++
The lumbar intervertebral disc can be injected with contrast,
local anesthetic, or other substances. Although observing the effect
of a local anesthetic injection on pain and function (analgesic
discography) can potentially provide useful information, the clinical
utility of this has not been well defined. However, the effect of
an applied mechanical or chemical stimulus on pain (provocative discography)
has demonstrated clinical utility.
++
In addition to provoking pain that can be compared with the patient’s
clinical symptoms, injecting contrast into the disc may demonstrate
pathology that is not otherwise revealed on conventional imaging
studies. Prior to the introduction of sophisticated imaging studies
such as computed tomography (CT) and magnetic resonance imaging
(MRI), lumbar discography was often used primarily as a radiologic
imaging study to complement myelography.25 Several
studies have confirmed the accuracy of lumbar discography as a radiologic
test in demonstrating both disc herniations and disc degeneration.2,26–31 With
the advent of MRI scanning, and in particular the use of gadolinium
enhancement in evaluating postoperative patients, the utility of
lumbar discography purely as a radiologic study has diminished.
However, both cadaver and clinical studies have demonstrated that
discography is more sensitive than MRI in detecting disc degeneration,
particularly when postdiscography CT scanning is added.2,28,31–33
++
The purpose of discography is to determine whether the intervertebral
disc is a source of clinical symptoms. Although interpretation of
the radiologic images obtained at the time of discography is important,
in contemporary practice discography is primarily a provocative
clinical test, rather than a radiologic imaging procedure.
+++
Provocative
Discography
++
A variety of pathologic processes affect the intervertebral disc,
potentially causing noxious stimulation of nerve endings. A precision
injection of contrast dye into the disc nucleus also stimulates nerve
endings. The stimulus applied with discography has two components:
a chemical stimulus resulting from contact between contrast dye
and sensitized tissues, and a mechanical stimulus resulting from
a fluid-distending stress. The underlying premise of discography
is that this applied stimulus replicates the clinical noxious stimulus
responsible for the patient’s symptoms, and that reproduction
of the patients clinical symptoms during the injection confirms
the disc as the source of pain.
++
As with any diagnostic test, it is important to know the false-positive
and false-negative rates associated with provocative discography,
which are used to calculate sensitivity and specificity. Defining
the false-positive and false-negative rates requires comparing the
test results against a gold standard.17,34,35 A
gold standard is a method for definitively establishing a diagnosis
and is typically obtained from biopsy, surgery, or long-term follow-up.17 Unfortunately,
in contrast to radicular pain, there is no absolute method to determine
the tissue origin of lumbar axial pain, and, therefore, no way to
determine the sensitivity and specificity of discography.
++
A variety of factors can lead to both false-positive and false-negative
results from provocative discography. One possible source of both
false-positive and false-negative injections is that the stimulus
applied with the injection may not be selective for the nerve endings
in the disc being studied. The nerve endings in the lumbar disc
are in the end plates and middle and outer annulus. Although pathologic
processes involving the end plate can occur,36,37 the
innervated portion of the disc most often affected by pathologic
processes is the annulus. For the provocative discography construct
to be valid, the injection must selectively affect the nerve endings
in the annulus of the disc being studied, which is the presumed
site of the clinical noxious stimulus. A number of authors have
suggested alternate sources for the pain of discography, other than
stimulation of annular nerve endings.38 Postulated
pain mechanisms include increased pressure at the end plates or
within the vertebral body,39 increased substance
P and VIP in the dorsal root ganglion,40 or transmission
of mechanical stimulation to the facet joints. Despite these hypotheses,
there is good evidence that the provocative response resulting from
discography is related to stimulation of nerve endings in the outer
annulus, rather than other factors.41–43 Although
relatively unusual, painful end-plate disruptions can also occur.36
++
The complex nature of anatomic structures can also lead to inaccurate
results from discography. Anatomic structures are typically composed
of several different types of tissues. A pathologic process can
potentially affect just one component of a structure, which may
not be the same component targeted by a precision injection. For
example, discogenic pain is commonly felt to be a result of annular
fissures originating in the nucleus and extending to the outer annulus,
which is where the majority of the disc nerve endings reside. During
discography, contrast is injected into the nucleus. If there are
not fissures, the contrast will be confined to the nucleus. However,
histologic studies have demonstrated that there can be middle or
outer annular abnormalities that are not contiguous with the nucleus.28 In
such cases, an injection into the nucleus could lead to a false-negative
result.
