequent
discography contrast injection, was carried out intraopera-
tively before excision of each disc. Each lavage was done
under the supervision of the lead investigator (G.J.S.). As
described in the Sackett and Haynes model for assessing di-
agnostic test validity, this Phase 1 assessment is a necessary,
213J.M. Cuellar et al. / The Spine Journal 10 (2010) 212–218
but not sufficient, first screening for test validity. That is, if
the test is positive in this group (ie, the finding of a specific
cytokine at high concentrations), this will provide evidence
against test validity.
Disc lavage technique
All symptomatic study participants underwent lumbar
discography to evaluate their chronic axial pain. The patient
was placed prone on a radiolucent table and prepared using
standard sterile procedure. Intravenous antibiotics were
given before the commencement of discography. After infil-
trating the subcutaneous tissue with 1% xylocaine with epi-
nephrine, a 22-gauge needle was used to enter the annulus,
followed by fluoroscopic placement confirmation. Approxi-
mately 1 to 3 mL of sterile physiologic saline was injected
into each disc space, and the lavage fluid was immediately
withdrawn back into the same syringe, yielding ~0.5 mL of
lavasate, which was then placed into 2-mL Eppendorf tubes
containing a protease inhibitor cocktail tablet (Roche Diag-
ered saline to reach a total volume of 130 mL (~0.045 tablet/
mL per sample) and immediately frozen at ?20?C. Samples
were then transported to the laboratory where the fluid was
stored at ?80?C until analysis.
A nucleogram with Omnipaque was performed at all
levels considered as potential pain sites, in addition to
a ‘‘control’’ injection at a level with negative findings on
MRI in each subject. Pain provocation, intensity, concor-
dance, and annular disruption were recorded.
The Bio-Plex system (BioRad, Hercules, CA, USA),
a multiplex bead array immunoassay, was used for analysis
of lavasate. This system uses a sandwich-style immunoassay
with capture antibodies linked to polystyrene beads and
detection antibodies conjugated to fluorophores. It has been
validated against standard enzyme-linked immunosorbent
assays in human blood . The immunoreactivity (immu-
nofluorescence) of 17 cytokines/chemokines (interferon
gamma [IFN-g], IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10,
IL-12, IL-13, IL-17, granulocyte colony-stimulating factor,
granulocyte macrophage colony-stimulating factor, tumor
tory protein-1Beta (MIP-1b) was measured using the pre-
mixed human inflammatory 17-plex panel. Each multiplex
assay was performed in duplicate and according to the man-
ufacturer’s specifications. Standard curves were included in
each run, and sample concentrations were calculated with
Bio-Plex Manager software from these fluorescence mea-
Statistical analyses were performed with SPSS v.14
(SPSS Inc., Chicago, IL, USA). The study was designed
by a prior power analysis. For N567 discs (study group
analysis), the power of the t test of significance of the Pear-
son correlation statistic with an alpha value of 0.05 is
99.7% for a large effect size (Pearson r50.5). The power
of the chi-square test of a two-by-two contingency table
analysis with an alpha value of 0.05 is 98.4% for a large
effect size (w statistic50.5). The power of the F test of sig-
nificance of multiple regression with an alpha value of 0.05
is 99.2% for a large effect size (F2statistic50.35).
A total of 83 discs from 24 patients were studied: 67 discs
from 21 patients with LBP (study group) and 16 discs from 3
Phase 1 control subjects (scoliosis patients) with no pain.
In the LBP group, there were 13 men and nine women
with mean age 44611 years (range 20–64) and mean pre-
operative VAS of 7.461.3. In the LBP groups, IFN-g
was the only cytokine tested that was consistently identi-
fied, with mean (6standard deviation), median, and range
of immunoreactivity corresponding to a concentration of
2766455, 136, and 4.7–2,814.0 pg/mL, respectively. Of
the other 16 cytokines, mean and median calculated con-
centrations were zero except for IL-2 (mean51.361.7,
There are no unique radiographic or pathologic markers
that accurately identify disc pathology that is causing
chronic low back pain.
inflammatory markersdidnotcorrelate withdegeneration
tivity was consistently present in patients with low back
modestly correlated with disc injection pain.
