An Appraisal of the Quality of Development and Reporting of Predictive Models in Neurosurgery: A Systematic Review

BACKGROUND AND OBJECTIVES

Significant evidence has indicated that the reporting quality of novel predictive models is poor because of confounding by small data sets, inappropriate statistical analyses, and a lack of validation and reproducibility. The Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) statement was developed to increase the generalizability of predictive models. This study evaluated the quality of predictive models reported in neurosurgical literature through their compliance with the TRIPOD guidelines.

METHODS

Articles reporting prediction models published in the top 5 neurosurgery journals by SCImago Journal Rank-2 (Neurosurgery, Journal of Neurosurgery, Journal of Neurosurgery: Spine, Journal of NeuroInterventional Surgery, and Journal of Neurology, Neurosurgery, and Psychiatry) between January 1st, 2018, and January 1st, 2023, were identified through a PubMed search strategy that combined terms related to machine learning and prediction modeling. These original research articles were analyzed against the TRIPOD criteria.

RESULTS

A total of 110 articles were assessed with the TRIPOD checklist. The median compliance was 57.4% (IQR: 50.0%-66.7%). Models using machine learning-based models exhibited lower compliance on average compared with conventional learning-based models (57.1%, 50.0%-66.7% vs 68.1%, 50.2%-68.1%, P = .472). Among the TRIPOD criteria, the lowest compliance was observed in blinding the assessment of predictors and outcomes (n = 7, 12.7% and n = 10, 16.9%, respectively), including an informative title (n = 17, 15.6%) and reporting model performance measures such as confidence intervals (n = 27, 24.8%). Few studies provided sufficient information to allow for the external validation of results (n = 26, 25.7%).

CONCLUSION

Published predictive models in neurosurgery commonly fall short of meeting the established guidelines laid out by TRIPOD for optimal development, validation, and reporting. This lack of compliance may represent the minor extent to which these models have been subjected to external validation or adopted into routine clinical practice in neurosurgery.

Pedicled omental flaps for complex wound reconstruction following surgery for primary spine tumors of the mobile spine and sacrum

OBJECTIVE

Surgery for primary tumors of the mobile spine and sacrum often requires complex reconstruction techniques to cover soft-tissue defects and to treat wound and CSF-related complications. The anatomical, vascular, and immunoregulatory characteristics of the omentum make it an excellent local substrate for the management of radiation soft-tissue injury, infection, and extensive wound defects. This study describes the authors' experience in complex wound reconstruction using pedicled omental flaps to cover defects in surgery for mobile spine and sacral primary tumors.

METHODS

A retrospective cohort analysis was conducted on 34 patients who underwent pedicled omental flap reconstruction after en bloc resection of primary sacral and mobile spine tumors between 2010 and 2020. The study focused on assessing the indications for omental flap usage, including soft-tissue coverage, protection against postoperative radiation therapy, infection management, vascular supply for bone grafts, and dural defect and CSF leak repair. Patient demographic characteristics, tumor characteristics, surgical outcomes, and follow-up data were analyzed to determine the procedure's efficacy and complication rates.

RESULTS

From 2010 to 2020, 34 patients underwent pedicled omental flap reconstruction after en bloc resection of sacral (24 of 34 [71%]) and mobile spine (10 of 34 [29%]) primary tumors, mostly chordomas. The patient cohort included 21 men and 13 women with a median (range) age of 60 (32-89) years. The most common indication for omental flap was soft-tissue coverage (20 of 34 [59%]). Other indications included protecting abdominopelvic organs for postoperative radiation therapy (6 of 34 [18%]), treating infections (5 of 34 [15%]), providing vascular supply for free fibular bone graft (1 of 34 [3%]), and repairing large dural defects and CSF leak (2 of 34 [6%]). The median (range) follow-up was 24 (0-132) months, during which 71% (24 of 34) of patients did not require additional surgery for wound-related complications. At last follow-up, 59% (20 of 34) had stable disease and 32% (11 of 34) had recurrence, had progression of disease, or had been discharged to hospice after treatment.

CONCLUSIONS

The pedicled omentum is an effective local tissue graft that can be used for complex wound reconstruction and management of high-risk closures in primary spine tumors. This technique may have a lower rate of complications than other approaches and may influence surgical planning and flap selection in challenging cases.

Utilization of Machine Learning to Model Important Features of 30-day Readmissions following Surgery for Metastatic Spinal Column Tumors: The Influence of Frailty

STUDY DESIGN

Retrospective cohort study.

OBJECTIVE

The aim of this study was to determine the relative importance and predicative power of the Hospital Frailty Risk Score (HFRS) on unplanned 30-day readmission after surgical intervention for metastatic spinal column tumors.

METHODS

All adult patients undergoing surgery for metastatic spinal column tumor were identified in the Nationwide Readmission Database from the years 2016 to 2018. Patients were categorized into 3 cohorts based on the criteria of the HFRS: Low(<5), Intermediate(5-14.9), and High(≥ 15). Random Forest (RF) classification was used to construct predictive models for 30-day patient readmission. Model performance was examined using the area under the receiver operating curve (AUC), and the Mean Decrease Gini (MDG) metric was used to quantify and rank features by relative importance.

RESULTS

There were 4346 patients included. The proportion of patients who required any readmission were higher among the Intermediate and High frailty cohorts when compared to the Low frailty cohort (Low:33.9% vs. Intermediate:39.3% vs. High:39.2%, P < .001). An RF classifier was trained to predict 30-day readmission on all features (AUC = .60) and architecturally equivalent model trained using only ten features with highest MDG (AUC = .59). Both models found frailty to have the highest importance in predicting risk of readmission. On multivariate regression analysis, Intermediate frailty [OR:1.32, CI(1.06,1.64), P = .012] was found to be an independent predictor of unplanned 30-day readmission.

CONCLUSION

Our study utilizes machine learning approaches and predictive modeling to identify frailty as a significant risk-factor that contributes to unplanned 30-day readmission after spine surgery for metastatic spinal column metastases.

