Neurosurgery Clinics of North America

A prospective, randomized, double-blinded study is the gold standard for evaluating the treatment of any medical condition or surgical intervention. Few would argue with the value of the data generated by the various investigational device exemption studies assessing the efficacy of lumbar and cervical arthroplasty devices. 1, 2, 3, 4, 5, 6 Nonetheless, the barriers to prospective, randomized, blinded studies limit their applicability to conditions such as spondylolisthesis, which presents with various symptoms and an array of treatment options. These studies are not only cost prohibitive but also have inclusion and exclusion criteria that can be so restrictive that the study population does not accurately reflect clinical practice. Nevertheless, we apply these data to determine treatment strategies for our patients, many of whom would not meet the inclusion criteria for the study most relevant to the care. Thus, the generalizability of a study is paramount to the decisions that surgeons make daily.

As we consider future studies that will lead us to optimize outcomes for the management of lumbar spondylolisthesis, it is important to recognize that the gold standard for scientific analysis in medicine may not be the only means to achieve our goals. Instead, the ideal vehicle to collect and analyze these data will likely be surgical registries. Registry data accurately reflect the current practice of spinal surgery because there are no inclusion or exclusion criteria, only surgeons enrolling patients they see every day in clinical practice. Every patient who undergoes a surgical procedure has the potential to be included and contributes valuable data.

Optimization of outcomes in lumbar spondylolisthesis will arise from expanding the enrollment of patients into registries, increasing the granularity of the data collected, and most importantly, rigorous and thoughtful analysis of that “big data.” In the years to come, barriers to participate in registries will cease to exist, with predictable expansion of registry use and a decrease in cost. Enrollment of patients promises to seamlessly incorporate itself into the practice of spine surgery and become virtually unnoticeable to the clinician. Data will continue to accumulate. Optimization of outcomes through data analysis will be inevitable. Prospective trials will still have a role in the assessment of spinal surgeries, but their cost, restrictiveness, and applicability will limit their use to answering very specific questions.

This article presents our current knowledge of lumbar spondylolisthesis gained from prospective randomized trials and surgical registries. Before an examination of the prospective clinical study and registry data on lumbar spondylolisthesis, a review of the surgical options for the management of spondylolisthesis will provide the groundwork and the context for the direction of future studies.

Even with the most recent prospective randomized trials and registry outcome studies on lumbar spondylolisthesis, we find no universal consensus on the management of spondylolisthesis. 7, 8, 9 The decision to proceed with one intervention over another may be as likely to be based on equal parts science and surgeon preference as it is on clinical symptoms, anatomy, and affected segment. Jeffrey Hammerbacher, professor of genetics and genomic science at the Icahn School of Medicine at Mount Sinai urges that “before you predict the future, you have to see the present.” He also adds, “but seeing the present clearly can be as hard as predicting the future.” 10 To understand where to place the emphasis of future studies, we first must examine the present management of spondylolisthesis

The options for surgical intervention break down into 3 broad categories: (1) decompression alone, (2) decompression with fusion, and (3) decompression with a motion preservation device. From here, the options increase in complexity (Fig. 1). The continued refinement of minimally invasive techniques, both percutaneous and mini-open, alongside the traditional midline approaches, create several additional branch points in each category. The first branch point in this decision tree, decompression versus decompression and fusion, has been an area of rigorous study and debate.

Fig. 1. The surgical management of spondylolisthesis may be distilled into 3 broad categories. Decompression versus decompression and fusion has been the focus of the most recent prospective randomized trials.(Courtesy of Barrow Neurological Institute, Phoenix, AZ; with permission.)

