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.

Chromosome-scale mega-haplotypes enable digital karyotyping of cancer aneuploidy.

Genomic instability is a frequently occurring feature of cancer that involves large-scale structural alterations. These somatic changes in chromosome structure include duplication of entire chromosome arms and aneuploidy where chromosomes are duplicated beyond normal diploid content. However, the accurate determination of aneuploidy events in cancer genomes is a challenge. Recent advances in sequencing technology allow the characterization of haplotypes that extend megabases along the human genome using high molecular weight (HMW) DNA. For this study, we employed a library preparation method in which sequence reads have barcodes linked to single HMW DNA molecules. Barcode-linked reads are used to generate extended haplotypes on the order of megabases. We developed a method that leverages haplotypes to identify chromosomal segmental alterations in cancer and uses this information to join haplotypes together, thus extending the range of phased variants. With this approach, we identified mega-haplotypes that encompass entire chromosome arms. We characterized the chromosomal arm changes and aneuploidy events in a manner that offers similar information as a traditional karyotype but with the benefit of DNA sequence resolution. We applied this approach to characterize aneuploidy and chromosomal alterations from a series of primary colorectal cancers.

Awake versus asleep deep brain stimulation for Parkinson's disease: a critical comparison and meta-analysis.

OBJECTIVE: No definitive comparative studies of the efficacy of 'awake' deep brain stimulation (DBS) for Parkinson's disease (PD) under local or general anaesthesia exist, and there remains significant debate within the field regarding differences in outcomes between these two techniques. METHODS: We conducted a literature review and meta-analysis of all published DBS for PD studies (n=2563) on PubMed from January 2004 to November 2015. Inclusion criteria included patient number >15, report of precision and/or clinical outcomes data, and at least 6 months of follow-up. There were 145 studies, 16 of which were under general anaesthesia. Data were pooled using an inverse-variance weighted, random effects meta-analytic model for observational data. RESULTS: There was no significant difference in mean target error between local and general anaesthesia, but there was a significantly less mean number of DBS lead passes with general anaesthesia (p=0.006). There were also significant decreases in DBS complications, with fewer intracerebral haemorrhages and infections with general anaesthesia (p<0.001). There were no significant differences in Unified Parkinson's Disease Rating Scale (UPDRS) Section II scores off medication, UPDRS III scores off and on medication or levodopa equivalent doses between the two techniques. Awake DBS cohorts had a significantly greater decrease in treatment-related side effects as measured by the UPDRS IV off medication score (78.4% awake vs 59.7% asleep, p=0.022). CONCLUSIONS: Our meta-analysis demonstrates that while DBS under general anaesthesia may lead to lower complication rates overall, awake DBS may lead to less treatment-induced side effects. Nevertheless, there were no significant differences in clinical motor outcomes between the two techniques. Thus, DBS under general anaesthesia can be considered at experienced centres in patients who are not candidates for traditional awake DBS or prefer the asleep alternative.

Single-Cell RNA-Sequencing in Glioma.

PURPOSE OF REVIEW: In this review, we seek to summarize the literature concerning the use of single-cell RNA-sequencing for CNS gliomas. RECENT FINDINGS: Single-cell analysis has revealed complex tumor heterogeneity, subpopulations of proliferating stem-like cells and expanded our view of tumor microenvironment influence in the disease process. Although bulk RNA-sequencing has guided our initial understanding of glioma genetics, this method does not accurately define the heterogeneous subpopulations found within these tumors. Single-cell techniques have appealing applications in cancer research, as diverse cell types and the tumor microenvironment have important implications in therapy. High cost and difficult protocols prevent widespread use of single-cell RNA-sequencing; however, continued innovation will improve accessibility and expand our of knowledge gliomas.

A review of potential applications of MR-guided focused ultrasound for targeting brain tumor therapy.

Magnetic resonance-guided focused ultrasound (MRgFUS) has been used extensively to ablate brain tissue in movement disorders, such as essential tremor. At a lower energy, MRgFUS can disrupt the blood-brain barrier (BBB) to allow passage of drugs. This focal disruption of the BBB can target systemic medications to specific portions of the brain, such as for brain tumors. Current methods to bypass the BBB are invasive, as the BBB is relatively impermeable to systemically delivered antineoplastic agents. Multiple healthy and brain tumor animal models have suggested that MRgFUS disrupts the BBB and focally increases the concentration of systemically delivered antitumor chemotherapy, immunotherapy, and gene therapy. In animal tumor models, combining MRgFUS with systemic drug delivery increases median survival times and delays tumor progression. Liposomes, modified microbubbles, and magnetic nanoparticles, combined with MRgFUS, more effectively deliver chemotherapy to brain tumors. MRgFUS has great potential to enhance brain tumor drug delivery, while limiting treatment toxicity to the healthy brain.