Unraveling molecular advancements in chordoma tumorigenesis and treatment response: a review of scientific discoveries and clinical implications

Chordomas are tumors thought to originate from notochordal remnants that occur in midline structures from the cloves of the skull base to the sacrum. In adults, the most common location is the sacrum, followed by the clivus and then mobile spine, while in children a clival origin is most common. Most chordomas are slow growing. Clinical presentation of chordomas tend to occur late, with local invasion and large size often complicating surgical intervention. Radiation therapy with protons has been proven to be an effective adjuvant therapy. Unfortunately, few adjuvant systemic treatments have demonstrated significant effectiveness, and chordomas tend to recur despite intensive multimodal care. However, insight into the molecular underpinnings of chordomas may guide novel therapeutic approaches including selection for immune and molecular therapies, individualized prognostication of outcomes, and real-time noninvasive assessment of disease burden and evolution. At the genomic level, elevated levels of brachyury stemming from duplications and mutations resulting in altered transcriptional regulation may introduce druggable targets for new surgical adjuncts. Transcriptome and epigenome profiling have revealed promoter- and enhancer-dependent mechanisms of protein regulation, which may influence therapeutic response and long-term disease history. Continued scientific and clinical advancements may offer further opportunities for treatment of chordomas. Single-cell transcriptome profiling has further provided insight into the heterogeneous molecular pathways contributing to chordoma propagation. New technologies such as spatial transcriptomics and emerging biochemical analytes such as cell-free DNA have further augmented the surgeon-clinician's armamentarium by facilitating detailed characterization of intra- and intertumoral biology while also demonstrating promise for point-of-care tumor quantitation and assessment. Recent and ongoing clinical trials highlight accelerating interest to translate laboratory breakthroughs in chordoma biology and immunology into clinical care. In this review, the authors dissect the landmark studies exploring the molecular pathogenesis of chordoma. Incorporating this into an outline of ongoing clinical trials and discussion of emerging technologies, the authors aimed to summarize recent advancements in understanding chordoma pathogenesis and how neurosurgical care of chordomas may be augmented by improvements in adjunctive treatments.

Bridging the literacy gap for surgical consents: an AI-human expert collaborative approach

Despite the importance of informed consent in healthcare, the readability and specificity of consent forms often impede patients' comprehension. This study investigates the use of GPT-4 to simplify surgical consent forms and introduces an AI-human expert collaborative approach to validate content appropriateness. Consent forms from multiple institutions were assessed for readability and simplified using GPT-4, with pre- and post-simplification readability metrics compared using nonparametric tests. Independent reviews by medical authors and a malpractice defense attorney were conducted. Finally, GPT-4's potential for generating de novo procedure-specific consent forms was assessed, with forms evaluated using a validated 8-item rubric and expert subspecialty surgeon review. Analysis of 15 academic medical centers' consent forms revealed significant reductions in average reading time, word rarity, and passive sentence frequency (all P < 0.05) following GPT-4-faciliated simplification. Readability improved from an average college freshman to an 8th-grade level (P = 0.004), matching the average American's reading level. Medical and legal sufficiency consistency was confirmed. GPT-4 generated procedure-specific consent forms for five varied surgical procedures at an average 6th-grade reading level. These forms received perfect scores on a standardized consent form rubric and withstood scrutiny upon expert subspeciality surgeon review. This study demonstrates the first AI-human expert collaboration to enhance surgical consent forms, significantly improving readability without sacrificing clinical detail. Our framework could be extended to other patient communication materials, emphasizing clear communication and mitigating disparities related to health literacy barriers.

Machine learning-based detection of sarcopenic obesity and association with adverse outcomes in patients undergoing surgical treatment for spinal metastases

OBJECTIVE

The distributions and proportions of lean and fat tissues may help better assess the prognosis and outcomes of patients with spinal metastases. Specifically, in obese patients, sarcopenia may be easily overlooked as a poor prognostic indicator. The role of this body phenotype, sarcopenic obesity (SO), has not been adequately studied among patients undergoing surgical treatment for spinal metastases. To this end, here the authors investigated the role of SO as a potential prognostic factor in patients undergoing surgical treatment for spinal metastases.

METHODS

The authors identified patients who underwent surgical treatment for spinal metastases between 2010 and 2020. A validated deep learning approach evaluated sarcopenia and adiposity on routine preoperative CT images. Based on composition analyses, patients were classified with SO or nonsarcopenic obesity. After nearest-neighbor propensity matching that accounted for confounders, the authors compared the rates and odds of postoperative complications, length of stay, 30-day readmission, and all-cause mortality at 90 days and 1 year between the SO and nonsarcopenic obesity groups.

RESULTS

A total of 62 patients with obesity underwent surgical treatment for spinal metastases during the study period. Of these, 37 patients had nonsarcopenic obesity and 25 had SO. After propensity matching, 50 records were evaluated that were equally composed of patients with nonsarcopenic obesity and SO (25 patients each). Patients with SO were noted to have increased odds of nonhome discharge (OR 6.0, 95% CI 1.69-21.26), 30-day readmission (OR 3.27, 95% CI 1.01-10.62), and 90-day (OR 4.85, 95% CI 1.29-18.26) and 1-year (OR 3.78, 95% CI 1.17-12.19) mortality, as well as increased time to mortality after surgery (12.60 ± 19.84 months vs 37.16 ± 35.19 months, p = 0.002; standardized mean difference 0.86). No significant differences were noted in terms of length of stay or postoperative complications when comparing the two groups (p > 0.05).

CONCLUSIONS

The SO phenotype was associated with increased odds of nonhome discharge, readmission, and postoperative mortality. This study suggests that SO may be an important prognostic factor to consider when developing care plans for patients with spinal metastases.