The continued refinement of minimally invasive techniques, both percutaneous and mini-open, alongside the traditional midline approaches create several additional branch points in each category. Fig. 2 shows the options available to patients in the decompression alone category. The distinguishing feature among these 3 groups is the preservation of the spinous process and posterior tension band. Although biomechanical studies have demonstrated greater instability in a segment with incremental removal of the posterior elements, the clinical implications of spinous process and the posterior tension band preservation on clinical outcomes have been difficult to discern. 11 One of the fair criticisms of the various randomized trials is that none of the studies included current minimally invasive techniques. It is difficult to speculate what effect preservation of the posterior tension band would have on the reoperation rate in a decompression cohort, which was as high as 34% in one of the randomized studies. 7

Fig. 2. Decompression alone for management of spondylolisthesis condenses into 3 options, both open and minimally invasive. The main distinguishing feature is the preservation of the posterior tension band and spinous process. (Courtesy of Barrow Neurological Institute, Phoenix, AZ; with permission.)

The greatest number of permutations resides in decompression and fusion. There is ample clinical evidence, ranging from level I to level IV, to support the various interventions listed in the flowchart in Fig. 3, 7, 8, 9, 12, 13, 14, 15, 16 Instrumented posterolateral fusion has the highest level of evidence supported by prospective randomized trials. 7, 8, 9 It is important to note that minimally invasive techniques were not included in these studies. There are 4 different lumbar interbody approaches: posterior lumbar interbody fusion, transforaminal lumbar interbody fusion (TLIF), anterior and oblique lumbar interbody fusion (ALIF), and transpsoas lumbar interbody fusion. The latter 2 options have been reported as stand-alone, or stand-alone or, more commonly, with additional posterior instrumentation, which can be performed through midline open, paramedian mini-open, and percutaneous approaches. 15

The recent trend in direct-to-patient marketing efforts has introduced a new component to the decision-making process. Patients now present requesting certain procedures brought to their attention by industry advertisements. That trend is inherently tied to the increasing options for decompression with motion preservation. Interspinous process devices and facet replacement systems collectively represent the third category of decompression with a motion preservation device (Fig. 4). 12, 13, 14, 17

Fig. 4. Decompression with motion preservation devices include interspinous process devices and facet replacement systems.(Courtesy of Barrow Neurological Institute, Phoenix, AZ; with permission.)

There are currently no less than 19 different surgical options available to a patient with symptomatic single-level spondylolisthesis. Obtaining level I evidence to establish one of these operations or even a group of operations as optimal is untenable. Powering such a trial in a statistically significant manner is daunting. The cost, enrollment, inclusion and exclusion criteria, randomization of patients, and curation of the data collection create an insurmountable barrier to considering such a trial. Nonetheless, the questions of which procedure demonstrates the greatest efficacy for a particular patient remain important. Evidence needs to be gathered to answer those questions, but the vehicle to answer those questions is not a randomized prospective trial. Instead, prospective enrollment of patients undergoing all of these procedures in a well-curated database drawn from several centers will guide the path of further investigation to help answer the question.

In 2000, the NIH funded the largest prospective study examining the efficacy of surgical management versus nonsurgical management in the 3 most common spinal conditions that afflict Americans: lumbar disc herniations, lumbar stenosis, and lumbar spondylolisthesis. Known as the Spine Outcome Research Trial (SPORT), the study enrolled patients from 13 medical centers in 11 states and randomized them to surgery or nonoperative measures. In the spondylolisthesis arm of the study, a total of 607 out of 892 eligible participants enrolled (304 into the randomized trial and 303 in the observational cohort). The randomized cohort placed 159 patients into the surgical arm and 145 patients into the nonoperative arm.

Eight years into the spondylolisthesis trial, 115 of 159 patients in the surgical arm had undergone surgery. The type of surgery these patients underwent was left entirely to the discretion of the surgeon. One of 4 procedures was performed: (1) posterior decompressive laminectomy alone, (2) decompression with posterolateral in situ fusion (uninstrumented), (3) decompression with instrumented posterolateral fusion, and (4) decompression with instrumented interbody fusion. Outcome measures included the Medical Outcomes Study 36-Item Short Form (SF-36), the Oswestry Disability Index (ODI), the Stenosis Bothersomeness Index, the Low Back Pain Bothersomeness Scale, and Leg Pain Bothersomeness.

