Friday, September 23, 2016
The degree of improvement following surgery for cervical spondylotic myelopathy (CSM) can vary substantially, with most patients experiencing partial recovery of function following surgery. Many studies have evaluated surgical outcome predictors, and a recent meta-analysis concluded that a longer duration of symptoms and more severe baseline symptoms predicted worse post-operative function.1 Diabetes has been evaluated as a risk factor for poor outcomes following surgery for CSM, but it has generally not had a strong association with outcomes.2 Given the unclear association between diabetes and CSM outcomes, Kusin et al. retrospectively analyzed the improvement in Nurick scores in 113 CSM patients treated surgically, 33 of whom had diabetes. Additionally, they stratified the diabetic patients based on their average perioperative glucose levels. They found that diabetic and non-diabetic patients had similar baseline Nurick scores of approximately 2.5, a score indicating the presence of a gait abnormality but no need for an assistive device in the average patient in the study. The diabetic patients improved significantly less on the Nurick score (1.1 vs. 1.6 points), and there was a strong inverse relationship between perioperative glucose level and Nurick score improvement. Stratifying the diabetes patients into groups with average perioperative glucose levels of above and below 150 mg/dL revealed that patients with poor perioperative glucose control had minimal improvement (0.5 points), while those with good control improved to a similar degree as non-diabetics (1.9 points). Multivariate analysis revealed that diabetics and smokers had less improvement, while worse baseline symptoms and a greater number of levels fused predicted greater improvement.
This study adds some important information to the CSM literature, namely that the level of glucose control in diabetics may affect outcomes. Prior studies that classified diabetes as a dichotomous variable may have missed an association between diabetes and outcomes if many of the patients had well-controlled diabetes and responded like non-diabetics. While the level of glucose control may affect CSM outcomes, the limitations of this study need to be considered before drawing strong conclusions. The authors used average perioperative glucose level as a surrogate for diabetes control, and hemoglobin A1c would have provided a much better picture of glucose control over a longer period of time. Additionally, the only outcome was the Nurick score, a relatively blunt instrument that may not detect subtle improvements in function. The JOA score is a more responsive item that would have been helpful to include. The study also included a relatively small number of diabetic patients and included the experience of a single surgeon. Its retrospective nature also limits the strength of the conclusions that can be drawn. Despite these limitations, the paper certainly raises the possibility that poorly controlled diabetics may have less improvement following surgery for CSM. Future studies will need to determine if better perioperative glucose control leads to better outcomes. A key question in CSM research at this point is determining if there are subgroups that do not improve more with surgery than with non-operative care. Given that CSM has a generally poor prognosis if treated without surgery, there is little data available on non-operative outcomes. However, there may be subgroups—i.e. poorly controlled diabetics, patients with a very long duration of symptoms, and patients with mild myelopathy—who don't benefit much from surgery. A long-term observational study that follows patients treated both with surgery and non-operative care would be required to answer these questions, and it is not clear if such a study will ever be performed.
Please read this article in the September 15 issue. Does this change how you consider treating diabetics with CSM? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Tetreault LA, Karpova A, Fehlings MG. Predictors of outcome in patients with degenerative cervical spondylotic myelopathy undergoing surgical treatment: results of a systematic review. Eur Spine J 2015;24 Suppl 2:236-51.
2. Tetreault LA, Kopjar B, Vaccaro A, et al. A clinical prediction model to determine outcomes in patients with cervical spondylotic myelopathy undergoing surgical treatment: data from the prospective, multi-center AOSpine North America study. J Bone Joint Surg Am 2013;95:1659-66.