++
Another potential source of inaccurate results from discography
is the change in central nervous system (CNS) nociceptive processing
that occurs with chronic pain. The neuroanatomic pathways mediating
acute pain behave as a hard-wired system, with a pure stimulus-response
relationship.44 However, these neuroanatomic pathways
are plastic, as they change with the development of chronic pain.
With chronic pain, central sensitization occurs and dorsal horn
cell activity no longer depends on peripheral tissue injury.44,45 A
pure stimulus-response relationship no longer exists. Both previously
innocuous stimuli to the dorsal horn and stimuli from outside the
original receptive field cause pain. As a result, the interpretation
of diagnostic injections based on an acute pain paradigm may be
inaccurate.45 In the presence of chronic pain,
it is possible that an anesthetic injection of an injured nerve
or structure may not produce complete pain relief, anesthetizing
an adjacent normal nerve or structure may relieve pain, and provoking
a normal structure or nerve may reproduce a patient’s clinical
pain.
++
Finally, psychological factors are important sources of false-positive
results with discography. There are two components to pain. The
first is the nociceptive process initiated by tissue injury, and
the second is the psychological and emotional reaction to nociception.
A patient with chronic pain should always be seen in the context
of these interacting factors. Measuring the response to a diagnostic
injection, and in particular to provocative discography, always
relies, to some extent, on a patients’ self-reports of
pain. Therefore, psychological factors can clearly affect the measurement
of the response to a diagnostic injection. As a result, when assessing
patients’ responses to diagnostic injections, the relative
contribution of nociception and psychological factors should be considered
and the reliability of patients’ self-reports of pain estimated.
++
Clearly, there are a number of potential sources of both false-positive
and false-negative responses with discography. In an effort to study
the potential for false-positive results, several studies have investigated
the ability of discography to provoke back pain in asymptomatic
subjects.
++
Holt, in 1968, reported a 36% rate of positive discography
in asymptomatic subjects, leading him to discredit the use of the
test.46 However, there were several methodologic
flaws with this study. The most notable flaws were that all of the
subjects were prisoners, a highly irritating contrast medium was
used, and, most importantly, Holt did not include a positive pain
response as a criterion for a positive injection (i.e., the criteria
for a positive result were based primarily on radiologic images).
++
Holt’s findings were subsequently refuted in a well-designed
study by Walsh and colleagues, who demonstrated a 0% rate
of positive discography in asymptomatic volunteers.47 Walsh
and colleagues studied ten asymptomatic subjects and seven patients
with chronic low back pain. The criteria for a positive result differed
between the two groups. For both groups, a positive result required
a 3-out-of-5 pain intensity (using a pain thermometer), two types
of pain behavior (as assessed by videotape review), and structural
degeneration. For the patients with chronic low back pain, a positive
result also required that the provoked pain be similar to their usual
pain. Obviously, it was not possible to evaluate the similarity
of pain in asymptomatic subjects, as they had no pain prior to the
injection. Among the asymptomatic subjects, five of ten had at least
one structurally abnormal disc; however, none satisfied the criteria
for a positive test. Thus, the false-positive rate in these asymptomatic
volunteers was 0%. Among the chronic low back pain patients,
all seven had at least one structurally abnormal disc, and six of
seven patients had at least one disc that satisfied the criteria
for a positive result. Overall, 13 discs were structurally abnormal,
with 7 being positive and 6 negative. Of note, two of the seven
had at least one disc that was structurally abnormal and was associated
with intense, but atypical, provoked pain, as well as pain behaviors.
In each case the test result was considered to be negative, given
that the provoked pain was different from the patient’s
typical pain.
++
The Walsh study was important for several reasons. Firstly, using
strict criteria for a positive test, including postinjection review
of videotaped responses, there was excellent interrater reliability. Diagnostic
tests that rely on an observer’s interpretation are not
clinically useful unless there is good interobserver reliability
(i.e., the same test applied to the same patient should always produce
the same result).77 Thus, Walsh and colleagues’ study
established reproducible criteria for a positive result from discography.