Racial/Ethnic Disparities Among Patients Undergoing Anterior Cervical Discectomy and Fusion or Posterior Cervical Decompression and Fusion for Cervical Spondylotic Myelopathy: A National Administrative Database Analysis

INTRODUCTION

The aim of this study was to investigate the impact of racial disparities on surgical outcomes for cervical spondylotic myelopathy (CSM).

METHODS

Adult patients undergoing anterior cervical discectomy and fusion (ACDF) or posterior cervical decompression and fusion (PCDF) for CSM were identified from the 2016 to 019 National Inpatient Sample Database using the International Classification of Diseases codes. Patients were categorized based on approach (ACDF or PCDF) and race/ethnicity (White, Black, Hispanic). Patient demographics, comorbidities, operative characteristics, adverse events, and health care resource utilization were assessed. Multivariate logistic regression analyses were used to identify independent predictors of extended length of stay (LOS), nonroutine discharge (NRD), and exorbitant costs.

RESULTS

A total of 46,500 patients were identified, of which 36,015 (77.5%) were White, 7465 (16.0%) were Black, and 3020 (6.5%) were Hispanic. Black and Hispanic patients had a greater comorbidity burden compared to White patients (P = 0.001) and a greater incidence of any postoperative complication (P = 0.001). Healthcare resource utilization were greater in the PCDF cohort than the ACDF cohort and greater in Black and Hispanic patients compared to White patients (P < 0.001). Black and Hispanic patient race were significantly associated with extended hospital LOS ([Black] odds ratio [OR]: 2.24, P < 0.001; [Hispanic] OR: 1.64, P < 0.001) and NRD ([Black] OR: 2.33, P < 0.001; [Hispanic] OR: 1.49, P = 0.016). Among patients who underwent PCDF, Black race was independently associated with extended hospital LOS ([Black] OR: 1.77, P < 0.001; [Hispanic] OR: 1.47, P = 0.167) and NRD ([Black] OR: 1.82, P < 0.001; [Hispanic] OR: 1.38, P = 0.052).

CONCLUSIONS

Our study suggests that patient race may influence patient outcomes and healthcare resource utilization following ACDF or PCDF for CSM.

Reliability of a Novel Classification System for Thoracic Disc Herniations

STUDY DESIGN

This is a cross-sectional survey.

OBJECTIVE

The aim was to assess the reliability of a proposed novel classification system for thoracic disc herniations (TDHs).

SUMMARY OF BACKGROUND DATA

TDHs are complex entities varying substantially in many factors, including size, location, and calcification. To date, no comprehensive system exists to categorize these lesions.

METHODS

Our proposed system classifies 5 types of TDHs using anatomic and clinical characteristics, with subtypes for calcification. Type 0 herniations are small (≤40% of spinal canal) TDHs without significant spinal cord or nerve root effacement; type 1 are small and paracentral; type 2 are small and central; type 3 are giant (>40% of spinal canal) and paracentral; and type 4 are giant and central. Patients with types 1 to 4 TDHs have correlative clinical and radiographic evidence of spinal cord compression. Twenty-one US spine surgeons with substantial TDH experience rated 10 illustrative cases to determine the system's reliability. Interobserver and intraobserver reliability were determined using the Fleiss kappa coefficient. Surgeons were also surveyed to obtain consensus on surgical approaches for the various TDH types.

RESULTS

High agreement was found for the classification system, with 80% (range 62% to 95%) overall agreement and high interrater and intrarater reliability (kappa 0.604 [moderate to substantial agreement] and kappa 0.630 [substantial agreement], respectively). All surgeons reported nonoperative management of type 0 TDHs. For type 1 TDHs, most respondents (71%) preferred posterior approaches. For type 2 TDHs, responses were roughly equivalent for anterolateral and posterior options. For types 3 and 4 TDHs, most respondents (72% and 68%, respectively) preferred anterolateral approaches.

CONCLUSIONS

This novel classification system can be used to reliably categorize TDHs, standardize description, and potentially guide the selection of surgical approach. Validation of this system with regard to treatment and clinical outcomes represents a line of future study.

The Impact of Preoperative Spinal Injection Timing on Postoperative Complications of Lumbar Decompression Surgery

BACKGROUND AND OBJECTIVES

Epidural steroid injections (ESIs) are commonly used for lower back pain management. The effect of these injections on lumbar decompression surgery outcomes is hitherto underexplored. The study objective was to determine the impact of ESIs on postoperative rates of medical and surgical complications and to define the appropriate interval before lumbar decompression surgery.

METHODS

This retrospective all-payer database analysis identified 587 651 adult patients undergoing one- to three-level laminectomies from January 2010 to October 2021. A 2:1 propensity score match accounting for comorbidities, levels of surgery, and demographics was performed to create two cohorts: (1) 43 674 patients who had received an ESI in the 90 days before laminectomy and (2) 87 348 patients who had not received an ESI. The primary outcome was the rates of medical and surgical complications between groups at 30 days postoperatively. Patients were divided into five cohorts based on injection time before surgery: 1 to 30 days, 31 to 45 days, 46 to 60 days, 61 to 75 days, and 76 to 90 days. Logistic regression was performed between groups to identify temporal associations of complication rates. Confidence intervals of 95% are provided when appropriate. P values < .01 were considered significant.

RESULTS

Rates of medical complications within 30 days of surgery were significantly higher in those with ESI compared with control (4.83% vs 3.9%, P < .001). Cerebrospinal fluid (CSF) leak rates were increased in the ESI group at 0.28% vs 0.1% (P < .001), but surgical site infection rates were not significantly different between groups (1.31% vs 1.42% P = .11). ESI performed within 30 days was associated with increased odds of CSF leak (OR: 5.32, 95% CI: 3.96-7.15).

CONCLUSION

Preoperative ESI increases the risk of CSF leak and medical complications after lumbar decompression. Because these complications were significantly associated with ESIs given 1 to 30 days before surgery, avoiding ESIs at least 30 days before surgery may be advisable.

AO Spine Guideline for the Use of Osteobiologics (AOGO) in Anterior Cervical Discectomy and Fusion for Spinal Degenerative Cases

STUDY DESIGN

Guideline.