Despite years of investigation, no unique marker or tech-
nique has been found to accurately identify primary dis-
provocative or anaesthetic discography, inflammatory
markers, and other candidates have thus far demon-
strated poor test validity. This has led some to suggest
this search is misdirected (the illness is not primarily
structural) and others to suggest the search will be
fruitful with the emergence of newer technologies.
While the current findings are intriguing, it is unclear if
IFN is simply a marker for degeneration (rather than
painful degeneration) and if disc puncture and lavage
is a safe and reproducible sampling method. However,
these results suggest further exploration is reasonable.
214J.M. Cuellar et al. / The Spine Journal 10 (2010) 212–218
median50.6, range50–9.3) and granulocyte macrophage
colony-stimulating factor (mean511.1620.8; median50;
range 5 0–83.8).
In the asymptomatic control group, there were two
males and one female with mean age of 1660.6 years
(range 15–16). The mean (6standard deviation) and
median concentration of IFN-g were 2.766.1 and 0.4 pg/
mL, respectively (range 0.0–23.0).
The level of IFN-g detected in lavage samples from
discs of symptomatic subjects was significantly greater than
that detected in discs from asymptomatic control subjects
(p5.003; one-way analysis of variance [ANOVA] control-
ling for age) (Fig. 1).
Analysis of IFN-g immunoreactivity in symptomatic
In the symptomatic group, IFN-g concentrations were
consistent with an exponential (log10) distribution but not
a normal distribution (Fig. 2; p !0.001). Subsequent
regression and correlation
performed with log10(IFN-g) concentrations. The results
did not change with the use of untransformed IFN-g
Analysis of IFN-g concentrations with respect to the
pain intensity rating during provocative discography results
was performed (Fig. 2). Using a cutoff pain intensity value
of 6 (as recommended by Walsh et al. ) to separate pos-
itive from negative painful injection, there was no signifi-
cant difference in IFN-g concentrations by t test (Fig. 3;
p5.43). Using a cutoff pain intensity value of 7 (as
recommended by Derby et al. ), there was a significant
difference in IFN-g concentrations (Fig. 3; p5.003).
These results did not change (ie, significance obtained
with a cutoff value of 7 but not with 6) using a matched
analysis in which IFN-g levels were normalized (either by
subtraction or division) by the ‘‘control level’’ of the disco-
gram (ie, the level tested by discogram with the lowest intra-
operative VAS score) or with the use of log10 IFN-g
measurements. Therewas no significant correlation between
IFN-g concentration and injection pain intensity with
ANOVA or log-linear analysis with the use of main effects
and two-way effects when the data were not dichotomized
into painful versus not painful groups with a cutoff value.
Analysis of IFN-g versus Pfirrmann grade on MRI
Analysis of IFN-g concentrations with respect to Pfirr-
mann grade on MRI was performed (Fig. 4) using a group-
ing of grades 1 to 2 as mild, grades 3 to 4 as moderate, and
grade 5 as severe. There was no significant difference in
IFN-g concentration by t test. The results did not change
with the use of log10IFN-g measurements, with the use
of ANOVA or log-linear analysis using main effects and
two-way effects when the data were not dichotomized pain-
ful and not painful groups with a cutoff value, or with
a matched analysis in which IFN-g levels were normalized
(by either subtraction or division) by the ‘‘control level’’ of
the discogram (ie, the level tested by discogram with the
lowest Pfirrmann MRI grade).