In the SPORT trial, 94% of patients treated in the surgical arm were treated with a decompression and fusion. An analysis of the fusion technique demonstrated that an instrumented posterolateral fusion (without an interbody cage) had greater than 10° of rotation and 4 mm of translation on flexion-extension radiographs. Interestingly, the interbody fusion group had the lowest incidence of dural tears. The final results at 8 years did not reveal a statistically significant advantage for one fusion method over another, nor was the study designed or powered for such an analysis.8 At the end of the trial, the patients who were treated surgically in the as-treated analysis (ie, those patients treated with surgery regardless of randomization) improved significantly more than those treated nonoperatively at years 6, 7, and 8 on all outcome measures.8

Although this level I evidence seemed to settle the issue regarding the role of surgery in patients with lumbar spondylolisthesis refractory to exhaustive nonoperative measures, it did not answer the question about the best type of surgery, that is, decompression or decompression and fusion. Two additional studies, both randomized and prospective, 1 from Sweden and 1 from the United States, attempted to answer that question but generated 2 very different results

In 2016, Försth and colleagues 9 published a randomized control trial on 247 patients, 135 of whom had spondylolisthesis. The investigators randomly assigned the enrolled patients in a 1:1 ratio to undergo decompression alone or decompression and fusion. Potentially confounding the study was the manner in which spondylolisthesis was handled. The diagnosis was made with static conventional lateral radiographs. The authors emphasize that, in their methods, flexion and extension radiographs were not obtained, and they elaborate in the discussion section the absence of a consensus on instability. The criteria to meet spondylolisthesis was at least 3 mm of translation of one vertebral body over another on a static radiograph. The randomization of the entire group was then stratified to the presence or absence of degenerative spondylolisthesis, placing an almost equal number in the fusion group (67 of 123 patients) and nonfusion group (68 of 124 patients). Similar to the SPORT study, the decompression or decompression and fusion was solely determined by the surgeon. At 5 years, the authors could not identify a statistically significant difference between decompression and fusion versus decompression alone. In addition, the authors observe that fusion was associated with higher costs but not with a greater clinical benefit at 2 years. 9

Published in the same edition of the New England Journal of Medicine was another prospective randomized controlled trial by Ghogawala and colleagues 7 attempting to answer the same question as the Swedish authors. In the Spinal Laminectomy versus Instrumented Pedicle Screw Fixation (SLIP) trial, 66 patients, all of whom had spondylolisthesis, were randomized to either decompression alone or decompression and fusion. The fusion was performed through a traditional midline open exposure with pedicle screw and fixation, decompression, and posterolateral fusion using iliac crest and local autograft. After 4 years of follow-up, the authors were able to identify a clear benefit in the laminectomy and fusion group over the decompression only group. The fusion group had a clinically meaningful improvement in overall physical health-related quality of life over the decompression group. Reoperation rates differed significantly between the 2 groups (14% in the decompression and fusion group and 34% in the decompression only group). 7

The Swedish study and the SLIP trial asked the same question. The prospectively randomized designs allowed both studies to achieve a high level of evidence. The 2 different conclusions do not change the value that each study contributes to the understanding of spondylolisthesis when viewed in the appropriate context. Perhaps more importantly, the different conclusions despite the similarities in the study concept and design offer insight into designing future studies, especially for defining the patient population.

In the Swedish study, Försth and colleagues concluded that fusion only increases the risk to the patient and the cost of managing this condition while not bestowing any meaningful improvement in outcome. In the SLIP trial, Ghogawala and coworkers demonstrated a statistically significantly lower rate of reoperation with fusion and better patient-related outcomes 4 years after surgery. The Swedish authors suggest that the difference in the studies arose because of both a higher dropout rate and a higher reoperation rate in the SLIP study. Försth and colleagues 9 believed that these confounding factors may have negatively affected the results of the SF-36 in the SLIP study, thereby leading to a different conclusion. However, the Swedish study authors make no mention of the differences between the patient populations. Ghogawala and colleagues 18 appropriately highlight those differences, emphasizing that the SLIP study only enrolled patients with spondylolisthesis, whereas the Swedish study had patients with lumbar stenosis and spondylolisthesis.