A pilot study on the use of cerebrospinal fluid cell-free DNA in intramedullary spinal ependymoma.

Cerebrospinal fluid (CSF) represents a promising source of cell-free DNA (cfDNA) for tumors of the central nervous system. A CSF-based liquid biopsy may obviate the need for riskier tissue biopsies and serve as a means for monitoring tumor recurrence or response to therapy. Spinal ependymomas most commonly occur in adults, and aggressive resection must be delicately balanced with the risk of injury to adjacent normal tissue. In patients with subtotal resection, recurrence commonly occurs. A CSF-based liquid biopsy matched to the patient's spinal ependymoma mutation profile has potential to be more sensitive then surveillance MRI, but the utility has not been well characterized for tumors of the spinal cord. In this study, we collected matched blood, tumor, and CSF samples from three adult patients with WHO grade II intramedullary spinal ependymoma. We performed whole exome sequencing on matched tumor and normal DNA to design Droplet Digital™ PCR (ddPCR) probes for tumor and wild-type mutations. We then interrogated CSF samples for tumor-derived cfDNA by performing ddPCR on extracted cfDNA. Tumor cfDNA was not reliably detected in the CSF of our cohort. Anatomic sequestration and low grade of intramedullary spinal cord tumors likely limits the role of CSF liquid biopsy.

Tumor DNA in cerebral spinal fluid reflects clinical course in a patient with melanoma leptomeningeal brain metastases.

Cerebral spinal fluid (CSF) from brain tumor patients contains tumor cellular and cell-free DNA (cfDNA), which provides a less-invasive and routinely accessible method to obtain tumor genomic information. In this report, we used droplet digital PCR to test mutant tumor DNA in CSF of a patient to monitor the treatment response of metastatic melanoma leptomeningeal disease (LMD). The primary melanoma was known to have a BRAF (V600E) mutation, and the patient was treated with whole brain radiotherapy and BRAF inhibitors. We collected 9 CSF samples over 6 months. The mutant cfDNA fraction gradually decreased from 53 % (time of diagnosis) to 0 (time of symptom alleviation) over the first 6 time points. Three months after clinical improvement, the patient returned with severe symptoms and the mutant cfDNA was again detected in CSF at high levels. The mutant DNA fraction corresponded well with the patient's clinical response. We used whole exome sequencing to examine the mutation profiles of the LMD tumor DNA in CSF before therapeutic response and after disease relapse, and discovered a canonical cancer mutation PTEN (R130*) at both time points. The cellular and cfDNA revealed similar mutation profiles, suggesting cfDNA is representative of LMD cells. This study demonstrates the potential of using cellular or cfDNA in CSF to monitor treatment response for LMD.

The "Liquid Biopsy": the Role of Circulating DNA and RNA in Central Nervous System Tumors.

The detection of tumor-derived circulating nucleic acids in patients with cancer, known as the "liquid biopsy," has expanded from use in plasma to other bodily fluids in an increasing number of malignancies. Circulating nucleic acids could be of particular use in central nervous system tumors as biopsy carries a 5-7 % risk of major morbidity. This application presents unique challenges that have limited the use of cell-free DNA and RNA in the diagnosis and monitoring of CNS tumors. Recent work suggests that cerebrospinal fluid may be a useful source of CNS tumor-derived circulating nucleic acids. In this review, we discuss the available data and future outlook on the use of the liquid biopsy for CNS tumors.

The strokes that killed Churchill, Roosevelt, and Stalin.

From February 4 to 11, 1945, President Franklin D. Roosevelt of the United States, Soviet Union Premier Joseph Stalin, and British Prime Minister Winston Churchill met near Yalta in Crimea to discuss how post-World War II (WWII) Europe should be organized. Within 2 decades of this conference, all 3 men had died. President Roosevelt died 2 months after the Yalta Conference due to a hemorrhagic stroke. Premier Stalin died 8 years later, also due to a hemorrhagic stroke. Finally, Prime Minister Churchill died 20 years after the conference because of complications due to stroke. At the time of Yalta, these 3 men were the leaders of the most powerful countries in the world. The subsequent deterioration of their health and eventual death had varying degrees of historical significance. Churchill's illness forced him to resign as British prime minister, and the events that unfolded immediately after his resignation included Britain's mismanagement of the Egyptian Suez Crisis and also a period of mistrust with the United States. Furthermore, Roosevelt was still president and Stalin was still premier at their times of passing, so their deaths carried huge political ramifications not only for their respective countries but also for international relations. The early death of Roosevelt, in particular, may have exacerbated post-WWII miscommunication between America and the Soviet Union-miscommunication that may have helped precipitate the Cold War.

Sports-related brain injuries: connecting pathology to diagnosis.