Analysis of discography versus MRI findings
Statistical analysis of the relationship between discogra-
phy and MRI findings was undertaken irrespective of IFN-
g concentration (Table 1). In the study group, there were no
Fig. 1. A comparison of interferon gamma concentration in picograms per
milliliter (pg/ml; on a log scale) between patients with low back pain
(N567 discs) who underwent discography and those without back pain
(asymptomatic scoliosis controls; N516 discs). The error bars represent
the standard error of the mean. The difference is significant by one-way
analysis of variance controlling for age (p5.008).
Fig. 2. A scatter plot showing interferon gamma concentration in pico-
grams per milliliter (pg/ml; on a log scale) from all lavage samples taken
during discograms of patients with axial pain. Patient-reported pain during
discography was rated on a visual analog scale (VAS) of 1 to 10 and is rep-
resented on the x-axis.
215J.M. Cuellar et al. / The Spine Journal 10 (2010) 212–218
partial)amongprovocativediscogrampain intensitywith in-
jection (VAS) and Pfirrmann MRI grade by Pearson correla-
the use of Spearman correlations, ANOVA, or log-linear
analysis or with the use of main effects and two-way effects.
Our understanding of the pathophysiology of axial back
pain on a molecular level continues to evolve but remains
an area of active investigation. On a clinical level,
differentiating the subset of discs that may cause seriously
painful back disability from either normal aging or clini-
cally painless degeneration is similarly difficult. Provoca-
tive discography and MRI are tests that physicians
consider in identifying an underlying source of back pain.
However, recent studies suggest that discography and
MRI are at best modestly predictive of outcomes following
interventions directed at suspected ‘‘discogenic’’ pain .
In light of these limitations, a clinical assay of biochemical
markers to complement other diagnostic modalities is
The study provides a Sackett and Haynes Phase 1 anal-
ysis of a possible diagnostic test for identifying discogenic
pain. This first-stage diagnostic test assessment is intended
to survey the results of lavage technique quantification of
inflammatory markers in discs from subjects suspected to
have discogenic pain and compare these to the findings in
discs definitively without discogenic pain (asymptomatic
scoliosis subjects) and discs highly unlikely to be the
Fig. 4. A comparison of interferon gamma concentration in picograms per
milliliter (on a log scale) among patients with magnetic resonance imaging
changes graded by Pfirrmann scores 1 to 2, 3 to 4, and 5. The error bars
represent the standard error of the mean. The differences are not significant
by analysis of variance (p5.77).
Fig. 3. (Top) A comparison of interferon gamma (IFN-g) concentration in
picograms per milliliter (on a log scale) between patients with self-
reported pain on discography rated on a visual analog scale (VAS) of 1
to 10, grouped by VAS of 1 to 5 and 6 to 10, with mean IFN-g concentra-
tion (6standard error of the mean) of 246648 and 3426145, respectively.
The differences are not significant by t test (p5.43). (Bottom) A compar-
ison of IFN-g concentration in picograms per milliliter (on a log scale)
between patients with pain on discography grouped by visual analog scale
of 1 to 6 and 7 to 10, with mean (6standard error of the mean) IFN-g
concentration of 213639 and 6866309, respectively. The differences are
significant by t test (p5.003).
Summary of statistical analysis among intraoperative provocative disco-
gram visual analog scale (VAS; 1–10 scale), magnetic resonance imaging
Pfirrmann grade (1–5 scale), and preoperative VAS (1–10 scale)
MRI Pfirrmann grade
r50.07; p5.59r50.12; p5.35
Data are Pearson correlation (r) and p values; there are no significant
differences (p value greater than .05). The results do not change with the
use of Spearman correlations.
216J.M. Cuellar et al. / The Spine Journal 10 (2010) 212–218
source of discogenic pain (morphologically normal discs
without pain on disc injection in LBP symptomatic sub-
jects). It should be emphasized that this Phase 1 study can-
not provide a high confidence in positive test validity.
However, if the test were repeatedly positive in discs and
subjects without discogenic pain, then the prima facie
evidence would indicate that this is very unlikely to be
a valid test with Phase 2 or Phase 3 investigations.