OBJECTIVES

To develop an international guideline (AOGO) about the use of osteobiologics in anterior cervical discectomy and fusion (ACDF) for treating degenerative spine conditions.

METHODS

The guideline development process was guided by AO Spine Knowledge Forum Degenerative (KF Degen) and followed the Guideline International Network McMaster Guideline Development Checklist. The process involved 73 participants with expertise in degenerative spine diseases and surgery from 22 countries. Fifteen systematic reviews were conducted addressing respective key topics and evidence was collected. The methodologist compiled the evidence into GRADE Evidence-to-Decision frameworks. Guideline panel members judged the outcomes and other criteria and made the final recommendations through consensus.

RESULTS

Five conditional recommendations were created. A conditional recommendation is about the use of allograft, autograft or a cage with an osteobiologic in primary ACDF surgery. Other conditional recommendations are about the use of osteobiologic for single- or multi-level ACDF, and for hybrid construct surgery. It is suggested that surgeons use other osteobiologics rather than human bone morphogenetic protein-2 (BMP-2) in common clinical situations. Surgeons are recommended to choose 1 graft over another or 1 osteobiologic over another primarily based on clinical situation, and the costs and availability of the materials.

CONCLUSION

This AOGO guideline is the first to provide recommendations for the use of osteobiologics in ACDF. Despite the comprehensive searches for evidence, there were few studies completed with small sample sizes and primarily as case series with inherent risks of bias. Therefore, high-quality clinical evidence is demanded to improve the guideline.

Demographic Representation in 3 Leading Artificial Intelligence Text-to-Image Generators

Importance

The progression of artificial intelligence (AI) text-to-image generators raises concerns of perpetuating societal biases, including profession-based stereotypes.

Objective

To gauge the demographic accuracy of surgeon representation by 3 prominent AI text-to-image models compared to real-world attending surgeons and trainees.

Design, Setting, and Participants

The study used a cross-sectional design, assessing the latest release of 3 leading publicly available AI text-to-image generators. Seven independent reviewers categorized AI-produced images. A total of 2400 images were analyzed, generated across 8 surgical specialties within each model. An additional 1200 images were evaluated based on geographic prompts for 3 countries. The study was conducted in May 2023. The 3 AI text-to-image generators were chosen due to their popularity at the time of this study. The measure of demographic characteristics was provided by the Association of American Medical Colleges subspecialty report, which references the American Medical Association master file for physician demographic characteristics across 50 states. Given changing demographic characteristics in trainees compared to attending surgeons, the decision was made to look into both groups separately. Race (non-White, defined as any race other than non-Hispanic White, and White) and gender (female and male) were assessed to evaluate known societal biases.

Exposures

Images were generated using a prompt template, "a photo of the face of a [blank]", with the blank replaced by a surgical specialty. Geographic-based prompting was evaluated by specifying the most populous countries on 3 continents (the US, Nigeria, and China).

Main Outcomes and Measures

The study compared representation of female and non-White surgeons in each model with real demographic data using χ2, Fisher exact, and proportion tests.

Results

There was a significantly higher mean representation of female (35.8% vs 14.7%; P < .001) and non-White (37.4% vs 22.8%; P < .001) surgeons among trainees than attending surgeons. DALL-E 2 reflected attending surgeons' true demographic data for female surgeons (15.9% vs 14.7%; P = .39) and non-White surgeons (22.6% vs 22.8%; P = .92) but underestimated trainees' representation for both female (15.9% vs 35.8%; P < .001) and non-White (22.6% vs 37.4%; P < .001) surgeons. In contrast, Midjourney and Stable Diffusion had significantly lower representation of images of female (0% and 1.8%, respectively; P < .001) and non-White (0.5% and 0.6%, respectively; P < .001) surgeons than DALL-E 2 or true demographic data. Geographic-based prompting increased non-White surgeon representation but did not alter female representation for all models in prompts specifying Nigeria and China.

Conclusion and Relevance

In this study, 2 leading publicly available text-to-image generators amplified societal biases, depicting over 98% surgeons as White and male. While 1 of the models depicted comparable demographic characteristics to real attending surgeons, all 3 models underestimated trainee representation. The study suggests the need for guardrails and robust feedback systems to minimize AI text-to-image generators magnifying stereotypes in professions such as surgery.

Treatment of intramedullary spinal cord tumors: a modified Delphi technique of the North American Spine Society Section of Spine Oncology

OBJECTIVE

Intramedullary spinal cord tumors (IMSCTs) are rare tumors with heterogeneous presentations and natural histories that complicate their management. Standardized guidelines are lacking on when to surgically intervene and the appropriate aggressiveness of resection, especially given the risk of new neurological deficits following resection of infiltrative tumors. Here, the authors present the results of a modified Delphi method using input from surgeons experienced with IMSCT removal to construct a framework for the operative management of IMSCTs based on the clinical, radiographic, and tumor-specific characteristics.

METHODS

A modified Delphi technique was conducted using a group of 14 neurosurgeons experienced in IMSCT resection. Three rounds of written correspondence, surveys, and videoconferencing were carried out. Participants were queried about clinical and radiographic criteria used to determine operative candidacy and guide decision-making. Members then completed a final survey indicating their choice of observation or surgery, choice of resection strategy, and decision to perform duraplasty, in response to a set of patient- and tumor-specific characteristics. Consensus was defined as ≥ 80% agreement, while responses with 70%-79% agreement were defined as agreement.

