Discography was first described in 1948 as an investigative technique for herniated nucleus pulposus. Since that time, new imaging techniques that are more appropriate for this diagnosis have been developed. Discography is currently used to determine whether the disk is the source of pain in patients with predominantly axial back or neck pain. 
During discography, contrast medium is injected into the disk and the patient's response to the injection is noted; provocation of pain that is similar to the patient's existing back or neck pain suggests that the disk might the source of the pain. Computed tomography (CT) is usually performed after discography to assess anatomical changes in the disk and to demonstrate intradiscal clefts and radial tears.
Early studies suggested that discography had a low specificity, but more recent studies have failed to induce pain in asymptomatic controls, suggesting that discography has utility in identifying patients with discogenic pain. Pain reproduction during discography in symptomatic individuals is variable, with a lower incidence of pain reproduction in patients with disk degeneration than in those with posterior tears of the anulus fibrosus or significant disk bulges.
Controlled clinical trials of discography are lacking, and a standard against which to compare is elusive. When comparing outcomes of fusion procedures, lumbar discography is sensitive but lacks specificity. See the images below.
Discs form the main connection between vertebrae. Their size varies depending on the adjacent vertebrae size and comprises approximately one quarter the length of the vertebral column.
These disks are composed of 4 parts: the nucleus pulposus in the middle, the annulus fibrosis surrounding the nucleus, and 2 end plates that are attached to the adjacent vertebral bodies (see the image below). They serve as force dissipators, transmitting compressive loads throughout a range of motion. The disks are thicker anteriorly and therefore contribute to normal cervical lordosis.
Internal Disc Disruption and Pain Provocation Theories
The cardinal lesion that renders a lumbar disk painful is internal disk disruption. The characteristic feature of internal disk disruption is a radial fissure extending to the innervated outer third of the annulus fibrosus. As radial fissures extend to the outer third of the annulus, nerve endings are exposed to the inflammatory and algogenic chemicals produced by nuclear degradation. As a radial fissure develops, fewer and fewer lamellae remain intact to bear the load. At some stage, the threshold for mechanical nociception will be attained, especially if the nerve endings have been chemically sensitized. Disk stimulation reveals this condition by showing a reduced threshold for mechanical stimulation of the disk.
In theory, discography provokes pain by the following mechanisms:
The injection of contrast material into the disk may increase intradiscal pressure. In an abnormal disk, stretching of the annular fibers of the disk may stimulate nerve endings.
The injection may result in some biochemical or neurochemical stimulation that causes pain.
The injection may increase pressure at the end plates, or pressure may be transferred to the vertebral body throughout the end plate, resulting in an increase in intravertebral pressure. This theory is supported by studies reporting disk injection resulting in end-plate deflection and increased specimen height.
The presence of pain on injection of a seemingly normal disk may be due to transfer of pressure from the injection to an abnormal, symptomatic adjacent disk, thus eliciting a positive pain response.
Discography should be performed only if adequate attempts at conservative therapy and noninvasive diagnostic tests, such as MRI, have failed to reveal the etiology of back pain.
Specific indications for discography include the following:
Persistent, severe symptoms when other diagnostic tests have failed to clearly confirm a suspected disk as a source of the pain
Evaluation of abnormal disks or recurrent pain from a previously operated disk or lateral disk herniation
Assessment of patients in whom surgery has failed, to determine whether pseudoarthrosis or a symptomatic disk in a posteriorly viewed segment could be the source of pain
Assessment of disks prior to fusion to determine whether the disks of the proposed fusion segment are symptomatic and whether the disks adjacent to this segment can support a fusion
Assessment of candidates for minimally invasive surgery who have a confirmed disk herniation
Complications associated with discography include spinal headache, meningitis, diskitis, intrathecal hemorrhage, arachnoiditis, severe reaction to accidental intradural injection, damage to the disk, urticaria, retroperitoneal hemorrhage, nausea (2%), seizures (4%), headache (10%), and increased pain (81%).
In rare cases, discography has been found to result in disk herniations. Five cases of acute lumbar disk herniation precipitated by discography have been reported. New-onset or a persistent exacerbation of radicular symptoms following provocative discography merits further investigation. 
The incidence of diskitis is 2-3% when a single-needle technique is used and 0.7% when a double-needle technique is used. The incidence of diskitis might decrease to less than 1% when prophylactic antibiotics are used.