The present study showed mixed results. IFN-g immu-
noreactivity was the most consistent positive assay detected
in the lavage fluid of discs sampled in symptomatic sub-
jects. Higher levels of IFN-g immunoreactivity were
obtained from discs with the greatest pain response on disc
injection. However, some IFN-g immunoreactivity (greater
than 10 pg/mL) was detected in all but one disc in the
symptomatic group, even from those discs with relatively
normal MRI findings or an apparently ‘‘negative control’’
discography designation. On the other hand, the samples
from disc lavage in the asymptomatic group did not show
appreciable IFN-g immunoreactivity.
Although various inflammatory cytokines have been
identified in the intervertebral disc as it undergoes degener-
ation, it is not known if any of these specific inflammatory
chemicals plays a significant role in axial LBP [13,21–23].
Inflammatory cytokines have been previously identified in
cerebrospinal fluid of patients with symptomatic degenera-
tive disc disease, although the source and concentration of
inflammatory mediators at the level of the symptomatic
nerve root are unknown [7,24]. It has been postulated that
inflammatory mechanisms are most likely involved in the
painful state in axial pain syndromes [4,6,24,25]. Identifi-
cation of matrix fragments, nonenzymatic glycation
products, and neuropeptides is representative of the many
ongoing products and processes that occur with aging
and degeneration but does not necessarily generate
A practical sampling technique for specific biomarkers
that may accurately predict serious LBP generated intrinsic
to that disc would greatly help our understanding of symp-
tomatic degenerative disc disease and perhaps direct appro-
priate clinical care. Although further research with more
detailed biochemical peptide analyses to delineate its role
is required to confirm our findings, this study suggests that
IFN-g immunoreactivity may be an important clinical
biomarker for some axial pain syndromes associated with
common degenerative findings.
It is interesting that our preliminary data suggest that the
correlation between discography findings, MRI findings,
and levels of biochemical markers of inflammation was
not strong in this patient population using this sampling
technique. This finding may reflect the heterogeneity of
cytokine concentrations in different parts of the disc. The
lavage technique will necessarily retrieve substances
preferentially from the nuclear center, as opposed to the
subannular or subligamentous space. Spot sampling in
the epidural space or along the annulus or in specific
annular fissures may yield different results. It is also prob-
able that some LBP syndromes with common degenerative
findings may be because of primarily mechanical or other
processes withouta strong
In conclusion, this study suggests that IFN-g immunore-
activity from a disc lavage sampling has at least prelimi-
nary Phase 1 assessment support for potential test validity
in discogenic pain. From these findings, Phase 2 and Phase
3 validity studies are reasonable to pursue. Phase 4 utility
studies may be performed concurrently to assess this meth-
od’s predictive value in outcome studies.
 Martin BI, Deyo RA, Mirza SK, et al. Expenditures and health status
among adults with back and neck problems. JAMA 2008;299:656–64.
 Boden SD, Davis DO, Dina TS, et al. Abnormal magnetic-resonance
scans of the lumbar spine in asymptomatic subjects. A prospective
investigation. J Bone Joint Surg Am 1990;72:403–8.
 Carragee EJ, Lincoln T, Parmar VS, Alamin T. A gold standard eval-
uation of the ‘‘discogenic pain’’ diagnosis as determined by provoc-
ative discography. Spine 2006;31:2115–23.
mimics nucleus pulposus-induced neuropathology. Molecular, histo-
logic, and behavioral comparisons in rats. Spine 2000;25:2975–80.
 Scuderi GJ, Brusovanik GV, Golish SR, et al. A critical evaluation of
discography in patients with lumbar intervertebral disc disease. Spine
pholipaseA2activity in lumbar disc herniations. Spine1990;15:674–8.
zyme activities in CSF of low back pain patients. Pain 1990;43:163–8.