RESULTS

Thirty-six total characteristics were assessed. There was consensus favoring surgical intervention for patients with new-onset myelopathy (86% agreement), chronic myelopathy (86%), or progression from mild to disabling numbness (86%), but disagreement for patients with mild numbness or chronic paraplegia. Age was not a determinant of operative candidacy except among frail patients, who were deemed more suitable for observation (93%). Well-circumscribed (93%) or posteriorly located tumors reaching the surface (86%) were consensus surgical lesions, and participants agreed that the presence of syringomyelia (71%) and peritumoral T2 signal change (79%) were favorable indications for surgery. There was consensus that complete loss of transcranial motor evoked potentials with a 50% decrease in the D-wave amplitude should halt further resection (93%). Preoperative symptoms seldom influenced choice of resection strategy, while a distinct cleavage plane (100%) or visible tumor-cord margins (100%) strongly favored gross-total resection.

CONCLUSIONS

The authors present a modified Delphi technique highlighting areas of consensus and agreement regarding surgical management of IMSCTs. Although not intended as a substitute for individual clinical decision-making, the results can help guide care of these patients. Additionally, areas of controversy meriting further investigation are highlighted.

Racial disparities in the management and outcomes of primary osseous neoplasms of the spine: a SEER analysis

PURPOSE

Primary osseous neoplasms of the spine, including Ewing's sarcoma, osteosarcoma, chondrosarcoma, and chordoma, are rare tumors with significant morbidity and mortality. The present study aims to identify the prevalence and impact of racial disparities on management and outcomes of patients with these malignancies.

METHODS

The 2000 to 2020 Surveillance, Epidemiology, and End Results (SEER) Registry, a cancer registry, was retrospectively reviewed to identify patients with Ewing's sarcoma, osteosarcoma, chondrosarcoma, or chordoma of the vertebral column or sacrum/pelvis. Study patients were divided into race-based cohorts: White, Black, Hispanic, and Other. Demographics, tumor characteristics, treatment variables, and mortality were assessed.

RESULTS

2,415 patients were identified, of which 69.8% were White, 5.8% Black, 16.1% Hispanic, and 8.4% classified as "Other". Tumor type varied significantly between cohorts, with osteosarcoma affecting a greater proportion of Black patients compared to the others (p < 0.001). A lower proportion of Black and Other race patients received surgery compared to White and Hispanic patients (p < 0.001). Utilization of chemotherapy was highest in the Hispanic cohort (p < 0.001), though use of radiotherapy was similar across cohorts (p = 0.123). Five-year survival (p < 0.001) and median survival were greatest in White patients (p < 0.001). Compared to non-Hispanic Whites, Hispanic (p < 0.001) and "Other" patients (p < 0.001) were associated with reduced survival.

CONCLUSION

Race may be associated with tumor characteristics at diagnosis (including subtype, size, and site), treatment utilization, and mortality, with non-White patients having lower survival compared to White patients. Further studies are necessary to identify underlying causes of these disparities and solutions for eliminating them.

A guide to finite element analysis models of the spine for clinicians

Finite element analysis (FEA) is a computer-based mathematical method commonly used in spine and orthopedic biomechanical research. Advances in computational power and engineering modeling and analysis software have enabled many recent technical applications of FEA. Through the use of FEA, a wide range of scenarios can be simulated, such as physiological processes, mechanisms of disease and injury, and the efficacy of surgical procedures. Such models have the potential to enhance clinical studies by allowing comparisons of surgical treatments that would be impractical to perform in human or animal studies, and by linking model results to treatment outcomes. While traditional ex vivo experiments are limited by variabilities in tissue, the complexity of test setup, cost, measurable biomechanical parameters, and the repeatability of experiments, FEA models can be used to measure a wide range of clinically relevant biomechanical parameters. Generic or patient-specific anatomical models can be modified to simulate different clinical and surgical conditions under simulated physiological conditions. Despite these capabilities, there is limited understanding of the clinical applicability and translational potential of FEA models. For spine surgeons, a comprehensive understanding of the key features, strengths, and limitations of FEA models of the spine and their ability to personalize treatment options and assist in clinical decision-making would significantly enhance the impact of FEA research. Furthermore, fostering collaborations between surgeons and engineers could augment the clinical use of these models. The purpose of this review was to highlight key features of FEA model building for clinicians. To illustrate these features, the authors present an example of the use of FEA models in comparing FDA-approved disc arthroplasty implants.

Impact of Preoperative Frailty on Outcomes in Patients with Cervical Spondylotic Myelopathy Undergoing Anterior vs. Posterior Cervical Surgery

Introduction: Frailty has been shown to negatively influence patient outcomes across many disease processes, including in the cervical spondylotic myelopathy (CSM) population. The aim of this study was to assess the impact that frailty has on patients with CSM who undergo anterior cervical discectomy and fusion (ACDF) or posterior cervical decompression and fusion (PCDF). Materials and Methods: A retrospective cohort study was performed using the 2016-2019 national inpatient sample. Adult patients (≥18 years old) undergoing ACDF only or PCDF only for CSM were identified using ICD codes. The patients were categorized based on receipt of ACDF or PCDF and pre-operative frailty status using the 11-item modified frailty index (mFI-11): pre-Frail (mFI = 1), frail (mFI = 2), or severely frail (mFI ≥ 3). Patient demographics, comorbidities, operative characteristics, perioperative adverse events (AEs), and healthcare resource utilization were assessed. Multivariate logistic regression analyses were used to identify independent predictors of extended length of stay (LOS) and non-routine discharge (NRD). Results: A total of 37,990 patients were identified, of which 16,665 (43.9%) were in the pre-frail cohort, 12,985 (34.2%) were in the frail cohort, and 8340 (22.0%) were in the severely frail cohort. The prevalence of many comorbidities varied significantly between frailty cohorts. Across all three frailty cohorts, the incidence of AEs was greater in patients who underwent PCDF, with dysphagia being significantly more common in patients who underwent ACDF. Additionally, the rate of adverse events significantly increased between ACDF and PCDF with respect to increasing frailty (p < 0.001). Regarding healthcare resource utilization, LOS and rate of NRD were significantly greater in patients who underwent PCDF in all three frailty cohorts, with these metrics increasing with frailty in both ACDF and PCDF cohorts (LOS: p < 0.001); NRD: p < 0.001). On a multivariate analysis of patients who underwent ACDF, frailty and severe frailty were found to be independent predictors of extended LOS [(frail) OR: 1.39, p < 0.001; (severely frail) OR: 2.25, p < 0.001] and NRD [(frail) OR: 1.49, p < 0.001; (severely frail) OR: 2.22, p < 0.001]. Similarly, in patients who underwent PCDF, frailty and severe frailty were found to be independent predictors of extended LOS [(frail) OR: 1.58, p < 0.001; (severely frail) OR: 2.45, p < 0.001] and NRD [(frail) OR: 1.55, p < 0.001; (severely frail) OR: 1.63, p < 0.001]. Conclusions: Our study suggests that preoperative frailty may impact outcomes after surgical treatment for CSM, with more frail patients having greater health care utilization and a higher rate of adverse events. The patients undergoing PCDF ensued increased health care utilization, compared to ACDF, whereas severely frail patients undergoing PCDF tended to have the longest length of stay and highest rate of non-routine discharge. Additional prospective studies are necessary to directly compare ACDF and PCDF in frail patients with CSM.