Friday, September 16, 2016
Bone morphogenetic protein-2 (BMP-2) generally increases fusion rates, though it is unclear which patients benefit from its use, and it has been linked to adverse events. Certain patient characteristics and surgical procedures are known to be associated with pseudarthrosis (i.e. smoking, prior non-union, long fusions to the pelvis), and these cases may benefit most from BMP-2 use. However, the FDA IDE trials evaluating BMP-2 studied primary one level lumbar fusions, which are known to have a relatively high fusion rate and reasonably good patient outcomes using autograft. As such, the benefits of BMP-2 in terms of fusion rates and patient reported outcomes were modest at best, and it is difficult to justify the cost and potentially associated adverse events if BMP-2 offers minimal advantage.1,2 Amber Laurie and her colleagues from Oregon obtained access to the Yale Open Data Access (YODA) project database including individual level data on over 1,200 patients enrolled in the FDA IDE trials and performed a meta-analysis with the goal of identifying patient subgroups who benefited more from BMP-2 use. They analyzed age, sex, body mass index (BMI), prior surgery, diabetes, work status, and smoking as covariates to determine if BMP-2 had a differential effect for subgroups defined by these characteristics. The primary outcomes were "overall success" (defined as a solid radiographic fusion, improvement on the ODI more than 15 points, no decline in neurologic function, and no "failure" requiring re-operation) and radiographic fusion at 2 years. Patient reported outcomes (i.e. ODI, SF-36) were analyzed as secondary outcomes. The most striking benefit was for smokers, who had a fusion rate of 91% at 2 years with BMP-2 compared to 71% with iliac crest bone graft (ICBG). For non-smokers, the fusion rates were 91% with BMP-2 and 88% for ICBG (no significant difference). BMP-2 also seemed to have a greater advantage for patients with a BMI of less than 25 (normal weight) compared to those with higher BMI. At 2 years for normal weight patients, the overall success rate was 64% with BMP-2 compared to 46% for ICBG. For the other weight categories, no significant differences in overall success were observed between the BMP-2 and ICBG groups. This seemed to be driven by a lower fusion rate at 2 years for the normal weight group treated with ICBG (78%) compared to the heavier groups (85-91% fusion rates with ICBG). There were no differences in patient reported outcomes across the subgroups stratified by graft type, though, overall, the BMP-2 patients had slightly better ODI and SF-36 outcomes than the ICBG patients.
The authors should be applauded for helping to define which fusion patients might benefit most from BMP-2 use. While some authors have painted a black and white picture about BMP-2 (it's either good or bad), in reality, BMP-2 is good for some patients and not helpful (and maybe harmful) for others. Unfortunately, there is not much high quality scientific data available to define these subgroups. The current study does suggest that smokers will have a higher fusion rate for a one-level fusion when BMP-2 is used. Interestingly, this did not translate to markedly better patient-reported outcomes for this group. The interaction between graft type and BMI is interesting, and it is unclear why the normal weight patients had a lower fusion rate with ICBG compared to the heavier patients. It is possible that some underweight or low normal weight patients had metabolic derangements contributing to lower fusion rates and benefited from BMP-2. It would have been interesting to see a similar analysis based on bone density, but these data were not available. The main limitation of this study is the mixing of patients treated with very different fusion techniques (i.e. ALIF, posterolateral instrumented fusion, and PLIF), and it is possible that the findings would have been different if the analysis had been stratified based on fusion technique. Unfortunately, such an analysis would have probably been underpowered. Based on these data, it seems as though there's little justification for using BMP-2 in one level fusions for non-smokers. For smokers, its use will likely improve the fusion rate, though it is unclear how much this will improve patient reported outcomes. For now, surgeons will have to use their judgment to guess which patients will benefit from BMP-2. Hopefully future studies will help define this population better.
Please read Ms. Laurie's study in the September 15 issue. Does this change in whom you consider using BMP-2? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Fu R, Selph S, McDonagh M, et al. Effectiveness and harms of recombinant human bone morphogenetic protein-2 in spine fusion: a systematic review and meta-analysis. Annals of internal medicine 2013;158:890-902.
2. Simmonds MC, Brown JV, Heirs MK, et al. Safety and effectiveness of recombinant human bone morphogenetic protein-2 for spinal fusion: a meta-analysis of individual-participant data. Annals of internal medicine 2013;158:877-89.
Friday, September 9, 2016
The spine literature is replete with papers demonstrating the poor correlation between imaging findings and symptoms. This concept goes back to Boden's seminal paper in 1990 showing a high rate of MRI "abnormalities" in asymptomatic patients.1 While there is some conflicting data regarding the relationship between lumbar stenosis severity and symptom severity, the literature has generally shown a poor or non-existent correlation between the two. The current study by Dr. Burgstaller and collegaues from Switzerland evaluated the relationship between stenosis severity and symptom severity in 150 lumbar stenosis patients, all of whom had neurogenic claudication and cross-sectional imaging demonstrating stenosis. They evaluated 23 radiographic parameters, including measures of central, lateral recess, and foraminal stenosis along with Pfirrman disk degeneration and Modic changes. They then evaluated the correlation between these parameters and back and leg pain scores. Similar to prior studies on this topic, they found no significant correlation between any of the radiographic findings and patient reported pain scores.
