Spinal Morphometry As A Novel Predictor For Recurrent Lumbar Disc Herniation Requiring Revision Surgery: Results of A Case Control Study
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.360932
Source of Support: None, Conflict of Interest: None
Keywords: Morphometry, predictive analysis, recurrent lumbar disc herniation, whole lumbar spine
Recurrent lumbar disc herniation (rLDH), defined as repeated disc herniation occurring more than 6 months after the primary surgery, is one of the most frequent indications for revision surgery in the first 2 years after index discectomy., Pain and functional outcomes after revision lumbar disc surgery are generally worse than those after primary surgery.,, Preoperative identification of predictors of rLDH hence becomes all the more relevant. Varied and conflicting risk factors for rLDH including modifiable lifestyle variables and biomechanical or surgical factors have been identified.,,,, While some spinal morphometric variables have been reported to be risk factors for primary disc herniation and spinal stenosis,,,,,,, these have not been evaluated as possible predictors of rLDH. Considering this gap of knowledge, the objective of this study was to evaluate if preoperative lumbar spinal morphometry could predict which patients require revision surgery for rLDH.
Study design and patient population
This was a retrospective matched case-control study with a case-control ratio of 1:4 and has been reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. All the 250 patients in the study cohort had undergone micro-discectomy for a single level lumbar disc prolapse over a 10-year period from January 2006 to January 2016. The “cases” in the study group consisted of 45 consecutive patients operated for rLDH. These patients had clinical and MRI evidence of recurrent disc herniation at the same level and location (central/paracentral) as before, after a symptom-free period of at least 6 months after the index surgery. The control group consisted of 180 patients matched for age, sex, and technique (laminectomy/fenestration) who did not develop symptomatic rLDH within a 3 year follow-up period after the index surgery. The hold-out validation set consisted of 25 new patients, of whom 5 were consecutive rLDH patients, and the remaining were patients without rLDH matched for age, sex, and technique.
Patients selected for the index surgery had either failed conservative treatment for at least 6 weeks or had presented with neurological deficits. Their MRI scans had shown a significant disc herniation corresponding anatomically to their symptoms and signs. The exclusion criteria were degenerative or isthmic spondylolisthesis and post-traumatic disc prolapses. Patients with recurrence of pain due to nondiscogenic canal or foraminal stenosis, epidural fibrosis, segmental instability, and disc herniation at a different level were excluded from the study group.
Informed written consent was obtained from all patients prior to surgery. A uniform surgical technique was followed by all neurosurgeons in the department, all of whom had similar experience levels at performing the procedure. Under general anesthesia, patients were positioned prone on a Wilson frame. After radiographic confirmation of the level, a posterior midline longitudinal incision was made. The paraspinal muscles were elevated sub-periosteally to approach the interlaminar space. A micro lumbar retractor was used to expose the interlaminar space. Following a fenestration or laminectomy, a transverse annulotomy and aggressive (”subtotal”) discectomy was then performed after removal of the extruded or protruded fragments. None of the patients underwent a “limited” discectomy restricted to the removal of only the extruded or loose disc fragments.
Clinical and demographic variables
A host of demographic and clinical variables including some of the previously identified predictors of rLDH were recorded. The demographic variables included age, lifestyle (sedentary versus active), body mass index, diabetes, and smoking while the clinical variables included the duration of symptoms and the presence or absence of radicular pain, neurogenic claudication, motor and sensory deficits, and bladder involvement.
Morphometric variables pertaining to the WLS and variables related to the herniated disc level were recorded. For all the measurements, a neurosurgery final year resident (VR) and a consultant radiologist (SN) functioned as independent observers, and the mean of their recordings of the continuous variables was taken for analysis. In case of any discrepancy in the recording of the categorical variables, the opinion of an arbitrator (ST) was taken. Both the observers were blinded to the group allocation (cases versus controls). A uniform imaging protocol was used for all patients in both groups. Axial T2 stack, sagittal T2, and coronal T2 STIR MR images of the lumbosacral spine were reviewed on the hospital radiographic system (Synapse, Fujifilm Health Systems, USA).
The morphometric variables included central canal dimensions (area, circumference, and AP diameters) [Figure 1]a and [Figure 1]b, interfacet distance (distance drawn between the medial most aspects of the facets) [Figure 1]c, foraminal zone width [Figure 1]d and dural sac dimensions (area, circumference, AP, and transverse diameters) [Figure 2]a, [Figure 2]b, [Figure 2]c at all levels of the WLS including the level of the herniated disc. Ligamentum flavum area at the disc level was measured as the cumulative ligament area on either side [Figure 2]d. For all these morphometric measurements, the ratios of the measurements to those of the L1 vertebra (rather than the absolute values) were used to eliminate biases due to varying builds of the patients.