The prognostic nutritional index (PNI) is independently associated with 90-day and 12-month mortality after metastatic spinal tumor surgery

INTRODUCTION

Estimated postoperative survival is an important consideration during the decision-making process for patients with spinal metastases. Nutritional status has been associated with poor outcomes and limited survival in the general cancer population. The objective of this study was to evaluate the predictive utility of the prognostic nutritional index (PNI) for postoperative mortality after spinal metastasis surgery.

METHODS

A total of 139 patients who underwent oncologic surgery for spinal metastases between April 2012 and August 2022 and had a minimum 90-day follow-up were included. PNI was calculated using preoperative serum albumin and total lymphocyte count, with PNI < 40 defined as low. The mean PNI of our cohort was 43 (standard deviation: 7.7). The primary endpoint was 90-day mortality, and the secondary endpoint was 12-month mortality. Multivariate logistic regression analyses were performed.

RESULTS

The 90-day mortality was 27% (37/139), and the 12-month mortality was 56% (51/91). After controlling for age, ECOG performance status, total psoas muscle cross-sectional area (TPA), and primary cancer site, the PNI was associated with 90-day mortality [odds ratio 0.86 (95% confidence interval 0.79-0.94); p = 0.001]. After controlling for ECOG performance status and primary cancer site, the PNI was associated with 12-month mortality [OR 0.89 (95% CI 0.82-0.97); p = 0.008]. Patients with a low PNI had a 50% mortality rate at 90 days and an 84% mortality rate at 12 months.

CONCLUSION

The PNI was independently associated with 90-day and 12-month mortality after metastatic spinal tumor surgery, independent of performance status, TPA, and primary cancer site.

Performance of ChatGPT and GPT-4 on Neurosurgery Written Board Examinations

BACKGROUND AND OBJECTIVES

Interest surrounding generative large language models (LLMs) has rapidly grown. Although ChatGPT (GPT-3.5), a general LLM, has shown near-passing performance on medical student board examinations, the performance of ChatGPT or its successor GPT-4 on specialized examinations and the factors affecting accuracy remain unclear. This study aims to assess the performance of ChatGPT and GPT-4 on a 500-question mock neurosurgical written board examination.

METHODS

The Self-Assessment Neurosurgery Examinations (SANS) American Board of Neurological Surgery Self-Assessment Examination 1 was used to evaluate ChatGPT and GPT-4. Questions were in single best answer, multiple-choice format. χ 2 , Fisher exact, and univariable logistic regression tests were used to assess performance differences in relation to question characteristics.

RESULTS

ChatGPT (GPT-3.5) and GPT-4 achieved scores of 73.4% (95% CI: 69.3%-77.2%) and 83.4% (95% CI: 79.8%-86.5%), respectively, relative to the user average of 72.8% (95% CI: 68.6%-76.6%). Both LLMs exceeded last year's passing threshold of 69%. Although scores between ChatGPT and question bank users were equivalent ( P = .963), GPT-4 outperformed both (both P < .001). GPT-4 answered every question answered correctly by ChatGPT and 37.6% (50/133) of remaining incorrect questions correctly. Among 12 question categories, GPT-4 significantly outperformed users in each but performed comparably with ChatGPT in 3 (functional, other general, and spine) and outperformed both users and ChatGPT for tumor questions. Increased word count (odds ratio = 0.89 of answering a question correctly per +10 words) and higher-order problem-solving (odds ratio = 0.40, P = .009) were associated with lower accuracy for ChatGPT, but not for GPT-4 (both P > .005). Multimodal input was not available at the time of this study; hence, on questions with image content, ChatGPT and GPT-4 answered 49.5% and 56.8% of questions correctly based on contextual context clues alone.

CONCLUSION

LLMs achieved passing scores on a mock 500-question neurosurgical written board examination, with GPT-4 significantly outperforming ChatGPT.

Outcomes Following 2-Level Cervical Interventions with Cage-and-Plate, Zero-Profile, or Arthroplasty Constructs

BACKGROUND

Though cage-and-plate constructs are widely used for disk height restoration in surgery for cervical disc disease, concerns over range of motion limitations and adjacent disc space violations have fueled the development of artificial disc and zero-profile constructs. This study investigated the outcomes of patients undergoing two-level cervical interventions via arthroplasty, cage-and-plate, or zero-profile constructs.

METHODS

Patients undergoing two-level anterior cervical procedures between 2010 and 2020 were identified using an all-payer claims database. Logistic regression models were utilized to develop criteria for a 1:1:1-exact match procedure. The primary outcome was the need for additional surgery within 30 months, and the secondary outcomes included medical and surgical complications observed within 30 days of index intervention. P values < 0.05 were considered statistically significant.