Although the effects of a heterogeneous population on the results are difficult to determine, it is noteworthy that over 50% of the patients enrolled in the Swedish study would not have met the enrollment criteria for the SLIP trial. Therefore, these 2 studies confirm an intuitive principle: the same question, studied similarly with 2 different populations using 2 different outcome measures, will generate 2 different conclusions. The heterogeneity of the patient population in the Swedish study means that the conclusion should be applied to an equally heterogeneous population. When considering a more homogeneous group, such as single-level spondylolisthesis with less than 3 mm of translation on flexion-extension studies, the SLIP data would be more appropriate. Perhaps the most important lesson learned from the Swedish study and SLIP trial is that the patient population should be defined as precisely and homogenously as possible in future prospective studies.

The prospective randomized SLIP trial and Swedish study attempted to answer a single question: does fusion improve outcomes in the surgical management of spondylolisthesis? However, it was not the intent of either of these studies to answer which type of fusion is preferable and under what circumstances. As would be expected in a randomized prospective trial, the techniques used in both trials were standard open approaches, without interbody spacers and biologics.

It is important to recognize that the techniques used in the Swedish study and SLIP trial do not reflect the current trend toward minimally invasive approaches to spine surgery. Ghogawala and colleagues acknowledge the trend by speculating in the SLIP study, “the less-invasive strategy of unilateral laminotomy with bilateral decompression may offer an advantage over traditional laminectomy because the midline ligamentous structures are preserved.” The same may be said for the fusion techniques. Many surgeons would appropriately question the applicability of these studies to their practice when their surgical approaches vary significantly from the techniques applied in these 2 studies. The logical request would be for a prospective randomized trial to assess surgical management versus nonsurgical management of each of the 19 different techniques.

However, if 2 studies similar in everything but patient population and outcome measures created dissonance, it would be impracticable to study the 19 different surgical options available to patients with spondylolisthesis in 2019 to an equivalent level of evidence. Another less burdensome method with a high enough level of evidence to guide the optimization of surgical outcomes is needed.

The recent rise of registries, such as the Neuropoint Alliance Quality Outcome Database (QOD) registry, presents an opportunity to answer questions in an unconstrained prospective manner while applying a different approach than a prospective randomized trial. Instead of defining a patient population shaped by inclusion and exclusion criteria and then applying a uniform method of intervention to analyze 2 groups, a registry captures as much data as possible. Surgeons participating in the QOD are in the continuous mode of enrolling patients and collecting demographic, radiographic, surgical procedure, and outcome data. Such an endeavor builds an ever-expanding data set. That data set is a better representation of the population surgeons treat in their respective practices and more accurately reflects the surgical techniques used to treat those patients. Conclusions drawn from an analysis of 66 patients (the sample size of the SLIP study) are difficult to apply to a heterogeneous population of approximately 40,000 patients in the United States who undergo lumbar fusion surgery for degenerative lumbar spondylolisthesis.19 It is even more tenuous to apply conclusions reached from an analysis of 247 Swedish patients, a culturally and racially homogeneous population, to the diversity of patients in the United States.

Registries have less-constraining inclusion criteria than clinical studies. Novel technologies can be immediately captured and analyzed long before a question arises worthy of a prospective randomized trial. Regional differences may be identified and incorporated into the analysis; several interventions may be included and studied. Although registry data may not be considered level I evidence, clear trends surfacing from an analysis of thousands of patients would be difficult to dismiss. The lesson learned from the SLIP trial and the Swedish study may be immediately applied in the analysis of data in a registry. The study patient population may be defined as narrowly and homogeneously as the granular data collected will allow. From such a data set, a very narrow and homogeneous population can be defined, and a focused question may be asked creating an environment for rigorous analysis and meaningful conclusions.