See the list below:
Prophylactic antibiotics should be considered.
Double-needle technique always should be used.
The injection should be performed with water-soluble contrast medium.
Accurate needle placement is required to avoid annular injections, which could produce false-positive results.
Injection against the vertebral end plate can cause a false-positive response.
Discography should be followed by CT scanning.
The information recorded should include the following:
- Resistance to the injection (ie, end point)
- Amount of contrast material injected (ie, maximum volume)
- Volume at which the patient experienced pain (ie, pain volume)
- Pattern of dye distribution (eg, diffusion, location of fissure, extravasation, herniations, Schmorl nodule)
- Pressure at which patient experienced "pressure sensation"
- Pressure at which patient experienced "pain"
- Pain response (ie, location, character, distribution, intensity, and concordance or discordance with the patient's typical pain and pain pattern)
- Pain intensity recorded on a 0-10 scale
See the list below:
Very careful attention should be paid to interpreting the pain response during the injection of each disk, including whether the pain is similar to or exactly like the symptoms for which the patient seeks relief. The location of the pain and its intensity should be noted.
Pain at low pressures is most likely due to chemical irritation. Low resistance generally is associated with a tear through the outer annulus. Pain at high pressures may be due to mechanical irritation, end-plate deflection, or stimulation of pressure receptors.
Generally, if a large volume of contrast can be injected, the disk is degenerated or has a fissure extending through the outer annular wall.
The specificity of discography findings could be increased if CT scanning findings are correlated. Annular disruption reaching the outer annulus fibrosus is a key factor in pain generation. Disk morphology, including annular disruptions extending beyond the outer annulus, may permit increased discography specificity. 
Dallas classification of discography results
See the list below:
Type 1: The discogram is normal manometrically, volumetrically, and radiographically and produced no pain. The discogram/CT scan showed contrast to be located centrally in the axial and sagittal projections. See the image below.Postdiscography CT scan showing normal disk contour.
Type 2: This is identical to type 1 except that it is positive for reproduction of pain.
Type 3: Annular tears lead to a radial fissure. This group is subdivided further into types 3a (ie, posterior radial fissure), 3b (ie, fissure radiates posterolateral), and 3c (ie, fissure extends lateral to a line drawn from the center of the disk tangential to the lateral border of the superior articulating process).
Type 4: Once the radial fissure reaches the periphery of the anulus fibrosus, nuclear material may protrude, causing the outer annulus to bulge.
Type 5: When the outer annular fibers rupture, nuclear material may extrude beneath the posterior longitudinal ligament and come in direct contact with either the dura or a nerve root.
Type 6: When the extruded fragment is no longer in continuity with the interspace, it is said to be sequestrated. Manometrically, volumetrically, and radiographically, the discograms are always abnormal. Familiar pain may be reproduced only if enough pressure is generated against the free fragment to cause stimulation of the pain-sensitive structures.
Type 7: The end stage of this degeneration is internal disk disruption, characterized by multiple annular tears. The discograms are abnormal manometrically and volumetrically, and familiar pain may or may not be reproduced. Radiographically, the contrast usually fills the entire interspace in a chaotic fashion. The discogram/CT scan shows contrast extravasation throughout multiple annular tears. See the image below.Postdiscography CT scan showing some abnormal dye spread.
Pros and Cons of Lumbar Discography
Discography provides valuable information to the clinician and the patient. CT myelography and MRI are used to detect disk herniations and other space-occupying lesions that may compromise lumbar nerve roots. CT scans and MRI are excellent for investigating radicular pain, but offer little in the investigation of back pain and somatic referred pain. [4, 5]
Opponents of discography refer to studies by Holt.  However, the means and methods used by Holt have been scathingly refuted.  His studies have been replicated using more stringent conditions, with blinded investigators, independent observers, and manometrically monitored discography. Under these conditions, lumbar disks do not hurt in asymptomatic individuals and disk stimulation is a highly specific diagnostic test. For a disk to be deemed painful, stimulation must reproduce the patient's accustomed pain, provided that stimulation of the disk above or below (preferably both) does not reproduce pain.
Some surgeons have proclaimed that by selecting the correct disk for treatment, discography leads to greater success rates than anterior lumbar fusion. Disks selected for treatment are those that are symptomatic on stimulation and that express loss of signal intensity on MRI. Failing to find a painful disk on discography should preclude surgery; so too should finding multiple painful disks or obtaining indeterminate results. Discography is only a diagnostic tool to test whether a disk is painful. Unjustified surgery can be prevented by heeding indeterminate or negative results.