Secondly, by demonstrating a 0% false-positive rate in
asymptomatic subjects, Walsh and colleagues effectively refuted
Holt’s assertion that the false-positive rate of discography
was so high as to make it useless. Thirdly, Walsh and colleagues
demonstrated that patients suffering from chronic low back pain
were capable of developing different types of pain in response to provocation
discography. According to their criteria, only provoked pain that
was similar to the patients typical symptoms constituted a positive
test. Atypical provoked pain, even if intense and accompanied by
pain behaviors, constituted a negative test.
++
The asymptomatic subjects studied by Walsh and colleagues were
all healthy volunteers, with an average age of 23. Caragee and colleagues
recently expanded on the Walsh study of asymptomatic subjects, by
studying a cohort of subjects who did not have low back pain, but
whose clinical characteristics more closely matched those of patients
with low back pain who typically present for discography.23 Thirty
subjects with no history of low back pain were recruited: ten had
previous cervical surgery with good results, ten had the same surgery
but had persistent chronic pain, and ten had primary somatization
disorders. Lumbar discography was performed and interpreted according
to the Walsh protocol. Four somatization patients dropped out before
beginning the study and two stopped the study after only one or
two discs were injected, and, therefore, were not included in the
study analysis.
++
Among the subjects with good results from previous cervical surgery,
seven of ten had at least one disc that had an outer annular rupture
(10 of 30 discs, total), while only one of ten had a positive result.
The patient with a positive test had a high Zung depression score.
Of the subjects with chronic pain, five of ten patients had at least
one disc that had an outer annular rupture (11 of 32 discs, total),
with four of ten having at least one positive disc. Of the 11 discs
with significant structural abnormalities, 7 were positive and four
were negative. Among the subjects with somatization disorder three
quarters had at least one disc that had an outer annular rupture
(6 of 13 discs, total), with three quarters having at least one
positive disc. Of the six discs with significant structural abnormalities,
two were positive and four were negative. Based on these data, Caragee
and colleagues concluded that in individuals with normal psychometrics
and without chronic pain, the rate of false-positives is very low
if strict criteria are applied, and that the false-positive rate increases
with increased annular disruption.
++
The study by Caragee and colleagues is important for a number
of reasons. Firstly, it confirms the finding by Walsh and colleagues
that in subjects without a history of low back pain, and without psychosocial
risk factors, provocation of a significant pain response with discography
is unusual, with an incidence of 0% in the Walsh study
and 10% in the Caragee study. It also confirms the Walsh
finding that although discs in this population are often structurally
abnormal (combining the studies, 12 of 20 subjects had at least
one structurally abnormal disc) they are no more likely to be positive
than a structurally normal disc. More importantly, the Caragee study
reveals that in subjects without a history of low back pain, but
with a history of chronic pain or a somatization disorder, provocation
of a significant pain response with discography is common, with
an incidence of 40% in the chronic pain group and 75% in
the somatization disorder group. Furthermore, the more disrupted
the annulus, the greater is the chance of a positive response.
++
Caragee and colleagues’ study is a powerful reminder
of the importance of psychosocial factors in modulating pain, while
also demonstrating the potential of false-positive responses with
discography. However, in assessing the importance of this information,
it is necessary to reconsider the premise of discography.
++
The premise of discography is that reproduction of a patients’ clinical
symptoms during the injection identifies the disc as the source
of pain. The rationale for its use is that the results can help discriminate
among the various structures that may be responsible for axial pain.
Therefore, to establish its validity, the criteria for a true positive
disc must be determined in the relevant population, which is back
pain sufferers. The Walsh data on patients with chronic low back
pain demonstrated that it is common for patients undergoing discography
to have intense pain that is very different in location and character
from their clinical symptoms. In the Walsh study, the criteria for
a positive test in the chronic low back pain population required
that provoked pain be similar to the patient’s clinical
symptoms. Unfortunately, without a gold standard for axial pain,
the validity of incorporating measures of familiarity of pain into
the criteria for a positive test cannot be precisely defined.17
++
Caragee and colleagues clearly demonstrated the potential for
false-positive responses with discography. However, given the premise
of discography, and the fact that patients with chronic low back
pain frequently have intense but atypical pain during discography,
it is difficult to know the significance of any pain response in
an asymptomatic subject.