This paper adds to the already substantial literature demonstrating the weak association between imaging findings and symptoms in patients with lumbar stenosis. One of the current challenges in diagnosing and treating stenosis appropriately is our poor understanding of the etiology of claudication symptoms. It is unclear if claudication symptoms result from a vascular cause, mechanical compression of the nerve roots, or some combination thereof. If the nerve symptoms result from a vascular etiology, is it due to a lack of arterial inflow or venous congestion? How does the chronicity of the development of stenosis affect symptoms? How does single vs. multilevel disease affect outcomes? Until we understand and can measure these factors, cross sectional imaging combined with a good history and physical exam will likely remain the key components in making the diagnosis. One limitation of this and many studies on the topic is the inclusion of only patients with symptomatic stenosis. It seems likely that once a threshold of symptomatic stenosis is crossed, further stenosis may not cause worsening symptoms. Additionally, that threshold is clearly different in different patients. A study with a true cross-sectional design including symptomatic and asymptomatic patients would likely show some relationship between stenosis severity and claudication symptoms. However, such a study would probably not be that helpful given that the degree of stenosis required to cause symptoms is so variable across patients. The authors are correct in suggesting that stenosis severity is an unreliable metric by which to define the condition of lumbar spinal stenosis. Unfortunately, a better diagnostic test will probably not be forthcoming until we better understand the etiology of the symptoms.
Please read Dr. Burgstaller's paper on this topic in the September 1 issue. Does this change your view of how we diagnose spinal stenosis? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Boden SD, Davis DO, Dina TS, Patronas NJ, Wiesel SW. Abnormal magnetic-resonance scans of the lumbar spine in asymptomatic subjects. A prospective investigation. J Bone Joint Surg Am 1990;72:403-8.
Friday, September 2, 2016
There are many known risk factors for complications and readmission following spine surgery, including medical comorbidities, obesity, smoking, large magnitude surgery, and psychosocial problems. Malnutrition has been implicated as a potentially modifiable risk factor given that low protein stores have been linked to immunosuppression, wound healing problems, and infection. Dr. Odogwa and his colleagues wanted to better understand the association between malnutrition and complications following spine surgery, so they retrospectively reviewed a database including 145 spine surgery patients who had baseline albumin levels in their records. They divided the patients into nourished (albumin > 3.5 g/dL, n= 105) and malnourished (albumin < 3.5, n=40) groups and looked at 30 day readmission as their primary outcome measure. Other complications were also analyzed as secondary outcomes. The malnourished group had a somewhat higher proportion of men compared to the malnourished group (53% vs. 42%) and also underwent fusion of more levels (median of 4 levels vs. 2 levels). They reported an overall 30 day readmission rate of 14%, with the malnourished group readmitted a rate of 28% compared to 10% for the nourished group (p=0.03). Medical complications such as pneumonia, urinary tract infection, and myocardial infarction were also more common in the malnourished group. Given the potential for confounding, the authors performed a multivariate analysis that demonstrated that malnutrion, number of levels fused, and length of surgery were independent predictors of 30 day readmission.
This is an interesting study as it suggests that malnutrition is a stronger predictor of readmission than age, body mass index, diabetes, and smoking. Unlike most other risk factors for complications, protein malnutrition is relatively easy to modify, and doing so could lead to a lower complication rate. Prior to making any strong conclusions, the inherent limitations of a retrospective, observational study need to be considered. The biggest threat to the validity of this study is the potential for confounding by both measured and unmeasured variables. The malnourished patients underwent surgeries of significantly greater magnitude than the nourished group, and number of levels fused was also an independent risk factor for readmission. While there were no patient reported outcomes, given the greater magnitude of surgery required for the malnourished patients, one can infer that these patients were more disabled and had more severe spinal disease, both of which could lead to increased readmission rates. One of the main limitations of the study is the poor characterization of the type of underlying spinal pathology being treated and details on the surgical techniques employed. It is unclear if all patients underwent fusion and how many had cervical vs. lumbar surgery, factors that would clearly influence readmission rates and complications. The number of patients included in the multivariate analysis is relatively low, and this can limit the power of the analysis to evaluate different risk factors. Despite these limitations, this and other studies have identified malnutrition as a risk factor for readmission and complications. Whether it is the cause of these outcomes or just a marker for other causative agents remains to be seen, so it is unknown if correcting hypoalbuminemia pre-operatively would improve outcomes. Fortunately, it is relatively easy to correct protein malnutrition, and doing so is essentially risk free. Future studies should prospectively study the effect of correcting malnutrition on complications and readmission rates.