Variables recorded at the herniated disc level were: Pfirmann grading of disc degeneration, type of disc herniation (protrusion or extrusion/sequestration), presence of a T2- high-intensity zone within the annulus [Figure 3]a, mid-sagittal disc height [Figure 3]b, and exiting root indentation. Root signal abnormality was noted on coronal T2 short inversion time inversion recovery (STIR) images. Facetal arthropathy was graded as mild, moderate, and severe depending upon the degree of joint space narrowing, arthropathy, and hypertrophy. Modic endplate changes of the adjoining endplates were noted. Lumbar lordosis was measured between the lines drawn along the T12 inferior endplate and S1 superior endplate. The range of motion at the disc level and for the entire lumbar spine (T12–S1) was calculated on dynamic radiographs.
Using the odds ratio and overall risk of developing rLDH from previous studies, a sample size of 225 (45 cases and 180 controls) with a case-control ratio of 1:4 was found to be adequate for attaining a power of 90%. Continuous variables were expressed as mean ± standard deviation while frequency distributions were used to describe the categorical variable. Bivariate analysis was used to identify preoperative variables that correlated with the occurrence of rLDH. Logistic regression analysis was done using a purposeful variable selection process. Variables with a P value cut-off of 0.10 in the bivariate analysis were selected for the multivariate analysis. Covariates were removed from the model if they were non-significant and not confounders. Significance was evaluated at a 0.1 alpha level, and confounding was defined as a change of >20% in the remaining parameter estimate as compared to the full model. An algorithm was developed employing the regression coefficients to quantify the impact of each predictor on the outcome. Receiver operator characteristic (ROC) curves were generated to calculate the area under the curve (AUC) and predictive values. The algorithm with the best predictive value was then selected. Validation of the model was performed on a hold out data set of 25 patients.
Demographic and clinico-radiological characteristics
The male: Female ratio was 32:13, and the mean age at presentation was 41.78+/- 9.61 years [Table 1]. The mean duration of symptoms prior to the index surgery was 30.86+/- 15.31 months. The most affected disc level was L4–5, with Pfirmann grade III disc degeneration seen in a majority of the patients. Of the 45 patients in the rLDH group, 20 had undergone fenestration and discectomy while 25 had undergone laminectomy and discectomy. The corresponding numbers in the control group were 80 and 100 patients, respectively.
Amongst the various clinical and radiological variables tested for bivariate associations with rLDH [Table 2], age, BMI, diabetes mellitus, smoking, motor deficits, interfacet distance at the disc level, and the WLS dural sac circumference and WLS interfacet distance had correlations with P values =/<0.10. These variables were then subjected to LR analysis.
Logistic regression model
The results of the LR analysis using a purposeful variable selection process are shown in [Table 3]. The LR model was statistically significant, χ2 (4) =15.98, P = 0.003, and had a good fit (P value for Hosmer and Lemeshow test > 0.05). It explained 42% (Nagelkerke R2) of the variance in the occurrence of rLDH and correctly classified 80.3% of cases. We tested different models for their ability to predict the outcome. The model with the best prediction was:
Total points = 0.04*a + 0.9*b – 5.3*c - 1.6*d - 1.3, where a = age in years, b = smoking (0 or 1), c = WLS dural sac circumference/L1 vertebral body circumference (VBC) and d = WLS interfacet distance/L1 VBC and -1.3 is the constant. We performed an ROC analysis with the total score and compared it against the occurrence of rLDH. The model was internally validated on a cohort of 25 patients and was found to have a fair accuracy in predicting rLDH (accuracy of 0.80 and AUC of 0.70). The sensitivity and specificity of this model were 0.83 and 0.62, respectively.
Prediction of rLDH
Given the growing focus on value-based care in health care, risk stratification is being increasingly used to augment the efficacy of informed decision-making and preoperative counseling. This is of great relevance to spine surgeons who have amongst the highest rates of malpractice claims amongst all surgical specialties., It, therefore, becomes all the more important for the spine surgeon to ensure that the patient understands the risks of him or her having a complication or recurrence of the problem. Recurrent herniation is a common complication of lumbar disc surgery that places considerable burden on the health care system, and this is reflected by the number of studies that have tried to address the problem.
