Machine learning-based prediction of outcomes of the endoscopic endonasal approach in Cushing disease: is the future coming?

OBJECTIVE: Machine learning (ML) is an innovative method to analyze large and complex data sets. The aim of this study was to evaluate the use of ML to identify predictors of early postsurgical and long-term outcomes in patients treated for Cushing disease (CD). METHODS: All consecutive patients in our center who underwent surgery for CD through the endoscopic endonasal approach were retrospectively reviewed. Study endpoints were gross-tumor removal (GTR), postsurgical remission, and long-term control of disease. Several demographic, radiological, and histological factors were assessed as potential predictors. For ML-based modeling, data were randomly divided into 2 sets with an 80% to 20% ratio for bootstrapped training and testing, respectively. Several algorithms were tested and tuned for the area under the curve (AUC). RESULTS: The study included 151 patients. GTR was achieved in 137 patients (91%), and postsurgical hypersecretion remission was achieved in 133 patients (88%). At last follow-up, 116 patients (77%) were still in remission after surgery and in 21 patients (14%), CD was controlled with complementary treatment (overall, of 131 cases, 87% were under control at follow-up). At internal validation, the endpoints were predicted with AUCs of 0.81-1.00, accuracy of 81%-100%, and Brier scores of 0.035-0.151. Tumor size and invasiveness and histological confirmation of adrenocorticotropic hormone (ACTH)-secreting cells were the main predictors for the 3 endpoints of interest. CONCLUSIONS: ML algorithms were used to train and internally validate robust models for all the endpoints, giving accurate outcome predictions in CD cases. This analytical method seems promising for potentially improving future patient care and counseling; however, careful clinical interpretation of the results remains necessary before any clinical adoption of ML. Moreover, further studies and increased sample sizes are definitely required before the widespread adoption of ML to the study of CD.

The T2-FLAIR-mismatch sign as an imaging biomarker for IDH and 1p/19q status in diffuse low-grade gliomas: a systematic review with a Bayesian approach to evaluation of diagnostic test performance.

OBJECTIVE: With the revised WHO 2016 classification of brain tumors, there has been increasing interest in imaging biomarkers to predict molecular status and improve the yield of genetic testing for diffuse low-grade gliomas (LGGs). The T2-FLAIR-mismatch sign has been suggested to be a highly specific radiographic marker of isocitrate dehydrogenase (IDH) gene mutation and 1p/19q codeletion status in diffuse LGGs. The presence of T2-FLAIR mismatch indicates a T2-hyperintense lesion that is hypointense on FLAIR with the exception of a hyperintense rim. METHODS: In accordance with PRISMA guidelines, we performed a systematic review of the Ovid Medline, Embase, Scopus, and Cochrane databases for reports of studies evaluating the diagnostic performance of T2-FLAIR mismatch in predicting the IDH and 1p/19q codeletion status in diffuse LGGs. Results were combined into a 2 × 2 format, and the following diagnostic performance parameters were calculated: sensitivity, specificity, positive predictive value, negative predictive value, and positive (LR+) and negative (LR-) likelihood ratios. In addition, we utilized Bayes theorem to calculate posttest probabilities as a function of known pretest probabilities from previous genome-wide association studies and the calculated LRs. Calculations were performed for 1) IDH mutation with 1p/19q codeletion (IDHmut-Codel), 2) IDH mutation without 1p/19q codeletion (IDHmut-Noncodel), 3) IDH mutation overall, and 4) 1p/19q codeletion overall. The QUADAS-2 (revised Quality Assessment of Diagnostic Accuracy Studies) tool was utilized for critical appraisal of included studies. RESULTS: A total of 4 studies were included, with inclusion of 2 separate cohorts from a study reporting testing and validation (n = 746). From pooled analysis of all cohorts, the following values were obtained for each molecular profile-IDHmut-Codel: sensitivity 30%, specificity 73%, LR+ 1.1, LR- 1.0; IDHmut-Noncodel: sensitivity 33.7%, specificity 98.5%, LR+ 22.5, LR- 0.7; IDH: sensitivity 32%, specificity 100%, LR+ 32.1, LR- 0.7; 1p/19q codeletion: sensitivity 0%, specificity 54%, LR+ 0.01, LR- 1.9. Bayes theorem was used to calculate the following posttest probabilities after a positive and negative result, respectively-IDHmut-Codel: 32.2% and 29.4%; IDHmut-Noncodel: 95% and 40%; IDH: 99.2% and 73.5%; 1p/19q codeletion: 0.4% and 35.1%. CONCLUSIONS: The T2-FLAIR-mismatch sign is an insensitive but highly specific marker of IDH mutation but not 1p/19q codeletion in diffuse LGGs, although there may be significant exceptions. These findings support the utility of T2-FLAIR mismatch as an imaging-based biomarker for positive selection of patients with IDH-mutant gliomas.

Machine learning applications for the prediction of surgical site infection in neurological operations.