Brain injuries are becoming increasingly common in athletes and represent an important diagnostic challenge. Early detection and management of brain injuries in sports are of utmost importance in preventing chronic neurological and psychiatric decline. These types of injuries incurred during sports are referred to as mild traumatic brain injuries, which represent a heterogeneous spectrum of disease. The most dramatic manifestation of chronic mild traumatic brain injuries is termed chronic traumatic encephalopathy, which is associated with profound neuropsychiatric deficits. Because chronic traumatic encephalopathy can only be diagnosed by postmortem examination, new diagnostic methodologies are needed for early detection and amelioration of disease burden. This review examines the pathology driving changes in athletes participating in high-impact sports and how this understanding can lead to innovations in neuroimaging and biomarker discovery.

Surgical outcomes of cervical spondylotic myelopathy: an analysis of a national, administrative, longitudinal database.

OBJECTIVE The authors performed a population-based analysis of national trends, costs, and outcomes associated with cervical spondylotic myelopathy (CSM) in the United States. They assessed postoperative complications, resource utilization, and predictors of costs, in this surgically treated CSM population. METHODS MarketScan data (2006-2010) were used to retrospectively analyze the complications and costs of different spine surgeries for CSM. The authors determined outcomes following anterior cervical discectomy and fusion (ACDF), posterior fusion, combined anterior/posterior fusion, and laminoplasty procedures. RESULTS The authors identified 35,962 CSM patients, comprising 5154 elderly (age ≥ 65 years) patients (mean 72.2 years, 54.9% male) and 30,808 nonelderly patients (mean 51.1 years, 49.3% male). They found an overall complication rate of 15.6% after ACDF, 29.2% after posterior fusion, 41.1% after combined anterior and posterior fusion, and 22.4% after laminoplasty. Following ACDF and posterior fusion, a significantly higher risk of complication was seen in the elderly compared with the nonelderly (reference group). The fusion level and comorbidity-adjusted ORs with 95% CIs for these groups were 1.54 (1.40-1.68) and 1.25 (1.06-1.46), respectively. In contrast, the elderly population had lower 30-day readmission rates in all 4 surgical cohorts (ACDF, 2.6%; posterior fusion, 5.3%; anterior/posterior fusion, 3.4%; and laminoplasty, 3.6%). The fusion level and comorbidity-adjusted odds ratios for 30-day readmissions for ACDF, posterior fusion, combined anterior and posterior fusion, and laminoplasty were 0.54 (0.44-0.68), 0.32 (0.24-0.44), 0.17 (0.08-0.38), and 0.39 (0.18-0.85), respectively. CONCLUSIONS The authors' analysis of the MarketScan database suggests a higher complication rate in the surgical treatment of CSM than previous national estimates. They found that elderly age (≥ 65 years) significantly increased complication risk following ACDF and posterior fusion. Elderly patients were less likely to experience a readmission within 30 days of surgery. Postoperative complication occurrence, and 30-day readmission were significant drivers of total cost within 90 days of the index surgical procedure.

Therapeutic strategies to improve drug delivery across the blood-brain barrier.

Resection of brain tumors is followed by chemotherapy and radiation to ablate remaining malignant cell populations. Targeting these populations stands to reduce tumor recurrence and offer the promise of more complete therapy. Thus, improving access to the tumor, while leaving normal brain tissue unscathed, is a critical pursuit. A central challenge in this endeavor lies in the limited delivery of therapeutics to the tumor itself. The blood-brain barrier (BBB) is responsible for much of this difficulty but also provides an essential separation from systemic circulation. Due to the BBB's physical and chemical constraints, many current therapies, from cytotoxic drugs to antibody-based proteins, cannot gain access to the tumor. This review describes the characteristics of the BBB and associated changes wrought by the presence of a tumor. Current strategies for enhancing the delivery of therapies across the BBB to the tumor will be discussed, with a distinction made between strategies that seek to disrupt the BBB and those that aim to circumvent it.

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.

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.

Quantification of cerebrospinal fluid tumor DNA in lung cancer patients with suspected leptomeningeal carcinomatosis.