Some authors believe that the test has no proven efficacy in improving patient outcomes, and that it leads to inappropriate surgery. These authors also believe that discography was popularized and adapted before validity and utility were determined. Disagreement on discography involves 3 major areas.
The first area of disagreement on discography is the concept of internal disk disruption as a symptom-producing complex. Proponents of discography theorize that the cardinal lesion that renders a lumbar disk painful is internal disk disruption. However, authors that oppose discography believe that the concept is a combination of a variety of anatomic and physiologic facts garnered from disparate sources and cobbled together to provide a theory to support the concept of "internal disk disruption."
The second area of disagreement is the contention that discography is important as an informational tool in "internal disk disruption" to help us understand what is or is not wrong. What is the point of an "informational tool" for the purpose of establishing a diagnosis for which no proven therapy exists?
The third area of disagreement is that discography leads to inappropriate surgery. Nachemson stated that the origin of back pain remains unknown in a majority of patients.  The benefit of surgery for low back pain and sciatica at the present time is proven in scientific matter only for disk herniation giving nerve root pain.
Review of Published Studies
Derby et al completed a multicenter retrospective study of long-term surgical and nonsurgical outcomes after lumbar discography in 96 patients.  After positive lumbar discogram, patients underwent interbody fusion alone, combined fusion, intertransverse fusion, or no surgery. Those who underwent interbody/combined fusion had significantly better outcomes than those who underwent intertransverse fusion. Nonsurgical patients had the worst outcomes overall.
Parker et al prospectively studied 23 patients treated by a single surgeon.  All underwent preoperative discography and were monitored for an average of 4 years postoperatively. Thirty-nine percent had a good-to-excellent result, 13% a fair result, and 48% a poor result.
Smith et al did a study of 25 patients who had a positive discogram of disk L4-5 or L5-S1; the analysis was retrospective, with mean follow-up of 5 years.  Sixty-eight percent had improved, 8% were the same, and 24% had worsened. Outcome was not correlated with disk level, gender, or smoking history. Patients who had improved had a shorter history of low back pain and an older age of onset (45 y vs 33 y). Of patients who had worsened, 67% had psychiatric disease. Eighty percent of those receiving workers compensation had improved.
Knox and Chapman performed a study in which 22 patients undergoing anterior lumbar interbody fusion for discogram-concordant lower back pain were evaluated retrospectively.  Results were poor in all 2-level fusions. In single-level fusions, 35% had good, 18% fair, and 47% poor results.
Wetzel et al did a retrospective review of 48 patients with low back pain who had discogram/CT then lumbar arthrodesis.  Forty-six percent were judged to have satisfactory clinical outcome at final follow-up.
Motimaya et al performed a retrospective study of 46 patients who underwent cervical disk examination by discography.  They then evaluated results of 14 of the 16 patients who underwent cervical spine fusion at those levels in accordance with positive results on discogram. The average symptomatic period prior to discography was 12 months, and cervical disk pain was localized in all 16 patients. After discectomy and anterior fusion, all 14 patients had good-to-excellent results at 6 months.
Siebenrock and Aebi retrospectively reviewed 27 patients who underwent fusion of 39 cervical levels for discogenic pain.  The source of pain in all patients was identified by positive discography. Fusions were performed via a ventral approach and included 22 1-level, 7 2-level, and one 3-level procedures; iliac bone graft was done in all patients. Seventy-three percent had good-to-excellent results, 23% had fair results, and 3.8% had poor results. More good-to-excellent results were seen after 2-level than after 1-level fusion (86% vs 62%). Furthermore, patients with pain to the upper limbs did better.
Connor and Darden did a retrospective review of 31 patients who had positive discogram.  Twenty-two underwent anterior cervical discectomy and fusion. One patient had a good-to-excellent result, 41% had good results, and 54% had fair to poor results. Whitecloud and Seago retrospectively reviewed 34 patients who had cervical arthrodesis after positive discogram.  Seventy percent of the patients had good-to-excellent results.