OBJECTIVE: Surgical site infection (SSI) following a neurosurgical operation is a complication that impacts morbidity, mortality, and economics. Currently, machine learning (ML) algorithms are used for outcome prediction in various neurosurgical aspects. The implementation of ML algorithms to learn from medical data may help in obtaining prognostic information on diseases, especially SSIs. The purpose of this study was to compare the performance of various ML models for predicting surgical infection after neurosurgical operations. METHODS: A retrospective cohort study was conducted on patients who had undergone neurosurgical operations at tertiary care hospitals between 2010 and 2017. Supervised ML algorithms, which included decision tree, naive Bayes with Laplace correction, k-nearest neighbors, and artificial neural networks, were trained and tested as binary classifiers (infection or no infection). To evaluate the ML models from the testing data set, their sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), as well as their accuracy, receiver operating characteristic curve, and area under the receiver operating characteristic curve (AUC) were analyzed. RESULTS: Data were available for 1471 patients in the study period. The SSI rate was 4.6%, and the type of SSI was superficial, deep, and organ/space in 1.2%, 0.8%, and 2.6% of cases, respectively. Using the backward stepwise method, the authors determined that the significant predictors of SSI in the multivariable Cox regression analysis were postoperative CSF leakage/subgaleal collection (HR 4.24, p < 0.001) and postoperative fever (HR 1.67, p = 0.04). Compared with other ML algorithms, the naive Bayes had the highest performance with sensitivity at 63%, specificity at 87%, PPV at 29%, NPV at 96%, and AUC at 76%. CONCLUSIONS: The naive Bayes algorithm is highlighted as an accurate ML method for predicting SSI after neurosurgical operations because of its reasonable accuracy. Thus, it can be used to effectively predict SSI in individual neurosurgical patients. Therefore, close monitoring and allocation of treatment strategies can be informed by ML predictions in general practice.

Hemodynamic and morphological characteristics of a growing cerebral aneurysm.

The growth of cerebral aneurysms is linked to local hemodynamic conditions, but the driving mechanisms of the growth are poorly understood. The goal of this study was to examine the association between intraaneurysmal hemodynamic features and areas of aneurysm growth, to present the key hemodynamic parameters essential for an accurate prediction of the growth, and to gain a deeper understanding of the underlying mechanisms. Patient-specific images of a growing cerebral aneurysm in 3 different growth stages acquired over a period of 40 months were segmented and reconstructed. A unique aspect of this patient-specific case study was that while one side of the aneurysm stayed stable, the other side continued to grow. This unique case enabled the authors to examine their aims in the same patient with parent and daughter arteries under the same inlet flow conditions. Pulsatile flow in the aneurysm models was simulated using computational fluid dynamics and was validated with in vitro experiments using particle image velocimetry measurements. The authors' detailed analysis of intrasaccular hemodynamics linked the growing regions of aneurysms to flow instabilities and complex vortex structures. Extremely low velocities were observed at or around the center of the unstable vortex structure, which matched well with the growing regions of the studied cerebral aneurysm. Furthermore, the authors observed that the aneurysm wall regions with a growth greater than 0.5 mm coincided with wall regions of lower (< 0.5 Pa) time-averaged wall shear stress (TAWSS), lower instantaneous (< 0.5 Pa) wall shear stress (WSS), and high (> 0.1) oscillatory shear index (OSI). To determine which set of parameters can best identify growing and nongrowing aneurysms, the authors performed statistical analysis for consecutive stages of the growing CA. The results demonstrated that the combination of TAWSS and the distance from the center of the vortical structure has the highest sensitivity and positive predictive value, and relatively high specificity and negative predictive value. These findings suggest that an unstable, recirculating flow structure within the aneurysm sac created in the region adjacent to the aneurysm wall with low TAWSS may be introduced as an accurate criterion to explain the hemodynamic conditions predisposing the aneurysm to growth. The authors' findings are based on one patient's data set, but the study lays out the justification for future large-scale verification. The authors' findings can assist clinicians in differentiating stable and growing aneurysms during preinterventional planning.

The use of contrast-enhanced, time-resolved magnetic resonance angiography in cerebrovascular pathology.

Digital subtraction angiography (DSA) has long been the imaging gold standard in the evaluation, treatment, and follow-up of cerebro- and spinovascular disorders. However, DSA has the disadvantages of invasiveness, contrast allergy or nephropathy, the impracticality of procedural preparation and recovery, and expense. Contrast-enhanced (CE), time-resolved (TR) magnetic resonance angiography (CE TR-MRA) is a sophisticated, relatively novel imaging modality that provides multiphasic contrast-enhanced visualization of the neurovasculature. Given the crucial role of angiography in all aspects of care for patients with complex neurovascular disorders, it is incumbent on those who care for these patients to understand the usefulness and pitfalls of novel imaging in this arena to ensure best practices, and to deliver cutting edge care to these patients in a way that minimizes cost, but does not compromise quality. CE TR-MRA has the potential to play an expanded role in the workup and follow-up across the spectrum of neurovascular disease, and this review is aimed to help neurosurgeons better understand how CE TR-MRA can be used to better manage patients in this cohort.

Machine learning-based preoperative predictive analytics for lumbar spinal stenosis.