Cerebrospinal fluid tumor-derived DNA (CSF-tDNA) analysis is a promising approach for monitoring the neoplastic processes of the central nervous system. We applied a lung cancer-specific sequencing panel (CAPP-Seq) to 81 CSF, blood, and tissue samples from 24 lung cancer patients who underwent lumbar puncture (LP) for suspected leptomeningeal disease (LMD). A subset of the cohort (N = 12) participated in a prospective trial of osimertinib for refractory LMD in which serial LPs were performed before and during treatment. CSF-tDNA variant allele fractions (VAFs) were significantly higher than plasma circulating tumor DNA (ctDNA) VAFs (median CSF-tDNA, 32.7%; median plasma ctDNA, 1.8%; P < 0.0001). Concentrations of tumor DNA in CSF and plasma were positively correlated (Spearman's ρ, 0.45; P = 0.03). For LMD diagnosis, cytology was 81.8% sensitive and CSF-tDNA was 91.7% sensitive. CSF-tDNA was also strongly prognostic for overall survival (HR = 7.1; P = 0.02). Among patients with progression on targeted therapy, resistance mutations, such as EGFR T790M and MET amplification, were common in peripheral blood but were rare in time-matched CSF, indicating differences in resistance mechanisms based on the anatomic compartment. In the osimertinib cohort, patients with CNS progression had increased CSF-tDNA VAFs at follow-up LP. Post-osimertinib CSF-tDNA VAF was strongly prognostic for CNS progression (HR = 6.2, P = 0.009). Detection of CSF-tDNA in lung cancer patients with suspected LMD is feasible and may have clinical utility. CSF-tDNA improves the sensitivity of LMD diagnosis, enables improved prognostication, and drives therapeutic strategies that account for spatial heterogeneity in resistance mechanisms.

Can We Use Artificial Intelligence Cluster Analysis to Identify Patients with Metastatic Breast Cancer to the Spine at Highest Risk of Postoperative Adverse Events?

OBJECTIVE: Group patients who required open surgery for metastatic breast cancer to the spine by functional level and metastatic disease characteristics to identify factors that predispose to poor outcomes. METHODS: A retrospective analysis included patients managed at 2 tertiary referral centers from 2008 to 2020. The primary outcome was a 90-day adverse event. A 2-step unsupervised cluster analysis stratified patients into cohorts using function at presentation, preoperative spine radiation, structural instability, epidural spinal cord compression (ESCC), neural deficits, and tumor location/hormone status. Comparisons were performed using χ2 test and one-way analysis of variance. RESULTS: Five patient "clusters" were identified. High function (HIGH) had thoracic metastases and an Eastern Cooperative Oncology Group (ECOG) score of 1.0 ± 0.8. Low function/irradiated (LOW + RADS) had preoperative radiation and the lowest Karnofsky scores (56.0 ± 10.6). Estrogen receptor or progesterone receptor (ER/PR) positive patients had >90% estrogen/progesterone positivity and moderate Karnofsky scores (74.0 ± 11.5). Lumbar/noncompressive (NON-COMP) had the fewest patients with ESCC grade 2 or 3 epidural disease (42.1%, P < 0.001). Low function/neurologic deficits (LOW + NEURO) had ESCC grade 2 or 3 disease and neurologic deficits. Adverse event rates were 25.0% in the HIGH group, 73.3% in LOW + RADS, 24.0% in ER/PR, 31.6% in NON-COMP, and 60.0% in LOW + NEURO (P = 0.003). CONCLUSIONS: Function at presentation, tumor hormone signature, radiation history, and epidural compression delineated postoperative trajectory. We believe our results can aid in expectation management and the identification of at-risk patients who may merit closer surveillance following surgical intervention.

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.

Assessment of Spinal Metastases Surgery Risk Stratification Tools in Breast Cancer by Molecular Subtype.

BACKGROUND: Breast cancer molecular features and modern therapies are not included in spine metastasis prediction algorithms. OBJECTIVE: To examine molecular differences and the impact of postoperative systemic therapy to improve prognosis prediction for spinal metastases surgery and aid surgical decision making. METHODS: This is a retrospective multi-institutional study of patients who underwent spine surgery for symptomatic breast cancer spine metastases from 2008 to 2021 at the Massachusetts General Hospital and Brigham and Women's Hospital. We studied overall survival, stratified by breast cancer molecular subtype, and calculated hazard ratios (HRs) adjusting for demographics, tumor characteristics, treatments, and laboratory values. We tested the performance of established models (Tokuhashi, Bauer, Skeletal Oncology Research Group, New England Spinal Metastases Score) to predict and compare all-cause. RESULTS: A total of 98 patients surgically treated for breast cancer spine metastases were identified (100% female sex; median age, 56 years [IQR, 36-84 years]). The 1-year probabilities of survival for hormone receptor positive, hormone receptor positive/human epidermal growth factor receptor 2+, human epidermal growth factor receptor 2+, and triple-negative breast cancer were 63% (45 of 71), 83% (10 of 12), 0% (0 of 3), and 12% (1 of 8), respectively ( P < .001). Patients with triple-negative breast cancer had a higher proportion of visceral metastases, brain metastases, and poor physical activity at baseline. Postoperative chemotherapy and endocrine therapy were associated with prolonged survival. The Skeletal Oncology Research Group prognostic model had the highest discrimination (area under the receiver operating characteristic, 0.77 [95% CI, 0.73-0.81]). The performance of all prognostic scores improved when preoperative molecular data and postoperative systemic treatment plans was considered. CONCLUSION: Spine metastases risk tools were able to predict prognosis at a significantly higher degree after accounting for molecular features which guide treatment response.

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.