To date, QOD analyses have answered questions regarding the cost-effectiveness of one procedure over another and the minimum follow-up required for outcome measures to accurately determine a clinical benefit. The QOD analyses have unveiled trends in patient satisfaction that correlate with diagnosis and have identified differences in minimally invasive versus open surgical techniques. The following section highlights some of the recent analyses of the QOD that remain outside the capacity or intent of randomized trials. The observations made from each of these studies have implications on the future management of spondylolisthesis.

Cost-efficacy will increase in importance over the next decade. As the cost of health care in the United States remains on a trajectory to reach one-fifth of the gross domestic product, the value that surgical procedures bring to health care will be measured by a ratio of quality over cost. Both the SLIP trial and Swedish study touch on this topic, but by the very circumstances of those studies, the studies did not include the most recent techniques in current practice. Length of stay has decreased along with the cost of the procedure. It would be difficult to perform a post hoc analysis of a randomized trial attempting to apply scenarios of new technology and measure the impact. Registry data are not constrained in this manner. For instance, the use of transforaminal approaches to achieve interbody fusions is becoming increasingly common in clinical practice, but that technique is not included in either prospective study (the SLIP trial and Swedish study used instrumented posterior lumbar fusion [PLF]), and it represents the minority of the surgical arm of the SPORT study. However, there is an additional cost to the addition of an interbody. A reasonable question is what effect it has on the cost of the procedure and whether that additional cost demonstrates value.

By analyzing the patients enrolled in the QOD, Carreon and colleagues 20 identified a sizable patient population and reported a cost analysis of TLIF versus PLF for the management of spondylolisthesis. These authors demonstrated that operative time, length of stay, complication rates, discharge disposition, and readmission rates were similar, but the quality-adjusted life years gained with TLIF were greater than with PLF. These conclusions have implications in payer policies and coverage determinations for patients. In another analysis of the QOD, McGirt and colleagues21 were able to examine minimally invasive techniques versus open techniques, yet another limitation of all prospective studies. The QOD allowed for the analysis of almost 2000 patients in 27 states who underwent lumbar fusion for the management of lumbar spondylolisthesis or lumbar stenosis. There were 467 patients in the minimally invasive cohort and 1480 in the open cohort. McGirt and colleagues identified reduced intraoperative blood loss, reduced length of stays, but no significant difference at 12 months postoperatively. Chan and colleagues22 analyzed QOD data and identified a clear difference in outcomes when stratifying the patient population undergoing fusion for spondylolisthesis by body mass index(BMI). In their analysis of 382 patients with a BMI over 30 and 415 patients with a BMI under 30, these investigators found that, in multivariate analyses, a high BMI (obesity) was independently associated with worse Numeric Rating Scale leg pain and EuroQoL-5D at 12 months. The most important aspect of these analyses is that the conclusions demonstrate the clinical value of interventions commonly performed in clinical practice but not adequately represented in the randomized controlled studies.

The importance of accurately determining the clinical benefit of a procedure is immeasurable in the value proposition of the health care of tomorrow. Commercial and government payers will evaluate and stop paying for interventions that do not demonstrate a clinical benefit. Outcome measures such as the ODI, SF-36, and visual analog scale for pain collected over time should reveal patient benefits; nonetheless, at what point is the clinical benefit realized? Examining the trends in outcome measures in the QOD allowed authors to define the time for the collection of measures such as ODI to meaningfully evaluate the effectiveness of a lumbar intervention. Asher and colleagues23 concluded that prospective longitudinal registries need a minimum of 12 months of outcome measures to determine the effectiveness of a procedure. Measures collected for less than 12 months will not accurately reveal the clinical benefit of an intervention. Defining the duration of follow-up achieved through examination of big data further bolsters prospective and retrospective study methods. Future analyses of registry data will contribute to our understanding of the value of procedures.

