What Is the Probability of Recovery of C5 Palsy After Cervical Spine Surgery Over Time? A Systematic Review and Meta-Analysis.

Abstract

As recovery from brachial neuropathy after C5 or more extensive palsies after cervical spine surgery is generally favorable and may occur even beyond 2 years, early surgical interventions (e.g. nerve transfers) may risk overtreatment. Estimates of recovery probability over time may help guide decisions about the timing and potential benefits of additional surgery. Bayesian analysis is helpful in estimating this probability, as it estimates the probability of recovery and updates it when new information becomes available, such as when nerve recovery does not occur over time. We asked: (1) Over time, up to 2 years, what is the probability of brachial neuropathy recovery after cervical spine surgery? (2) What variables are independently associated with time to nerve recovery? (3) Over time, up to 2 years, what is the probability of recovery accounting for variables independently associated with time to recovery? We performed a meta-analysis of individual patient data following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We systematically searched PubMed, Embase, and Cochrane Library for English articles reporting on recovery in individuals with brachial neuropathy after cervical spine surgery. We excluded studies if they did not report individual participants' Medical Research Council (MRC) scores or time to recovery. The search yielded 4308 unique records. After screening titles and abstracts, 3962 were excluded, leaving 346 articles for full-text review. Fifteen full texts could not be retrieved, and 312 were excluded after full-text screening, resulting in 19 included studies encompassing 77 patients with 86 C5 palsies. Median age was 65 years (interquartile range [IQR] 56-72), 81% were male, and median follow-up was 12 months (IQR 5.9-19). Risk of bias was assessed using the MINORS criteria and was considered relatively low (median score 18, range 9-22). We defined recovery as an MRC score greater than or equal to 3 for biceps and/or deltoid function. Recovery over time was modeled using a Bayesian network based on two conditional probabilities: (1) the probability that the neuropathy is recoverable (neuropraxia, or recoverable axonotmesis) by 2 years, determined by survival analysis (88%, 95% confidence interval [CI] 80-95%); and (2) the probability that the neuropathy did not resolve at each timepoint, derived from the individual patient data on time of recovery. We tested for variables independently associated with time to nerve recovery with multivariable Cox-proportional hazards. If the palsy hasn't resolved by 16 months, the probability of recovery is still likely better than chance, 57% (range 42%-78%). If the palsy hasn't resolved by 20 months, the probability of nerve recovery is 24% (range 14%-45%). Cervical spondylotic radiculo-myelopathy (HR 1.9, 95% CI 1.1-3.4, p=0.016) compared to ossification of the posterior longitudinal ligament, and MRC2 (HR 2.0, 95% CI 1.1-3.6, p=0.023) compared to grade 0-1 were independently associated with an increased recovery. However, patients with an initial greater MRC score, who don't recover the first year, have a lower probability of recovery the second year after palsy than people with an initially lower MRC score. These findings may help reduce physicians and patient anxiety about the uncertainty of recovery, potentially reducing the urge to "do something". Deferring early nerve transfers and instead considering muscle transfers after 1.5 to 2 years if the palsy persists seems to minimize overtreatment. Larger future studies could further refine individualized recovery predictions over time.