With regard to the prediction of rLDH, individual studies that have attempted to identify risk factors for rLDH have demonstrated heterogeneous and often conflicting results.,,,, Smoking, obesity, diabetes mellitus, biomechanical factors, and a “limited” type of discectomy are amongst the many factors that have been cited as risk factors for rLDH. A recent study used multiple radiological variables and proposed a model with seven predictors: BMI, disc height index, hypermobility of the spinal segment, lumbar lordosis, smoking, disc protrusion, and Pfirrmann grade III disc degeneration. This study was, however, limited by possible selection bias and an over-fitted statistical model. Pooled estimates from a systematic review of 17 studies, identified smoking, disc protrusion, and diabetes to be significant predictors of rLDH while factors such as gender, BMI, occupation, level, and side of herniation were found to be irrelevant.,
Morphometrics in spine surgery
Most of the studies related to lumbar disc disease have analyzed variables ranging from demographic and lifestyle indices to radiographic factors such as disc morphology and grade of degeneration,, foraminal and neural compromise, facetal degeneration, and range of motion. Moving aside from these conventional variables, researchers are increasingly utilizing morphometrics to predict outcomes after spine surgery.,,, Keeping this in mind, we have included both, a set of previously identified predictors of rLDH, as well as some spinal morphometric variables that could potentially have a bearing on the occurrence of rLDH.
Many of the morphometric variables used in our study have been previously analyzed separately in studies on lumbar degenerative disease. None of these has however been analyzed with respect to rLDH. Ligamentum flavum area has been previously found to be a good morphological index of canal stenosis. Increasing interfacet distance has been reported to be protective against degenerative changes. Spinal canal and dural sac dimensions have been found to correlate with back pain,, claudication distance, health-related quality of life, and the eventual selection for surgery. Spinal canal and dural sac dimensions have also been able to predict, which patients with back pain eventually require surgery.,
Relevance of spinal canal and dural dimensions in rLDH
Disc degeneration and some cases of spinal stenosis have been recognized to have developmental links., Developmental spinal stenosis has been recognized as a risk factor for revision surgery as these patients are prone to developing symptoms at multiple levels. The impact of this entity on lumbar spine disease remains unclear, and its possible relevance in rLDH has not been evaluated. Akin to the cervical spine, where the unique kinematic traits in developmentally narrow canals have been demonstrated to predispose to accelerated pathological changes, it is conceivable that patients with narrower lumbar dural sacs or canals may be at higher risk of developing symptomatic disc degenerations and rLDH than their counterparts with normal canal morphology.
In our study, other than increasing age and smoking, two of the predictive variables turned out to be WLS morphometric variables. A smaller WLS interfacet distance and dural sac circumference turned out to be strong predictors indicating that patients with smaller canals or dural sacs are predisposed to developing symptomatic rLDH at any given level. This could be attributed to their having more neural compromise as compared to patients with larger canals or dural sacs for a given degree of disc prolapse.
Implications of our study
Smoking was the only directly modifiable risk factor for rLDH that we identified.
Our study suggests that the pathogenesis of rLDH is probably multifactorial, with developmental factors related to canal and dural sac dimensions playing a strong role in determining, which patient eventually becomes symptomatic enough to warrant repeat surgery. Though these morphometric dimensions mainly have developmental connotations, some studies have noted an occupational influence on some of these dimensions. For example, people subjected to chronic stress due to occupational workloads on their spines tend to develop wider interfacet distances. According to our analysis, this morphological alteration could actually protect them from rLDH if they were to ever undergo surgery for a lumbar disc herniation.
Using our fairly accurate and sensitive predictive model and identifying the “at-risk” population with smaller bony canals or dural sacs might make preoperative counseling for patients undergoing lumbar disc surgery more meaningful. This could further also aid in initiating early preventative measures, implementing lifestyle/activity modifications, selecting a more aggressive discectomy (over a “limited” discectomy) technique, and perhaps even lead to the development of novel and precision-based therapeutics in future.
Limitations of the study
The study has the inherent limitations of that of a retrospective analysis. Though a subtotal discectomy technique was followed in all cases, the degree of aggressiveness amongst different surgeons could be a confounding factor while analyzing rLDH. The predictive value for our model holds good for a relatively short 3 year follow-up period after the index surgery. Furthermore, our results will need external validation in larger sample sizes and in different patient populations. Morphometric analysis was performed on the patients' preoperative MRI images alone. It may be of added interest to evaluate changes in dural sac morphometrics after surgery and to assess if these changes influence long term outcomes. It would also be interesting to analyze whether spinal morphometrics influence outcomes in patients operated by minimally invasive techniques, a rapidly evolving trend in the realm of spine surgery.,
Larger mean lumbar bony canals and dural sacs protect from the occurrence of symptomatic rLDH. These WLS morphometric variables should be included in future risk stratification algorithms for lumbar disc disease. In addition to the previously recognized risk factors, our study points to an underlying developmental predisposition for rLDH.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Declaration of patient consent
Informed consent was obtained from all individual participants included in the study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3]