++
Although the sensitivity and specificity of provocative discography
cannot be precisely defined, it is important to remember that the
ultimate criterion for a diagnostic test is whether the patient
is better off as a result. If a test can predict the response to
treatment, and is reliable and reproducible, then it may be clinically
useful,17,34,35 even without a defined sensitivity
and specificity. The primary indication for provocative discography
is to determine whether a patient with chronic spinal pain, who
has failed aggressive efforts at conservative care, can be helped
with spinal fusion.
++
In contrast to the surgical treatment of radiculopathy, the surgical
treatment of axial pain is controversial, as studies have demonstrated
a wide disparity in outcomes.7–15,48 This
disparity has been attributed to a number of factors, including
type of fusion (interbody versus intertransverse, instrumented versus
noninstrumented), approach (anterior versus posterior versus 360
degree), surgeon variability, and methodologic differences. The
results from discography have been an important part of the preoperative
evaluation in most of these studies, but typically, the criteria
for a positive test have not been strictly defined. The validity
of the criteria used to define a positive discogram is another variable
that could potentially affect surgical outcome.
+++
Pressure-Controlled
Discography
++
In an effort to develop criteria for discography that can be
used to predict surgical outcome, Derby and colleagues reported
on a cohort of patients who underwent provocative discography under pressure
monitoring.41 Although the pathophysiology of lumbar
discogenic pain is still uncertain, there is presumptive evidence
that it results from both mechanical stimulation of nociceptors
in the annulus as well as by chemical irritation by enzymes and
breakdown products involved in the degradative process.41
++
Physiologic loading of the disc creates horizontal and vertical
stresses within the nucleus, annulus, and end plates of the disc
that are directly related to the weight of the body above the segment and
any added moment stresses resulting from body position. The relationship
between intradiscal pressure and body position has been quantified
by several investigators.19,49 Derby and colleagues
hypothesized that some discs were more sensitive to chemical stimuli
than mechanical stimuli. Pain at discography which occurred at low
pressures, below the typical weighted values, would result from
chemical stimulation of the outer annulus by contact with contrast
dye. Pain occurring at higher pressures would result from mechanical
stimulation of the annulus by the fluid-distending stress of discography.
++
In order to establish the criteria for chemical and mechanical
stimulation, Derby and colleagues used data from a preliminary study
on disc pressure measurements at the time of discography.41 In
a preliminary study, they combined provocative discography with
measurement of intradiscal pressure, comparing results from discography
performed in the lying position with results from discography performed
in the sitting position. The criteria for a positive result was
6-out-of-10 concordant pain. In normal discs, the average opening
pressure, representing the intrinsic pressure of the disc, was 27
pounds per square inch (psi) in the side-lying position and 85 psi
in the sitting position. As the degree of degeneration increased,
the opening pressure decreased in both positions. However, the threefold
difference between the opening pressure in the sitting position
versus the lying position was maintained between equally degenerated
discs. In the majority of discs, concordant pain provocation occurred
when contrast first reached the outer annulus, with the maximal
pain response usually occurring at pressures only 10 to 30 psi above
the opening pressure. From these findings, Derby and colleagues
concluded that in degenerated discs with annular disruption, pain
provocation during discography is usually caused by low-pressure
stimulation of an irritable outer annulus by a chemical stimulus.
++
Based on this information, the authors created a protocol for
grading the sensitivity of the disc annulus that could be used to
predict surgical outcome. Four categories of discs were defined.
In chemical discs, pain is provoked at minimal pressure; 15 psi
above opening pressure was chosen as the threshold for a chemical
disc, as this is well below the mechanical load resulting from sitting.
In mechanical discs, pain is provoked at pressures between standing
and lying; that is, between 15 and 50 psi above opening pressure.
In indeterminate discs, pain occurs between 51 and 90 psi above
opening pressure, and in normal discs, there is no pain.
++
This classification system was applied to a consecutive series
of patients referred for lumbar discography prior to potential fusion
surgery. Following discography, patients were returned to the care
of their referring surgeons, who independently decided whether surgery
was indicated, and if so, whether it should be an intertransverse
or interbody fusion. The disc classification was not reported to
the surgeon.