RESULTS

133,831 patients were identified as undergoing two-level anterior cervical interventions. Seven thousand three hundred seventy-one records were analyzed through a 1:1:1 match. Patients who received zero-profile versus cage-and-plate constructs had significantly decreased odds of requiring additional surgery within 30 months (Odds Ratio [OR] 0.64; 95% Confidence Interval [CI] 0.51-0.81). However, postoperative medical complications were increased among patients who received zero-profile constructs compared to cage-and-plate (OR 1.59; 95%CI 1.07-2.37). Patients who underwent arthroplasty also had decreased odds for additional surgery versus cage-and-plate (OR 0.75; 95%CI 0.60-0.93). There was no significant difference between arthroplasty and cage-and-plate constructs in developing postoperative surgical or medical complications.

CONCLUSIONS

Among patients undergoing two-level interventions, cage-and-plate constructs were associated with increased odds of additional surgery within 30 months following index procedures when compared to zero-profile constructs or arthroplasty.

Atlantoaxial instability associated with ALDH18A1 mutation

We report a 10-year-old boy with a de novo pathogenic variant in ALDH18A1, a rare form of metabolic cutis laxa, which was complicated by atlantoaxial instability and spinal cord compression following a fall from standing height. The patient required emergent cervical spine fusion and decompression followed by a 2-month hospitalization and rehabilitation. In addition to the core clinical features of joint and skin laxity, hypotonia, and developmental delays, we expand the connective tissue phenotype by adding a new potential feature of cervical spine instability. Patients with pathogenic variants in ALDH18A1 may warrant cervical spine screening to minimize possible morbidity. Neurosurgeons, geneticists, primary care providers, and families should be aware of the increased risk of severe cervical injury from minor trauma.

Performance of ChatGPT, GPT-4, and Google Bard on a Neurosurgery Oral Boards Preparation Question Bank

BACKGROUND AND OBJECTIVES

General large language models (LLMs), such as ChatGPT (GPT-3.5), have demonstrated the capability to pass multiple-choice medical board examinations. However, comparative accuracy of different LLMs and LLM performance on assessments of predominantly higher-order management questions is poorly understood. We aimed to assess the performance of 3 LLMs (GPT-3.5, GPT-4, and Google Bard) on a question bank designed specifically for neurosurgery oral boards examination preparation.

METHODS

The 149-question Self-Assessment Neurosurgery Examination Indications Examination was used to query LLM accuracy. Questions were inputted in a single best answer, multiple-choice format. χ 2 , Fisher exact, and univariable logistic regression tests assessed differences in performance by question characteristics.

RESULTS

On a question bank with predominantly higher-order questions (85.2%), ChatGPT (GPT-3.5) and GPT-4 answered 62.4% (95% CI: 54.1%-70.1%) and 82.6% (95% CI: 75.2%-88.1%) of questions correctly, respectively. By contrast, Bard scored 44.2% (66/149, 95% CI: 36.2%-52.6%). GPT-3.5 and GPT-4 demonstrated significantly higher scores than Bard (both P < .01), and GPT-4 outperformed GPT-3.5 ( P = .023). Among 6 subspecialties, GPT-4 had significantly higher accuracy in the Spine category relative to GPT-3.5 and in 4 categories relative to Bard (all P < .01). Incorporation of higher-order problem solving was associated with lower question accuracy for GPT-3.5 (odds ratio [OR] = 0.80, P = .042) and Bard (OR = 0.76, P = .014), but not GPT-4 (OR = 0.86, P = .085). GPT-4's performance on imaging-related questions surpassed GPT-3.5's (68.6% vs 47.1%, P = .044) and was comparable with Bard's (68.6% vs 66.7%, P = 1.000). However, GPT-4 demonstrated significantly lower rates of "hallucination" on imaging-related questions than both GPT-3.5 (2.3% vs 57.1%, P < .001) and Bard (2.3% vs 27.3%, P = .002). Lack of question text description for questions predicted significantly higher odds of hallucination for GPT-3.5 (OR = 1.45, P = .012) and Bard (OR = 2.09, P < .001).

CONCLUSION

On a question bank of predominantly higher-order management case scenarios for neurosurgery oral boards preparation, GPT-4 achieved a score of 82.6%, outperforming ChatGPT and Google Bard.

Predictive value of six nutrition biomarkers in oncological spine surgery: a performance assessment for prediction of mortality and wound infection

OBJECTIVE

Assessment of nutritional status is fundamental in cancer patients. The objective of this study was to assess the predictive ability of 6 nutritional biomarkers for postoperative mortality and wound infection after metastatic spinal tumor surgery.

METHODS

A total of 139 patients who underwent oncological surgery for metastatic spine disease between April 2012 and August 2022 and had a minimum follow-up of 90 days were included. Six unique nutritional biomarkers were assessed: Prognostic Nutritional Index (PNI), Nutritional Risk Index (NRI), Controlling Nutritional Status Score (CONUT), total psoas cross-sectional area (TPA), body mass index (BMI), and body weight. Study endpoints were 90-day mortality rate, 12-month mortality rate, and wound infection. The discriminative ability of each of these markers was assessed with the c-statistic. A multivariate analysis was done for each of the biomarkers after a univariate analysis was first performed.

RESULTS

The 90-day mortality rate was 27% (37 of 139). The biomarkers and respective c-statistics were as follows: PNI (0.74), NRI (0.75), CONUT (0.71), TPA (0.64), BMI (0.59), and body weight (0.60). The 12-month mortality rate was 56% (51 of 91). The biomarkers and respective c-statistics were as follows: PNI (0.72), NRI (0.73), CONUT (0.70), TPA (0.63), BMI (0.59), and body weight (0.60). The wound infection rate was 8% (11 of 139). The biomarkers and respective c-statistics were as follows: PNI (0.57), NRI (0.53), CONUT (0.55), TPA (0.57), BMI (0.48), and body weight (0.52). The PNI, NRI, and CONUT all predicted 90-day and 12-month mortality after multivariate regression analysis. No association between nutrition and wound infection was found.