Heggeness et al performed a retrospective review of 83 patients presenting with disabling back pain who had previously undergone surgical treatment for herniated nucleus pulposus.  They found a high incidence (72%) of concordant pain with discography of the previously operated level. Only 34% of the previously operated disks demonstrated posterior extravasation of discography dye. Persistence of a posterior annular defect was associated with a higher incidence of concordant pain.
Brightbill et al reported on a clinical series of 7 patients who had surgically proven internal disk disruption, normal results on MRI, and abnormal morphology on discogram.  They concluded that discography may be useful in patients with persistent symptoms despite a normal or equivocal MRI.
Bernard prospectively studied 250 patients with low back pain who underwent lumbar discography followed by CT scan.  In 93% of the patients, the combined discogram/CT provided significant information regarding equivocal or multiple level abnormalities and type of herniation, defining surgical options, and evaluating previously operated spines. In 94% of the patients, discography/CT correctly predicted disk herniation as protruded, extruded, sequestrated, or internally disrupted.
Lam et al did a prospective blinded study in which they found a significant correlation between abnormal disk morphology and the high-intensity zone (HIZ) on MRI.  In morphologically abnormal disks (grades 3, 4, 5), a significant correlation between HIZ and reproduction of exact or similar pain is typical. Sensitivity, specificity, and positive predictive value for pain reproduction were 81%, 79%, and 87%, respectively. See the image below.
Saifuddin et al retrospectively reviewed 99 lumbar discogram reports in which 260 disks were injected and 179 were abnormal.  They found that pain experienced in buttock, hip, groin, or lower limb can arise from the posterior annulus without direct involvement of the root.
Simmons et al performed a study in which 164 patients, all with low back pain, underwent discography and MRI.  Discography and MRI results correlated in 80% of the cases. Of abnormal disks, 76% reproduced symptoms on discography.
Antti-Poika et al did a prospective study of 279 injected disks in 100 patients.  Exact reproduction of pain on injection was more common in fissured or ruptured disks. The results indicated that discography had a sensitivity of 81% and specificity of 64% for pain. Additional information yielded by follow-up CT scan was minimal.
Schellhas et al conducted a retrospective study of patients until records of 100 HIZ disks in 63 patients were found.  Eighty-seven of the 100 disks tested were found to be concordantly painful. All 87 showed annular tears to the outer third of the anulus fibrosus. Of the 67 non-HIZ disks also studied, 64 were nonconcordant and of lower sensation intensity. In patients with symptomatic lower back pain, the HIZ is a reliable marker of painful outer annular disruption.
Scuderi GJ et al embarked on a prospective observational study of 48 continuous patients with symptomatic lumbar degenerative disk disease to correlate concordant pain on discography with MRI grade and biochemical markers of inflammation in a clinical setting.  The correlations between demographic, discogram, and radiographic variables are weak. The authors concluded that the response to discogram cannot be predicted by non-invasive means. The disk lavage method was unable to identify the presence of specific inflammatory peptides with multiplex immunoassays and enzyme-linked immunoassay (ELISA).
Caragee et al conducted a prospective study of 8 patients (24 disks) with no history of lower back pain who had undergone posterior iliac bone graft (1999). He found that 50% experienced concordant pain of the usual gluteal area. Thus, the ability of a patient to separate concordant pain on discography may be less meaningful than often assumed.
In another study by Caragee, 26 patients without lower back pain were studied prospectively after discography.  He surmised that, in a subject group with no lower back pain but with significant emotional and chronic problems, discography might result in reports of significant back pain. In fact, he found that for at least 1 year after injection, 66% of the somatization group and 40% of the abnormal psychometric test result group did indeed have significant back pain. Of 11 subjects with normal psychometric test results, none reported significant long-term back pain after discography.
In yet another study by Carragee, he reviewed a clinical series showing that pain intensity during discography injection is influenced strongly by the subject's emotional and psychological profile. Pain reproduction could not be used reliably to confirm the location of pain source; it was related primarily to the penetration of the outer annulus by the dye. 
Finally, Carragee also prospectively studied patients with and without lower back pain after undergoing laminotomy and subsequent discectomy.  Of 240 patients who had undergone single-level discectomy, 20 asymptomatic patients with normal psychometric test results were recruited for 3-level discography. A control group consisted of 27 symptomatic patients who had undergone single-level discectomy. The asymptomatic patients had a 40% rate of positive injection, while the symptomatic group had a rate of 63%.