OBJECTIVEPatient-reported outcome measures (PROMs) following decompression surgery for lumbar spinal stenosis (LSS) demonstrate considerable heterogeneity. Individualized prediction tools can provide valuable insights for shared decision-making. The authors aim to evaluate the feasibility of predicting short- and long-term PROMs, reoperations, and perioperative parameters by machine learning (ML) methods.METHODSData were derived from a prospective registry. All patients had undergone single- or multilevel mini-open facet-sparing decompression for LSS. The prediction models were trained using various ML-based algorithms to predict the endpoints of interest. Models were selected by area under the receiver operating characteristic curve (AUC). The endpoints were dichotomized by minimum clinically important difference (MCID) and included 6-week and 12-month numeric rating scales for back pain (NRS-BP) and leg pain (NRS-LP) severity and the Oswestry Disability Index (ODI), as well as prolonged surgery (> 45 minutes), extended length of hospital stay (> 28 hours), and reoperations.RESULTSA total of 635 patients were included. The average age was 62 ± 10 years, and 333 patients (52%) were male. At 6 weeks, MCID was seen in 63%, 76%, and 61% of patients for ODI, NRS-LP, and NRS-BP, respectively. At internal validation, the models predicted MCID in these variables with accuracies of 69%, 76%, and 85%, and with AUCs of 0.75, 0.79, and 0.92. At 12 months, 66%, 63%, and 51% of patients reported MCID; the observed accuracies were 62%, 74%, and 66%, with AUCs of 0.68, 0.72, and 0.79. Reoperations occurred in 60 patients (9.5%), of which 27 (4.3%) occurred at the index level. Overall and index-level reoperations were predicted with 69% and 63% accuracy, respectively, and with AUCs of 0.66 and 0.61. In 15%, a length of surgery greater than 45 minutes was observed and predicted with 78% accuracy and AUC of 0.54. Only 15% of patients were admitted to the hospital for longer than 28 hours. The developed ML-based model enabled prediction of extended hospital stay with an accuracy of 77% and AUC of 0.58.CONCLUSIONSPreoperative prediction of a range of clinically relevant endpoints in decompression surgery for LSS using ML is feasible, and may enable enhanced informed patient consent and personalized shared decision-making. Access to individualized preoperative predictive analytics for outcome and treatment risks may represent a further step in the evolution of surgical care for patients with LSS.

A comparison of digital subtraction angiography and computed tomography angiography for the diagnosis of penetrating cerebrovascular injury.

OBJECTIVE: Penetrating cerebrovascular injury (PCVI) is a subset of traumatic brain injury (TBI) comprising a broad spectrum of cerebrovascular pathology, including traumatic pseudoaneurysms, direct arterial injury, venous sinus stenosis or occlusion, and traumatic dural arteriovenous fistulas. These can result in immediate or delayed vascular injury and consequent neurological morbidity. Current TBI guidelines recommend cerebrovascular imaging for detection, but there is no consensus on the optimum modality. The aim of this retrospective cohort study was to compare CT angiography (CTA) and digital subtraction angiography (DSA) for the diagnosis of PCVI. METHODS: The records of all patients presenting to two level I trauma centers in the United States between January 2010 and July 2016 with penetrating head or neck trauma were reviewed. Only those who had undergone both CTA and DSA were included. Clinical and neuroimaging data were collected, and PCVIs were stratified using a modified Biffl grading scheme. DSA and CTA results were then compared. RESULTS: Of 312 patients with penetrating trauma over the study period, 56 patients (91% male, mean age 32 years) with PCVI met inclusion criteria and constituted the study cohort. The mechanism of injury was a gunshot wound in 86% (48/56) of patients. Twenty-four (43%) patients had sustained an angiographically confirmed arterial or venous injury. Compared with DSA as the gold standard, CTA had a sensitivity and specificity of 72% and 63%, respectively, for identifying PCVI. CTA had a positive predictive value of 61% and negative predictive value of 70%. Seven patients (13%) required immediate endovascular treatment of PCVI; in 3 (43%) of these patients, the injury was not identified on CTA. Twenty-two patients (39%) underwent delayed DSA an average of 25 days after injury; 2 (9%) of these patients were found to harbor new pathological conditions requiring treatment. CONCLUSIONS: In this retrospective analysis of PCVI at two large trauma centers, CTA demonstrated low sensitivity, specificity, and positive and negative predictive values for the diagnosis of PCVI. These findings suggest that DSA provides better accuracy than CTA in the diagnosis of both immediate and delayed PCVI and should be considered for patients experiencing penetrating head or neck trauma.

5-ALA-induced fluorescence as a marker for diagnostic tissue in stereotactic biopsies of intracranial lymphomas: experience in 41 patients.

OBJECTIVE Stereotactic needle biopsies are usually performed for histopathological confirmation of intracranial lymphomas to guide adequate treatment. During biopsy, intraoperative histopathology is an effective tool to avoid acquisition of nondiagnostic samples. In the last years, 5-aminolevulinic acid (5-ALA)-induced fluorescence has been increasingly used for visualization of diagnostic brain tumor tissue during stereotactic biopsies. Recently, visible fluorescence was reported in the first cases of intracranial lymphomas as well. The aim of this study is thus to investigate the technical and clinical utility of 5-ALA-induced fluorescence in a large series of stereotactic biopsies for intracranial lymphoma. METHODS This prospective study recruited adult patients who underwent frameless stereotactic needle biopsy for a radiologically suspected intracranial lymphoma after oral 5-ALA administration. During biopsy, samples from the tumor region were collected for histopathological analysis, and presence of fluorescence (strong, vague, or no fluorescence) was assessed with a modified neurosurgical microscope. In tumors with available biopsy samples from at least 2 different regions the intratumoral fluorescence homogeneity was additionally investigated. Furthermore, the influence of potential preoperative corticosteroid treatment or immunosuppression on fluorescence was analyzed. Histopathological tumor diagnosis was established and all collected biopsy samples were screened for diagnostic lymphoma tissue. RESULTS The final study cohort included 41 patients with intracranial lymphoma. Stereotactic biopsies with assistance of 5-ALA were technically feasible in all cases. Strong fluorescence was found as maximum level in 30 patients (75%), vague fluorescence in 2 patients (4%), and no visible fluorescence in 9 patients (21%). In 28 cases, samples were obtained from at least 2 different tumor regions; homogenous intratumoral fluorescence was found in 16 of those cases (57%) and inhomogeneous intratumoral fluorescence in 12 (43%). According to histopathological analysis, all samples with strong or vague fluorescence contained diagnostic lymphoma tissue, resulting in a positive predictive value of 100%. Analysis showed no influence of preoperative corticosteroids or immunosuppression on fluorescence. CONCLUSIONS The data obtained in this study demonstrate the technical and clinical utility of 5-ALA-induced fluorescence in stereotactic biopsies of intracranial lymphomas. Thus, 5-ALA can serve as a useful tool to select patients not requiring intraoperative histopathology, and its application should markedly reduce operation time and related costs in the future.