CONCLUSIONS

In this study, nutritional status was associated with postoperative mortality following oncological spine surgery. Three biomarkers predicted outcome independent of variables such as performance status or primary cancer. Future validation of these metrics is needed.

Implication of nutritional status for adverse outcomes after surgery for metastatic spine tumors

OBJECTIVE

Surgery for metastatic spinal tumors can have a substantial impact on patients' quality of life by alleviating pain, improving function, and correcting spinal instability when indicated. The decision to operate is difficult because many patients with cancer are frail. Studies have highlighted the importance of preoperative nutritional status assessments; however, little is known about which aspects of nutrition accurately inform clinical outcomes. This study investigates the interaction and prognostic importance of various nutritional and frailty measures in patients with spinal metastases.

METHODS

A retrospective analysis of consecutive patients who underwent surgery for spinal metastases between 2014 and 2020 at the Massachusetts General Hospital was performed. Patients were stratified according to the New England Spinal Metastasis Score (NESMS). Frailty was assessed using the metastatic spinal tumor frailty index. Nutrition was assessed using the prognostic nutritional index (PNI), preoperative body mass index, albumin, albumin-to-globulin ratio, and platelet-to-lymphocyte ratio. Outcomes included postoperative survival and complication rates, with focus on wound-related complications.

RESULTS

This study included 154 individuals (39% female; mean [SD] age 63.23 [13.14] years). NESMS 0 and NESMS 3 demonstrated the highest proportions of severely frail patients (56.2%) and nonfrail patients (16.1%), respectively. Patients with normal nutritional status (albumin-to-globulin ratio and PNI) had a better prognosis than those with poor nutritional status when stratified by NESMS. Multivariable regression adjusted for NESMS and frailty showed that a PNI > 40.4 was significantly associated with decreased odds of 90-day complications (OR 0.93, 95% CI 0.85-0.98). After accounting for age, sex, primary tumor pathology, physical function, nutritional status, and frailty, a preoperative nutrition consultation was associated with a decrease in postoperative wound-related complications (average marginal effect -5.00%; 95% CI -1.50% to -8.9%).

CONCLUSIONS

The PNI was most predictive of complications and may be a key biomarker for risk stratification in the 90 days following surgery. Nutrition consultation was associated with a reduced risk of wound-related complications, attesting to the importance of this preoperative intervention. These findings suggest that nutrition plays an important role in the postsurgical course and should be considered when developing a treatment plan for spinal metastases.

Surgery and proton radiation therapy for pediatric base of skull chordomas: Long-term clinical outcomes for 204 patients

BACKGROUND

Data on clinical outcomes for base of skull (BOS) chordomas in the pediatric population is limited. We report patient outcomes after surgery and proton radiotherapy (PRT).

METHODS

Pediatric patients with BOS chordomas were treated with PRT or combined proton/photon approach (proton-based; for most, 80% proton/20% photon) at the Massachusetts General Hospital from 1981 to 2021. Endpoints of interest were overall survival (OS), disease-specific survival, progression-free survival (PFS), freedom from local recurrence (LC), and freedom from distant failure (DC).

RESULTS

Of 204 patients, median age at diagnosis was 11.1 years (range, 1-21). Chordoma location included 59% upper and/or middle clivus, 36% lower clivus, 4% craniocervical junction, and 1% nasal cavity. Fifteen (7%) received pre-RT chemotherapy. Forty-seven (23%) received PRT, and 157 (77%) received comboRT. Median total dose was 76.7 Gy (RBE) (range, 59.3-83.3). At a median follow-up of 10 years (interquartile range, 5-16 years), 56 recurred. Median OS and PFS were 26 and 25 years, with 5-, 10-, and 20-year OS and PFS rates of 84% and 74%, 78% and 69%, and 64% and 64%, respectively. Multivariable actuarial analyses showed poorly differentiated subtype, radiographical progression prior to RT, larger treatment volume, and lower clivus location to be prognostic factors for worse OS, PFS, and LC. RT was well tolerated at a median follow-up of 9 years (interquartile range, 4-16 years). Side effects included 166 patients (80%) with mild/moderate acute toxicities, 24 (12%) patients with late toxicities, and 4 (2%) who developed secondary radiation-related malignancies.

CONCLUSION

This is the largest cohort of BOS chordomas in the literature, pediatric and/or adult. High-dose PRT following surgical resection is effective with low rates of late toxicity.

Hospital Frailty Risk Score and Healthcare Resource Utilization After Surgery for Primary Spinal Intradural/Cord Tumors

OBJECTIVE

The Hospital Frailty Risk Score (HFRS) is a metric that measures frailty among patients in large national datasets using ICD-10 codes. While other metrics have been utilized to demonstrate the association between frailty and poor outcomes in spine oncology, none have examined the HFRS. The aim of this study was to investigate the impact of frailty using the HFRS on complications, length of stay, cost of admission, and discharge disposition in patients undergoing surgery for primary tumors of the spinal cord and meninges.

METHODS

A retrospective cohort study was performed using the Nationwide Inpatient Sample database from 2016 to 2018. Adult patients undergoing surgery for primary tumors of the spinal cord and meninges were identified using ICD-10-CM codes. Patients were categorized into 2 cohorts based on HFRS score: Non-Frail (HFRS<5) and Frail (HFRS≥5). Patient characteristics, treatment, perioperative complications, LOS, discharge disposition, and cost of admission were assessed.

RESULTS

Of the 5955 patients identified, 1260 (21.2%) were Frail. On average, the Frail cohort was nearly 8 years older (P < .001) and experienced more postoperative complications (P = .001). The Frail cohort experienced longer LOS (P < .001), a higher rate of non-routine discharge (P = .001), and a greater mean cost of admission (P < .001). Frailty was found to be an independent predictor of extended LOS (P < .001) and non-routine discharge (P < .001).

CONCLUSION

Our study is the first to use the HFRS to assess the impact of frailty on patients with primary spinal tumors. We found that frailty was associated with prolonged LOS, non-routine discharge, and increased hospital costs.