Big data may be defined as the study and application of expanding data sets that are too large and complex for traditional data-processing methods. The power of the data collected will only be as good as the granularity measured. The value of the conclusions will only be as good as the rigor of the analysis of those data. The ability to identify trends that surface with particular interventions or patient populations will drive further inquiry. It may be the analysis of registry data that reveals the next question that would need to be answered by a prospective randomized trial. Analysis of the results of a prospective trial will lead to further refinement of the registry. The process will be cyclical. The result will be a larger body of evidence supporting the conclusions reached. The SLIP II study is a perfect example of a prospective trial leading to a registry. Recognizing the shortcomings of both the Swedish study and the original SLIP trial, the authors of the SLIP trial created a nonrandomized registry that will enroll 1000 patients with grade I spondylolisthesis and monitor the patients for 5 years. A formal cost-effectiveness analysis has also been added to the study.24

In Steve Lohr’s book DATA-ISM: The Revolution Transforming Decision Making, Consumer Behavior and Almost Everything Else, the author begins his examination of the impact the large databases have in our world with the following thoughtful, almost philosophic statement: “The promise of big data is the story, is far richer and more detailed than ever before, making it suddenly possible to see more and learn faster… to make the invisible visible.” Paraphrasing Lohr’s sentence, the promise of registries in spine surgery is that the data collected today will reveal obvious directions for intervention or investigations in the future.

It is conceivable that, in the next decade, patients who present with lumbar spondylolisthesis may have treatment plans guided, if not determined, almost entirely by the big data that have been collected in this decade. The hundreds of thousands of data points that our current generation of surgeons gather as part of their participation in the various surgical registries will begin to unveil clear directions for particular patients with a particular diagnosis, which are not seen in prospective randomized trials or appreciated in clinical practice by groups or individuals. Those directions are not as obvious today as they will be after collection and processing of the ever-growing data of tomorrow. Optimization of outcomes will be the inevitable upshot of the magnitude of the data set and its analyses. The more granular the data in the data set, the more potential it will have to “make the invisible visible.”

The promise of data analytics in the management of lumbar spondylolisthesis is to identify the best option for an individual on the basis of sex, duration of symptoms, vertebra level, BMI, and pelvic incidence-lumbar lordosis mismatch that offers the greatest amount of relief and mitigates the risk of adjacent segment degeneration or any subsequent surgery. What do we emphasize in future studies so that we optimize the management and outcomes of lumbar spondylolisthesis? The answer is striking a balance between registry data and prospective trials.

Big data are not without challenges and potential pitfalls. Each shortcoming must be acknowledged and routinely addressed to maintain the value of the registry. Lohr lists these challenges as data capture, storage, analysis, sharing, transfer, querying, updating, and privacy.10 The data breaches that social media and internet search engine companies have experienced demonstrate the vulnerability of big data, especially when there is a financial interest at stake. As the data from spine registries become increasingly powerful over time, monetary worth may transform it into a potential target. Another potential pitfall is the veracity of the data. The importance of maintaining a high level of veracity of the data is inherently tied to its value. As the barriers of entry to registries decrease, maintaining the veracity of data will be increasingly important. A low level of data veracity contributes to “noise” within the data. As the noise increases, the potential impact of any conclusion decreases.

In the final analysis of all the randomized trials and all the registry data, the only definitive conclusion that we can reach is that no hypothesis regarding intervention on the spine can ever be proven or disproven by a single study or even a combination of studies. Such hypotheses will be proven through studies that are refined again and again to establish the role of one procedure over another. Randomized trials and registries will be a component of identifying the patient, the timing, and the type of intervention from the menu of surgical options. A balance between prospective registries and randomized trials will guide the future studies that will demonstrate the optimal treatments and outcomes for lumbar spondylolisthesis in the patients of tomorrow.