Grubb and Kelly conducted a retrospective study of 173 cervical discograms over 12 years.  Of the 807 disks injected, 50% yielded concordant pain response. More than half of the discograms yielded 3 or more painful disks (more than expected).
Manabu performed a study in which 101 lumbar disks in 39 patients were studied with MRI and discography. He found that although lumbar disks with posterior combined annular tears are likely to produce pain on discography, this provoked pain was not predictive of discogenic pain.
Provocation discography (PD) is used in confirming or refuting a particular disc as a source of a patient's pain. The premise: "If a particular disc is painful, then stressing it should reproduce the patient's usual pain. If the disc is not the source of a patient's pain, then stressing it either should not be painful or should produce pain that is not the patient's familiar or accustomed pain" (International Spine Intervention Society [ISIS] Guidelines, Chapter 7). 
Despite its shortcomings, PD is the current standard for the diagnosis of painful internal disc disruption. Until recently, the injection of local anesthetic into the disc had not been well-studied, with only a handful of studies published over the last 60 years. 
To better diagnose and treat discogenic pain, interest has been renewed in various methods of analgesic discography (AD), including the recent introduction of functional analgesic discography (FAD). Analgesic and provocation discography are, in many ways, opposite sides of the same coin. Whereas PD attempts to confirm the disc as a pain source by reproducing a patient's usual symptoms, AD attempts to relieve those symptoms. Analgesic discography can be used alone or in combination with PD. FAD is a combination of analgesic discography and functional testing.
Alamin compared the ability of FAD and PD to predict favorable surgical outcome in 41 patients.  All patients underwent preprocedure functional testing to determine which activities were painful and rated the pain that each activity elicited. Standard provocation discography was then performed on all patients. A balloon-tipped catheter was inserted into patients' PD positive discs before leaving the operating suite. In the recovery room, the catheter(s) were sequentially injected first with normal saline followed by testing with 0.75 mL of local anesthetic (4% Xylocaine or 0.75% bupivacaine). The patients then underwent repeat functional testing within 20 minutes in positions that typically provoked their pain.
In a 2007 NASS abstract, Alamin reports the following: "7 of the 41 (17%) patients had 2-level findings on provocation discography that were reduced to 1-level findings on the FAD test. 11 patients (27%) had positive provocation discograms that were negative on FAD testing. Two patients (5%) had a negative provocation discogram and yet pain relief on the FAD. 21 patients (51%) had confirmatory findings on the FAD test. DRAM profile of distressed depressive (DD) or distressed somatic (DS) was a significant predictor of negative findings on the FAD test."  Alamin’s reported 44% false-positive PD rate per patient (27% of patients with single-level positive PD had a negative FAD; 17% with two-level positive PD reduced to one level with FAD) has not been substantiated by other researchers to date.
Although no studies have directly compared surgical outcomes in patients with positive PD results vs patients with positive AD results, several studies attest to the ability of each to identify a painful internally disrupted disc. In patients with positive results from PD, Derby et al showed that surgeons achieved good to excellent outcomes following single level interbody fusion supplemented with pedicle screw fixation and posterior lateral fusion in 62% of patients having a SF-36 mental component summary score (MCS) of greater than 40. 
Carragee, using much stricter success criteria, reported surgical outcome following single-level fusions for internal disk disruption (IDD) with positive discograms ranging from 33% for highly effective results to 75% meeting minimum FDA standards of a ≥2/10 decrease in Visual Analogue Scale (VAS) score and ≥10 decrease in ODI. 
Most recently, Cooper et al showed that discography predicted response to fusion.  Discograms were assigned scores based on the integrated shape and imaging system (ISIS) scoring system. Positive discograms (score >70) and indeterminate discograms (scores >50) predicted response to fusion surgery. The breakpoint was 50; with an ISIS discogram score >50, patients who underwent fusion surgery were 5 times more likely to return to ≥25% of normal daily activities, 3.4 times more likely to return to ≥50% daily activities, and 3.3 times more likely to have less pain compared with patients with a similar ISIS score who elected not to have surgery.
The use of discography remains controversial. However, the literature supports the use of discography in selected patients. Particular applications include patients with persistent pain in whom disk abnormality is suspected but noninvasive tests have not provided diagnostic information or needs to be correlated with clinical symptoms. Discography can also be helpful in the assessment of disks in patients in whom fusion is being considered and in whom pain remains after surgery.