Utility of deep neural networks in predicting gross-total resection after transsphenoidal surgery for pituitary adenoma: a pilot study.

OBJECTIVEGross-total resection (GTR) is often the primary surgical goal in transsphenoidal surgery for pituitary adenoma. Existing classifications are effective at predicting GTR but are often hampered by limited discriminatory ability in moderate cases and by poor interrater agreement. Deep learning, a subset of machine learning, has recently established itself as highly effective in forecasting medical outcomes. In this pilot study, the authors aimed to evaluate the utility of using deep learning to predict GTR after transsphenoidal surgery for pituitary adenoma.METHODSData from a prospective registry were used. The authors trained a deep neural network to predict GTR from 16 preoperatively available radiological and procedural variables. Class imbalance adjustment, cross-validation, and random dropout were applied to prevent overfitting and ensure robustness of the predictive model. The authors subsequently compared the deep learning model to a conventional logistic regression model and to the Knosp classification as a gold standard.RESULTSOverall, 140 patients who underwent endoscopic transsphenoidal surgery were included. GTR was achieved in 95 patients (68%), with a mean extent of resection of 96.8% ± 10.6%. Intraoperative high-field MRI was used in 116 (83%) procedures. The deep learning model achieved excellent area under the curve (AUC; 0.96), accuracy (91%), sensitivity (94%), and specificity (89%). This represents an improvement in comparison with the Knosp classification (AUC: 0.87, accuracy: 81%, sensitivity: 92%, specificity: 70%) and a statistically significant improvement in comparison with logistic regression (AUC: 0.86, accuracy: 82%, sensitivity: 81%, specificity: 83%) (all p < 0.001).CONCLUSIONSIn this pilot study, the authors demonstrated the utility of applying deep learning to preoperatively predict the likelihood of GTR with excellent performance. Further training and validation in a prospective multicentric cohort will enable the development of an easy-to-use interface for use in clinical practice.

Relative lumbar lordosis and lordosis distribution index: individualized pelvic incidence-based proportional parameters that quantify lumbar lordosis more precisely than the concept of pelvic incidence minus lumbar lordosis.

OBJECTIVE The subtraction of lumbar lordosis (LL) from the pelvic incidence (PI) offers an estimate of the LL required for a given PI value. Relative LL (RLL) and the lordosis distribution index (LDI) are PI-based individualized measures. RLL quantifies the magnitude of lordosis relative to the ideal lordosis as defined by the magnitude of PI. LDI defines the magnitude of lower arc lordosis in proportion to total lordosis. The aim of this study was to compare RLL and PI - LL for their ability to predict postoperative complications and their correlations with health-related quality of life (HRQOL) scores. METHODS Inclusion criteria were ≥ 4 levels of fusion and ≥ 2 years of follow-up. Mechanical complications were proximal junctional kyphosis/proximal junctional failure, distal junctional kyphosis/distal junctional failure, rod breakage, and implant-related complications. Correlations between PI - LL, RLL, PI, and HRQOL were analyzed using the Pearson correlation coefficient. Mechanical complication rates in PI - LL, RLL, LDI, RLL, and LDI interpreted together, and RLL subgroups for each PI - LL category were compared using chi-square tests and the exact test. Predictive models for mechanical complications with RLL and PI - LL were analyzed using binomial logistic regressions. RESULTS Two hundred twenty-two patients (168 women, 54 men) were included. The mean age was 52.2 ± 19.3 years (range 18-84 years). The mean follow-up was 28.8 ± 8.2 months (range 24-62 months). There was a significant correlation between PI - LL and PI (r = 0.441, p < 0.001), threatening the use of PI - LL to quantify spinopelvic mismatch for different PI values. RLL was not correlated with PI (r = -0.093, p > 0.05); therefore, it was able to quantify divergence from ideal lordosis for all PI values. Compared with PI - LL, RLL had stronger correlations with HRQOL scores (p < 0.05). Discrimination performance was better for the model with RLL than for PI - LL. The agreement between RLL and PI - LL was high (κ = 0.943, p < 0.001), moderate (κ = 0.455, p < 0.001), and poor (κ = -0.154, p = 0.343), respectively, for large, average, and small PI sizes. When analyzed by RLL, each PI - LL category was further divided into distinct groups of patients who had different mechanical complication rates (p < 0.001). CONCLUSIONS Using the formula of PI - LL may be insufficient to quantify normolordosis for the whole spectrum of PI values when applied as an absolute numeric value in conjunction with previously reported population-based average thresholds of 10° and 20°. Schwab PI - LL groups were found to constitute an inhomogeneous group of patients. RLL offers an individualized quantification of LL for all PI sizes. Compared with PI - LL, RLL showed a greater association with both mechanical complications and HRQOL. The use of RLL and LDI together, instead of PI - LL, for surgical planning may result in lower mechanical complication rates and better long-term HRQOL.