Please read Dr. Odogwa's article in the September 1 issue. Will this article motivate you to check pre-operative nutrition labs and prescribe dietary supplementation when indicated? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
Friday, August 26, 2016
The use of bone morphogenetic protein (BMP) in spine surgery remains controversial, and one of the most hotly debated topics is its association with cancer. The potential link between BMP and cancer was initially identified by Carragee et al. in 2011, and this group published another paper showing a stronger association with a high dose application of BMP-2 (AMPLIFY) in 2013.1,2 These studies were criticized for the low numbers involved and the heterogeneous cancer types reported. For example, of the 20 cancer events in the AMPLIFY group, 5 were skin cancers in a single patient and 3 more were skin cancers in another patient. Further independent analyses of the Medtronic BMP-2 data reported an increased risk of cancer, though the authors of these studies concluded that the numbers were small, the surveillance was uncertain, and the evidence supporting the link with cancer was relatively weak.3,4 With the Medtronic data having raised the possibility of an association between BMP-2 and cancer, multiple investigators performed large database analyses in order to have sufficient power to more definitively answer the question. Cooper et al. analyzed the Medicare database and found no relationship between BMP and cancer, while Lad et al. performed a similar study using the MarketScan database with the same findings.5,6 On this background, Dr. Dettori and colleagues analyzed administrative databases in Washington State that captured data on hospital discharges, cancer diagnoses, and death. They identified over 4,000 spinal fusion patients treated with BMP from 2002-2010, and matched them in a 1:3 fashion to over 12,000 non-BMP fusion patients based on age, sex, and year of treatment. The average available follow-up in the database was 4.7 years. They found an unadjusted hazard ratio of 1.07 for the risk of cancer in the BMP patients compared to the non-BMP patients, with a hazard ratio of 0.93 after adjustment for the site of surgery (i.e. cervical vs. lumbar). Neither of these were statistically significant. There were no significant differences in the rates of cancer deaths between the two groups.
This is now the third administrative database study to show no association between BMP and cancer. While the Medtronic dataset raised the possibility of a link between BMP and cancer, these further studies using large numbers of patients have called this into question. The limitations of large database studies always need to be considered when interpreting the results of these analyses. All databases are only as good as the data that is inputted, and errors in capturing BMP use and cancer likely exist. However, errors in capturing cancer diagnoses probably affect both groups equally, so this is unlikely to introduce bias. Additionally, these databases only include discharges, cancer diagnoses, and deaths within the state of Washington, so patients who migrate out of state would be lost to follow-up. The authors addressed this by performing an analysis of only patients who had renewed their license within 5 years of surgery (indicating they remained in-state residents for at least a portion of the follow-up) and found the same results. Similar to the issue of misclassification in the databases, migration would also likely affect both groups equally. While the indications for BMP use in spinal surgery and its cost-effectiveness remain debatable, the evidence indicating an association between BMP and cancer continues to weaken.
Please read Dr. Dettori's article in the August 15 issue. Does this change how you view the association between BMP and cancer? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Carragee EJ, Chu G, Rohatgi R, et al. Cancer risk after use of recombinant bone morphogenetic protein-2 for spinal arthrodesis. The Journal of bone and joint surgery American volume 2013;95:1537-45.
2. Carragee EJ, Hurwitz EL, Weiner BK. A critical review of recombinant human bone morphogenetic protein-2 trials in spinal surgery: emerging safety concerns and lessons learned. The spine journal : official journal of the North American Spine Society 2011;11:471-91.
3. Fu R, Selph S, McDonagh M, et al. Effectiveness and harms of recombinant human bone morphogenetic protein-2 in spine fusion: a systematic review and meta-analysis. Annals of internal medicine 2013;158:890-902.
4. Simmonds MC, Brown JV, Heirs MK, et al. Safety and effectiveness of recombinant human bone morphogenetic protein-2 for spinal fusion: a meta-analysis of individual-participant data. Annals of internal medicine 2013;158:877-89.
5. Cooper GS, Kou TD. Risk of cancer after lumbar fusion surgery with recombinant human bone morphogenic protein-2 (rh-BMP-2). Spine (Phila Pa 1976) 2013;38:1862-8.
6. Lad SP, Bagley JH, Karikari IO, et al. Cancer after spinal fusion: the role of bone morphogenetic protein. Neurosurgery 2013;73:440-9.