 Lindh C, Thornwall M, Hansen AC, et al. Neuropeptide-converting
enzymes in cerebrospinal fluid: activities increased in pain from her-
niated lumbar disc, but not from coxarthrosis. Acta Orthop Scand
 Imasato H, Nagata K, Hashimoto S, et al. Objective evaluation of
pain in various spinal diseases: neuropeptide immunoreactivity in
the cerebrospinal fluid. Spinal Cord 1997;35:757–62.
 Ahn SH, Cho YW, Ahn MW, et al. mRNA expression of cytokines
and chemokines in herniated lumbar intervertebral discs. Spine
 Olmarker K, Rydevik B. Selective inhibition of tumor necrosis fac-
tor-alpha prevents nucleus pulposus-induced thrombus formation,
intraneural edema, and reduction of nerve conduction velocity: possi-
ble implications for future pharmacologic treatment strategies of sci-
atica. Spine 2001;26:863–9.
 Kang JD, Georgescu HI, McIntyre-Larkin L, et al. Herniated lumbar
intervertebral discs spontaneously produce matrix metalloproteinases,
 Takahashi H, Suguro T, Okazima Y, et al. Inflammatory cytokines in
the herniated disc of the lumbar spine. Spine 1996;21:218–24.
 Burke JG, Watson RW, McCormack D, et al. Intervertebral discs
which cause low back pain secrete high levels of proinflammatory
mediators. J Bone Joint Surg Br 2002;84:196–201.
 Koes BW, Sholten R, Mens J, Bouter LM. Epidural steroid injection
for low back pain and sciatica: an updated systematic review of ran-
domized clinical trials. Pain Digest 1999;9:241–7.
 Pfirrmann CW, Metzdorf A, Zanetti M, et al. Magnetic resonance
classification of lumbar intervertebral disc degeneration. Spine
217J.M. Cuellar et al. / The Spine Journal 10 (2010) 212–218
 Sackett SL, Haynes RB. The architecture of diagnostic research. BMJ Download full-text
 de Jager W, te Velthuis H, Prakken BJ, et al. Simultaneous detec-
tion of 15 human cytokines in a single sample of stimulated
peripheral blood mononuclear cells. Clin Diagn Lab Immunol
 Walsh TR, Weinstein JN, Spratt KF, et al. Lumbar discography in
normal subjects. A controlled, prospective study. J Bone Joint Surg
 Derby R, Kim BJ, Lee SH, et al. Comparison of discographic find-
ings in asymptomatic subject discs and the negative discs of
chronic LBP patients: can discography distinguish asymptomatic
discs among morphologically abnormal discs? Spine J 2005;5:
 Gronblad M, Virri J, Tolonen J, et al. A controlled immunohisto-
chemical study of inflammatory cells in disc herniation tissue. Spine
 Goupille P, Jayson MI, Valat JP, Freemont AJ. The role of inflamma-
tion in disk herniation-associated radiculopathy. Semin Arthritis
 Carlton SM,Hargett GL.
Ca(2þ)/calmodulin-dependent protein kinase II alpha-containing
dorsal root ganglion neurons in the rat: colocalization with isolectin
Griffonia simplicifolia, calcitonin gene-related peptide, or vanilloid
receptor 1. J Comp Neurol 2002;448:102–10.
 Brisby H, Olmarker K, Larsson K, et al. Proinflammatory cytokines
in cerebrospinal fluid and serum in patients with disc herniation and
sciatica. Eur Spine J 2002;11:62–6.
 Olmarker K. Radicular pain—recent pathophysiologic concepts and
therapeutic implications. Schmerz 2001;15:425–9.
 Kang JD, Stefanovic-Racic M, McIntyre LA, et al. Toward
and herniation. Contributions of nitric oxide, interleukins, prostaglan-
din E2, and matrix metalloproteinases. Spine 1997;22:1065–73.
218 J.M. Cuellar et al. / The Spine Journal 10 (2010) 212–218