Factors triggering an additional resection and determining residual tumor volume on intraoperative MRI: analysis from a prospective single-center registry of supratentorial gliomas.

OBJECTIVE: In this analysis, the authors sought to identify variables triggering an additional resection (AR) and determining residual intraoperative tumor volume in 1.5-T intraoperative MRI (iMRI)-guided glioma resections. METHODS: A consecutive case series of 224 supratentorial glioma resections (WHO Grades I-IV) from a prospective iMRI registry (inclusion dates January 2011-April 2013) was examined with univariate and multiple regression models including volumetric data, tumor-related, and surgeon-related factors. The surgeon's expectation of an AR, in response to a questionnaire completed prior to iMRI, was evaluated using contingency analysis. A machine-learning prediction model was applied to consider if anticipation of intraoperative findings permits preoperative identification of ideal iMRI cases. RESULTS: An AR was performed in 70% of cases after iMRI, but did not translate into an accumulated risk for neurological morbidity after surgery (p = 0.77 for deficits in cases with AR vs no AR). New severe persistent deficits occurred in 6.7% of patients. Initial tumor volume determined frequency of ARs and was independently correlated with larger tumor remnants delineated on iMRI (p < 0.0001). Larger iMRI volume was further associated with eloquent location (p = 0.010) and recurrent tumors (p < 0.0001), and with WHO grade (p = 0.0113). Greater surgical experience had no significant influence on the course of surgery. The surgeon's capability of ruling out an AR prior to iMRI turned out to incorporate guesswork (negative predictive value 43.6%). In a prediction model, AR could only be anticipated with 65% accuracy after integration of confounding variables. CONCLUSIONS: Routine use of iMRI in glioma surgery is a safe and reliable method for resection guidance and is characterized by frequent ARs after scanning. Tumor-related factors were identified that influenced the course of surgery and intraoperative decision-making, and iMRI had a common value for surgeons of all experience levels. Commonly, the subjective intraoperative impression of the extent of resection had to be revised after iMRI review, which underscores the manifold potential of iMRI guidance. In combination with the failure to identify ideal iMRI cases preoperatively, this study supports a generous, tumor-oriented rather than surgeon-oriented indication for iMRI in glioma surgery.

Intraoperative neurophysiological mapping and monitoring in spinal tumor surgery: sirens or indispensable tools?

Spinal tumor (ST) surgery carries the risk of new neurological deficits in the postoperative period. Intraoperative neurophysiological monitoring and mapping (IONM) represents an effective method of identifying and monitoring in real time the functional integrity of both the spinal cord (SC) and the nerve roots (NRs). Despite consensus favoring the use of IONM in ST surgery, in this era of evidence-based medicine, there is still a need to demonstrate the effective role of IONM in ST surgery in achieving an oncological cure, optimizing patient safety, and considering medicolegal aspects. Thus, neurosurgeons are asked to establish which techniques are considered indispensable. In the present study, the authors focused on the rationale for and the accuracy (sensitivity, specificity, and positive and negative predictive values) of IONM in ST surgery in light of more recent evidence in the literature, with specific emphasis on the role of IONM in reducing the incidence of postoperative neurological deficits. This review confirms the role of IONM as a useful tool in the workup for ST surgery. Individual monitoring and mapping techniques are clearly not sufficient to account for the complex function of the SC and NRs. Conversely, multimodal IONM is highly sensitive and specific for anticipating neurological injury during ST surgery and represents an important tool for preserving neuronal structures and achieving an optimal postoperative functional outcome.

Visualization of nerve fibers and their relationship to peripheral nerve tumors by diffusion tensor imaging.

OBJECT The majority of growing and/or symptomatic peripheral nerve tumors are schwannomas and neurofibromas. They are almost always benign and can usually be resected while minimizing motor and sensory deficits if approached with the proper expertise and techniques. Intraoperative electrophysiological stimulation and recording techniques allow the surgeon to map the surface of the tumor in an effort to identify and thus avoid damaging functioning nerve fibers. Recently, MR diffusion tensor imaging (DTI) techniques have permitted the visualization of axons, because of their anisotropic properties, in peripheral nerves. The object of this study was to compare the distribution of nerve fibers as revealed by direct electrical stimulation with that seen on preoperative MR DTI. METHODS The authors conducted a retrospective chart review of patients with a peripheral nerve or nerve root tumor between March 2012 and January 2014. Diffusion tensor imaging and intraoperative data had been prospectively collected for patients with peripheral nerve tumors that were resected. Preoperative identification of the nerve fiber location in relation to the nerve tumor surface as seen on DTI studies was compared with the nerve fiber's intraoperative localization using electrophysiological stimulation and recordings. RESULTS In 23 patients eligible for study there was good correlation between nerve fiber location on DTI and its anatomical location seen intraoperatively. Diffusion tensor imaging demonstrated the relationship of nerve fibers relative to the tumor with 95.7% sensitivity, 66.7% specificity, 75% positive predictive value, and 93.8% negative predictive value. CONCLUSIONS Preoperative DTI techniques are useful in helping the peripheral nerve surgeon to both determine the risks involved in resecting a nerve tumor and plan the safest surgical approach.