++
The subjects were contacted at two follow-up intervals, at a
mean time of 16 and 32 months, with the overall outcome classified
as favorable or unfavorable depending on the results from three
different outcome tools. Looking at all surgical cases combined,
there was no significant difference in outcome between patients
undergoing interbody versus intertransverse fusion, with both groups having
approximately a 50% favorable outcome.
++
However, among patients classified as having a chemically sensitive
disc, there was a highly significant difference in outcome between
patients undergoing interbody versus intertransverse fusion. Within
that group, 89% of the interbody fusion patients had a
favorable outcome, while only 20% of the intertransverse
fusion patients had a favorable outcome. Patients with chemically sensitive
discs who did not have surgery of any kind had an 88% unfavorable
outcome. There was no significant difference in patient demographics,
including the percentage of patients with worker’s compensation
claims, between the patients with favorable outcomes and unfavorable outcomes.
Other than workers’ compensation status, psychosocial risk
factors were not assessed.
++
Until now, the clinical significance of degenerative disc disease
in a patient with axial pain was uncertain, as treatments directed
specifically at the disc (i.e., fusion) led to variable outcomes. Based
on the data of Derby and colleagues, it now appears that there is
a subset of patients with degenerative disc disease who have chemically
sensitive discs, and who have outcomes with surgery that rival those
of patients undergoing partial disc excision for herniated nucleus
pulposus. The surgery performed must be an interbody fusion, presumably
because the disc is completely excised, therefore removing the source
of the noxious stimulus. If these results stand up to long-term
follow-up, and are replicated by other investigators, then the use
of pressure-controlled discography as a diagnostic test to predict
patients who will benefit from surgical fusion will be validated.
++
In addition to potentially having the ability to predict outcome,
adding pressure monitoring to provocative discography improves interobserver
reliability and, therefore, reproducibility. Assessing the response
to discography requires measuring pain before and after the injection.
There are three components to pain: its intensity, location, and
character. If the location and character of the pain provoked at
discography is similar to or exactly the same as the patient’s
clinical symptoms, it satisfies the criteria for concordant pain.
The intensity of pain is measured both by the patient’s
self-report (e.g., using a numerical rating) and by observed pain
behaviors. However, the intensity of provoked pain is dependent
on the intensity of the stimulus. In simple terms, the harder one pushes
on the syringe the more likely the disc is to hurt. By measuring
intradiscal pressures, the intensity of the stimulus can be quantified,
allowing more reliable comparisons between patients and discographers.
Although it is possible to estimate injection pressures manually,
using a controlled inflation syringe with digital pressure readout
provides a precise value.
++
There are two approaches to the lumbar disc: posterior and lateral.50 The
posterior approach necessitates a dural puncture, and therefore
should be avoided. Disc puncture is typically performed with a 22-
or 25-gauge needle). There is some evidence that using an introducer
needle can reduce the risk of infection, although this is not a
universal practice.51 Although rarely encountered,
a variety of complications are possible with discography, including
neural injury, bleeding, and intradural leakage of injected substances.38,52,53 There
have been case reports of disc herniations resulting from discography.54,55 Canine
studies have had conflicting results on the potential for disc injury
during discography; however, the weight of evidence in humans suggests
that this is not a significant problem.55–58
++
The most significant risk associated with discography is infection.
The rate of discitis reported in the literature is as high as 1.3% per
disc, and serious morbidity has resulted.51,59,60 However, practice
audits at centers performing a large volume of discography have
demonstrated infection rates as low as 0 out of 10,000 (R. Derby,
personal communication). There is experimental evidence that prophylactic
antibiotics, both intravenous and intradiscal, can prevent discitis.61–64 As
a result, many practitioners routinely administer prophylactic antibiotics,
particularly to high-risk patients such as diabetics.
+++
Summary of Lumbar
Discography
++
The primary utility of discography is as a provocative clinical
test for the evaluation of axial spinal pain. The usefulness of
any diagnostic study is critically dependent on the critical diagnosis and
the pretest probability that a particular disorder is present.17,34,35 Therefore,
the results from discography must always be interpreted in the context
of the patient’s clinical presentation. In particular,
prior to discography there should be a careful assessment of psychosocial
risk factors.