Diagnostic accuracy of CT scan-based criteria compared with surgical exploration for the analysis of cervical fusion and nonunion.

OBJECTIVE: Computed tomography (CT) scans are accepted as the imaging standard of reference to define union after anterior cervical discectomy and fusion (ACDF). However, ideal CT criteria to diagnose union have not been identified or validated. The objective of this study was to analyze the diagnostic value of 9 CT-based criteria and identify the ideal criteria among them to assess cervical fusion after ACDF using surgical exploration as the standard of reference. METHODS: The authors performed a retrospective radiographic study of a single surgeon's prospective assessment of osseous fusion during cervical revision surgery by analyzing complete radiographic data in 44 patients who underwent anterior cervical revision surgery due to symptomatic suspected nonunion or adjacent level disease. All patients received standard preoperative CT scans, which were assessed by an independent radiologist to evaluate 9 diagnostic criteria for osseous union. During revision surgery, scar tissue was removed and manual segmental translation tests were performed. Nonunion was defined by visualized motion at the treated ACDF level. RESULTS: In total, 44 patients were included in the study (30 men; patient age 54 ± 6 years, BMI 28 ± 5 kg/m2). For analysis of fusion, 75 cervical levels were explored, of which 61 levels (81%) showed intraoperative movement indicating nonunion. Statistical analysis showed that of the 9 parameters used to diagnose bone union, "bridging bone on ≥ 3 CT slices" yielded the highest sensitivity (100%) and specificity (58%). Multivariate analysis revealed that prediction accuracy was not increased if several criteria were combined to determine fusion. CONCLUSIONS: The authors found that the best indicator of bone union was the item bridging bone on ≥ 3 CT slices. Combining the scoring of more than one criterion did not increase the diagnostic accuracy.

Preoperative computed tomography perfusion in pediatric moyamoya disease: a single-institution experience.

OBJECTIVE: Moyamoya disease is a progressive occlusive arteriopathy for which surgical revascularization is indicated. In this retrospective study, the authors investigated the use of preoperative CT perfusion with the aim of establishing pathological data references. METHODS: The authors reviewed the medical records of children with moyamoya disease treated surgically at one institution between 2016 and 2019. Preoperative CT perfusion studies were used to quantify mean transit time (MTT), cerebral blood volume (CBV), cerebral blood flow (CBF), and time to peak (TTP) for the anterior, middle, and posterior cerebral artery vascular territories for each patient. CT perfusion parameter ratios (diseased/healthy hemispheres) and absolute differences were compared between diseased and normal vascular territories (defined by catheter angiography studies). Sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values for CT perfusion parameters for severe angiographic moyamoya were calculated. RESULTS: Nine children (89% female) had preoperative CT perfusion data; 5 of them had evidence of unilateral hemispheric disease and 4 had bilateral disease. The mean age at revascularization was 77 months (range 40-144 months). The etiology of disease was neurofibromatosis type 1 (3 patients), Down syndrome (2), primary moyamoya disease (2), cerebral proliferative angiopathy (1), and sickle cell disease (1). Five patients had undergone unilateral revascularization. Among these patients, pathological vascular territories demonstrated increased MTT in 66% of samples, increased TTP in 66%, decreased CBF in 47%, and increased CBV in 87%. Severe moyamoya (Suzuki stage ≥ 4) had diseased/healthy ratios ≥ 1 for MTT in 78% of cases, for TTP in 89%, for CBF in 67%, and for CBV in 89%. The MTT and TTP region of interest ratio ≥ 1 demonstrated 89% sensitivity, 67% specificity, 80% PPV, and 80% NPV for the prediction of severe angiographic moyamoya disease. CONCLUSIONS: Pathological hemispheres in these children with moyamoya disease demonstrated increased MTT, TTP, and CBV and decreased CBF. The authors' results suggest that preoperative CT perfusion may, with high sensitivity, be useful in deciphering perfusion mismatch in brain tissue in children with moyamoya disease. More severe angiographic disease displays a more distinct correlation, allowing surgeons to recognize when to intervene in these patients.

Utility of neuromonitoring during lumbar pedicle subtraction osteotomy for adult spinal deformity.

OBJECTIVE: The benefits and utility of routine neuromonitoring with motor and somatosensory evoked potentials during lumbar spine surgery remain unclear. This study assesses measures of performance and utility of transcranial motor evoked potentials (MEPs) during lumbar pedicle subtraction osteotomy (PSO). METHODS: This is a retrospective study of a single-surgeon cohort of consecutive adult spinal deformity (ASD) patients who underwent lumbar PSO from 2006 to 2016. A blinded neurophysiologist reviewed individual cases for MEP changes. Multivariate analysis was performed to determine whether changes correlated with neurological deficits. Measures of performance were calculated. RESULTS: A total of 242 lumbar PSO cases were included. MEP changes occurred in 38 (15.7%) cases; the changes were transient in 21 cases (55.3%) and permanent in 17 (44.7%). Of the patients with permanent changes, 9 (52.9%) had no recovery and 8 (47.1%) had partial recovery of MEP signals. Changes occurred at a mean time of 8.8 minutes following PSO closure (range: during closure to 55 minutes after closure). The mean percentage of MEP signal loss was 72.9%. The overall complication rate was 25.2%, and the incidence of new neurological deficits was 4.1%. On multivariate analysis, MEP signal loss of at least 50% was not associated with complication (p = 0.495) or able to predict postoperative neurological deficits (p = 0.429). Of the 38 cases in which MEP changes were observed, the observation represented a true-positive finding in only 3 cases. Postoperative neurological deficits without MEP changes occurred in 7 cases. Calculated measures of performance were as follows: sensitivity 30.0%, specificity 84.9%, positive predictive value 7.9%, and negative predictive value 96.6%. Regarding the specific characteristics of the MEP changes, only a signal loss of 80% or greater was significantly associated with a higher rate of neurological deficit (23.0% vs 0.0% for loss of less than 80%, p = 0.021); changes of less than 80% were not associated with postoperative deficits. CONCLUSIONS: Neuromonitoring has a low positive predictive value and low sensitivity for detecting new neurological deficits. Even when neuromonitoring is unchanged, patients can still have new neurological deficits. The utility of transcranial MEP monitoring for lumbar PSO remains unclear but there may be advantages to its use.