++
An underlying assumption of the rationale for diagnostic injections,
including provocative discography, is that it is possible to accurately
measure pain. Measuring the change in pain after an injection relies
to a large extent on a patient’s self-report. Psychosocial
factors affect the reaction to changes in nociceptive input, and,
therefore, self-reports of pain. If patients are psychologically distressed
the rationale for the injection may be invalid. This can potentially
lead to both false-positive and false-negative responses. In assessing
patients’ responses to diagnostic injections, the relative
contribution of nociception and psychological factors should be
considered, and the reliability of patients’ self-reports
of pain estimated. Caragee and associates demonstrated the potential
for psychological factors to affect the results from discography.
Although more study is needed in this area, at a minimum it is important
to be aware of psychosocial risk factors and understand how they
might affect interpretation of the results from diagnostic injections.
Moreover, the implications of psychosocial risk factors on eventual
treatment, regardless of any results on diagnostic injections, should
be considered. As an example, patients with somatization disorders
are probably not good candidates for spinal fusion for axial pain.
Therefore, discography is not a clinically useful test in those
individuals and should only be performed in exceptional circumstances.
++
If a patient is being considered for surgical fusion for axial
pain, pressure-controlled discography is indicated to determine
whether they have chemically sensitized discs, in which case the
data of Derby and colleagues suggests a high likelihood of success
with an anterior interbody fusion.41 At present
it is not clear what the appropriate treatment should be for discs
falling into either the mechanically sensitive or the indeterminate
category. Preliminary information suggests that intertransverse
fusion alone may be effective for mechanically sensitive discs.41 Future
research should focus on validating the disc classification system
proposed by Derby and colleagues, and, in particular, further studying
the predictive value of the mechanically sensitive and indeterminate
categories.
++
Although not the primary focus of discography, it is important
to assess the radiologic findings. The images should be reviewed
to confirm that the needle was placed into the nucleus and that
the subsequent dye injection fills the nucleus. Annular injections,
injections into the space between the annulus and the nuclear cavity,
and venous uptake are all possible and may invalidate the results.65,66
++
Several classification schemes have been developed to describe
annular pathology as visualized by discography.25,33,67–70 Regardless
of the exact classification scheme used, it is important to note
the degree of annular degeneration, the presence of annular fissures,
and whether the annulus is competent or incompetent. If a patient
has a convincing pain response but no evidence of a radial annular
fissure on discography, postdiscography CT scanning should be considered,
as some discs that appear normal on discography are found to be
disrupted on CT discography.33 In addition to annular
pathology, both Schmorl’s nodes and end-plate disruptions
should be noted, as they may be clinically significant.36,37
++
If a patient has at least one disc that is both normal structurally
and does not elicit a pain response, that is considered by some
surgeons to serve as a control disc. A control injection may be
helpful in deciding whether a pain response at another disc is a
true positive result or reflects an exaggerated reaction to nociception.
However, a more valid control would probably be a structurally abnormal
disc. Although the significance of the results from control injections
has not been formally validated, clinical experience suggests that
if at least one structurally abnormal disc does not hurt, then pain
provoked at another disc is more likely to be a true positive.
++
As a final note, there are some surgeons who do not believe that
discography is necessary prior to fusion, as they feel that the
diagnosis of discogenic pain can be made by clinical and radiographic criteria.
There are substantial data suggesting that the clinical examination
is of minimal use in discriminating between potential axial pain
generators.71,72 A possible exception to this is
a McKenzie mechanical assessment, which may be able to predict the
results from discography.73 There have been several
studies demonstrating that MRI cannot reliably predict which discs
are painful on discography, at least to the level of confidence
required to rely solely on MRI for surgical decision making.2,31,74 A
high-intensity zone in the posterior annulus, as visualized on MRI, has
recently been proposed as a marker for painful discs.75,76 Although
highly specific, the sensitivity of this finding is only 26%,
which limits the usefulness of the high-intensity zone in selecting
patients for surgery.33 If a patient undergoes
a fusion for lumbar axial pain without preoperative discography,
both the patient and the surgeon should be aware that the level
adjacent to the planned fusion may be a source of clinical symptoms,
regardless of the findings on MRI scan.