Surgical treatment of meningiomas located in the rolandic area: the role of navigated transcranial magnetic stimulation for preoperative planning, surgical strategy, and prediction of arachnoidal cleavage and motor outcome.

OBJECTIVE: Surgical treatment of convexity meningiomas is usually considered a low-risk procedure. Nevertheless, the risk of postoperative motor deficits is higher (7.1%-24.7% of all cases) for lesions located in the rolandic region, especially when an arachnoidal cleavage plane with the motor pathway is not identifiable. The authors analyzed the possible role of navigated transcranial magnetic stimulation (nTMS) for planning resection of rolandic meningiomas and predicting the presence or lack of an intraoperative arachnoidal cleavage plane as well as the postoperative motor outcome. METHODS: Clinical data were retrospectively collected from surgical cases involving patients affected by convexity, parasagittal, or falx meningiomas involving the rolandic region, who received preoperative nTMS mapping of the motor cortex (M1) and nTMS-based diffusion tensor imaging (DTI) fiber tracking of the corticospinal tract before surgery at 2 different neurosurgical centers. Surgeons' self-reported evaluation of the impact of nTMS-based mapping on surgical strategy was analyzed. Moreover, the nTMS mapping accuracy was evaluated in comparison with intraoperative neurophysiological mapping (IONM). Lastly, we assessed the role of nTMS as well as other pre- and intraoperative parameters for predicting the patients' motor outcome and the presence or absence of an intraoperative arachnoidal cleavage plane. RESULTS: Forty-seven patients were included in this study. The nTMS-based planning was considered useful in 89.3% of cases, and a change of the surgical strategy was observed in 42.5% of cases. The agreement of nTMS-based planning and IONM-based strategy in 35 patients was 94.2%. A new permanent motor deficit occurred in 8.5% of cases (4 of 47). A higher resting motor threshold (RMT) and the lack of an intraoperative arachnoidal cleavage plane were the only independent predictors of a poor motor outcome (p = 0.04 and p = 0.02, respectively). Moreover, a higher RMT and perilesional edema also predicted the lack of an arachnoidal cleavage plane (p = 0.01 and p = 0.03, respectively). Preoperative motor status, T2 cleft sign, contrast-enhancement pattern, and tumor volume had no predictive value. CONCLUSIONS: nTMS-based motor mapping is a useful tool for presurgical assessment of rolandic meningiomas, especially when a clear cleavage plane with M1 is not present. Moreover, the RMT can indicate the presence or absence of an intraoperative cleavage plane and predict the motor outcome, thereby helping to identify high-risk patients before surgery.

Intraoperative neurophysiological monitoring of the bulbocavernosus reflex during surgery for conus spinal lipoma: what are the warning criteria?

OBJECTIVEDespite the surge in the intraoperative use of the bulbocavernosus reflex (BCR) during lumbosacral surgeries, there are as yet no widely accepted BCR warning criteria for use with intraoperative neurophysiological monitoring (IONM). The author's aim was to find clinically acceptable warning criteria for use in IONM of the BCR.METHODSRecords of IONM of the BCR in 164 operations in 163 patients (median age 5 months) with a conus spinal lipoma who underwent surgery between August 2002 and May 2016 were retrospectively analyzed. The outcomes of IONM of the BCR were grouped by the residual amplitude at the end of surgery: group 1, ≥ 50%; group 2, 25%-50% (including the lower bound, but not the upper); and group 3, < 25%. Cases in which the BCR was lost were separately assessed as a subgroup of group 3. The postoperative urinary complication rate was used to verify the warning criteria zones.RESULTSThe BCR could be monitored in 149 surgeries (90.9%). There were 118 surgeries (79.2%) in group 1, 18 (12.1%) in group 2, and 13 (8.7%) in group 3. Two surgeries (11.1%) in group 2 and 6 (46.2%) in group 3 resulted in urinary complications. In the group 3 subgroup (lost BCR), all 5 surgeries resulted in urinary complications. The cutoff value of the BCR amplitude reduction was placed between groups 1 and 2 (zone 1: cutoff 50%), groups 2 and 3 (zone 2: cutoff 25%), and group 3 and its subgroup (zone 3: cutoff zero, present or lost). In zone 1, the positive predictive value (PPV) was 25.8% and the negative predictive value (NPV) was 100%. In zone 2, the PPV was 53.8% and the NPV 98.5%. In zone 3, the PPV was 100% and the NPV 97.9%. The PPV was highest in zone 3. The NPV was highest in zone 1, but its PPV was low (25.8%).CONCLUSIONSThe "lost or remained" criterion of BCR amplitude (zone 3: cutoff zero) can be used as a predictor of postoperative urinary function. As a warning criterion, the cutoff value of the BCR amplitude reduction at 75% (zone 2) may be used. This preliminary clinical report on the warning criteria for the BCR may contribute to improving the safety of surgery for conus spinal lipoma.

Predicting nonroutine discharge after elective spine surgery: external validation of machine learning algorithms.

OBJECTIVE: Nonroutine discharge after elective spine surgery increases healthcare costs, negatively impacts patient satisfaction, and exposes patients to additional hospital-acquired complications. Therefore, prediction of nonroutine discharge in this population may improve clinical management. The authors previously developed a machine learning algorithm from national data that predicts risk of nonhome discharge for patients undergoing surgery for lumbar disc disorders. In this paper the authors externally validate their algorithm in an independent institutional population of neurosurgical spine patients. METHODS: Medical records from elective inpatient surgery for lumbar disc herniation or degeneration in the Transitional Care Program at Brigham and Women's Hospital (2013-2015) were retrospectively reviewed. Variables included age, sex, BMI, American Society of Anesthesiologists (ASA) class, preoperative functional status, number of fusion levels, comorbidities, preoperative laboratory values, and discharge disposition. Nonroutine discharge was defined as postoperative discharge to any setting other than home. The discrimination (c-statistic), calibration, and positive and negative predictive values (PPVs and NPVs) of the algorithm were assessed in the institutional sample. RESULTS: Overall, 144 patients underwent elective inpatient surgery for lumbar disc disorders with a nonroutine discharge rate of 6.9% (n = 10). The median patient age was 50 years and 45.1% of patients were female. Most patients were ASA class II (66.0%), had 1 or 2 levels fused (80.6%), and had no diabetes (91.7%). The median hematocrit level was 41.2%. The neural network algorithm generalized well to the institutional data, with a c-statistic (area under the receiver operating characteristic curve) of 0.89, calibration slope of 1.09, and calibration intercept of -0.08. At a threshold of 0.25, the PPV was 0.50 and the NPV was 0.97. CONCLUSIONS: This institutional external validation of a previously developed machine learning algorithm suggests a reliable method for identifying patients with lumbar disc disorder at risk for nonroutine discharge. Performance in the institutional cohort was comparable to performance in the derivation cohort and represents an improved predictive value over clinician intuition. This finding substantiates initial use of this algorithm in clinical practice. This tool may be used by multidisciplinary teams of case managers and spine surgeons to strategically invest additional time and resources into postoperative plans for this population.

The value of visible 5-ALA fluorescence and quantitative protoporphyrin IX analysis for improved surgery of suspected low-grade gliomas.

OBJECTIVE: In patients with suspected diffusely infiltrating low-grade gliomas (LGG), the prognosis is dependent especially on extent of resection and precision of tissue sampling. Unfortunately, visible 5-aminolevulinic acid (5-ALA) fluorescence is usually only present in high-grade gliomas (HGGs), and most LGGs cannot be visualized. Recently, spectroscopic probes were introduced allowing in vivo quantitative analysis of intratumoral 5-ALA-induced protoporphyrin IX (PpIX) accumulation. The aim of this study was to intraoperatively investigate the value of visible 5-ALA fluorescence and quantitative PpIX analysis in suspected diffusely infiltrating LGG. METHODS: Patients with radiologically suspected diffusely infiltrating LGG were prospectively recruited, and 5-ALA was preoperatively administered. During resection, visual fluorescence and absolute tissue PpIX concentration (CPpIX) measured by a spectroscopic handheld probe were determined in different intratumoral areas. Subsequently, corresponding tissue samples were safely collected for histopathological analysis. Tumor diagnosis was established according to the World Health Organization 2016 criteria. Additionally, the tumor grade and percentage of tumor cells were investigated in each sample. RESULTS: All together, 69 samples were collected from 22 patients with histopathologically confirmed diffusely infiltrating glioma. Visible fluorescence was detected in focal areas in most HGGs (79%), but in none of the 8 LGGs. The mean CPpIX was significantly higher in fluorescing samples than in nonfluorescing samples (0.693 µg/ml and 0.008 µg/ml, respectively; p < 0.001). A significantly higher mean percentage of tumor cells was found in samples with visible fluorescence compared to samples with no fluorescence (62% and 34%, respectively; p = 0.005), and significant correlation of CPpIX and percentage of tumor cells was found (r = 0.362, p = 0.002). Moreover, high-grade histology was significantly more common in fluorescing samples than in nonfluorescing samples (p = 0.001), whereas no statistically significant difference in mean CPpIX was noted between HGG and LGG samples. Correlation between maximum CPpIX and overall tumor grade was highly significant (p = 0.005). Finally, 14 (40%) of 35 tumor samples with no visible fluorescence and 16 (50%) of 32 LGG samples showed significantly increased CPpIX (cutoff value: 0.005 µg/ml). CONCLUSIONS: Visible 5-ALA fluorescence is able to detect focal intratumoral areas of malignant transformation, and additional quantitative PpIX analysis is especially useful to visualize mainly LGG tissue that usually remains undetected by conventional fluorescence. Thus, both techniques will support the neurosurgeon in achieving maximal safe resection and increased precision of tissue sampling during surgery for suspected LGG.Clinical trial registration no.: NCT01116661 (clinicaltrials.gov).