Velociraptor: Cross-Platform Quantitative Search Using Hallmark Cell Features.

A key challenge for single cell discovery analysis is to identify new cell types, describe them quantitatively, and seek these novel cells in new studies often using a different platform. Over the last decade, tools were developed to address identification and quantitative description of cells in human tissues and tumors. However, automated validation of populations at the single cell level has struggled due to the cytometry field's reliance on hierarchical, ordered use of features and on platform-specific rules for data processing and analysis. Here we present Velociraptor, a workflow that implements Marker Enrichment Modeling in three cross-platform modules: 1) identification of cells specific to disease states, 2) description of hallmark features for each cell and population, and 3) searching for cells matching one or more hallmark feature sets in a new dataset. A key advance is that Velociraptor registers cells between datasets, including between flow cytometry and quantitative imaging using different, overlapping feature sets. Four datasets were used to challenge Velociraptor and reveal new biological insights. Working at the individual sample level, Velociraptor tracked the abundance of clinically significant glioblastoma brain tumor cell subsets and characterized the cells that predominate in recurrent tumors as a close match for rare, negative prognostic cells originally observed in matched pre-treatment tumors. In patients with inborn errors of immunity, Velociraptor identified genotype-specific cells associated with GATA2 haploinsufficiency. Finally, in cross-platform analysis of immune cells in multiplex imaging of breast cancer, Velociraptor sought and correctly identified memory T cell subsets in tumors. Different phenotypic descriptions generated by algorithms or humans were shown to be effective as search inputs, indicating that cell identity need not be described in terms of per-feature cutoffs or strict hierarchical analyses. Velociraptor thus identifies cells based on hallmark feature sets, such as protein expression signatures, and works effectively with data from multiple sources, including suspension flow cytometry, imaging, and search text based on known or theoretical cell features.

IL-8 Instructs Macrophage Identity in Lateral Ventricle Contacting Glioblastoma.

Adult IDH-wildtype glioblastoma (GBM) is a highly aggressive brain tumor with no established immunotherapy or targeted therapy. Recently, CD32+ HLA-DRhi macrophages were shown to have displaced resident microglia in GBM tumors that contact the lateral ventricle stem cell niche. Since these lateral ventricle contacting GBM tumors have especially poor outcomes, identifying the origin and role of these CD32+ macrophages is likely critical to developing successful GBM immunotherapies. Here, we identify these CD32+ cells as M_IL-8 macrophages and establish that IL-8 is sufficient and necessary for tumor cells to instruct healthy macrophages into CD32+ M_IL-8 M2 macrophages. In ex vivo experiments with conditioned medium from primary human tumor cells, inhibitory antibodies to IL-8 blocked the generation of CD32+ M_IL-8 cells. Finally, using a set of 73 GBM tumors, IL-8 protein is shown to be present in GBM tumor cells in vivo and especially common in tumors contacting the lateral ventricle. These results provide a mechanistic origin for CD32+ macrophages that predominate in the microenvironment of the most aggressive GBM tumors. IL-8 and CD32+ macrophages should now be explored as targets in combination with GBM immunotherapies, especially for patients whose tumors present with radiographic contact with the ventricular-subventricular zone stem cell niche.

An immunosuppressed microenvironment distinguishes lateral ventricle-contacting glioblastomas.

Radiographic contact of glioblastoma (GBM) tumors with the lateral ventricle and adjacent stem cell niche correlates with poor patient prognosis, but the cellular basis of this difference is unclear. Here, we reveal and functionally characterize distinct immune microenvironments that predominate in subtypes of GBM distinguished by proximity to the lateral ventricle. Mass cytometry analysis of isocitrate dehydrogenase wild-type human tumors identified elevated T cell checkpoint receptor expression and greater abundance of a specific CD32+CD44+HLA-DRhi macrophage population in ventricle-contacting GBM. Multiple computational analysis approaches, phospho-specific cytometry, and focal resection of GBMs validated and extended these findings. Phospho-flow quantified cytokine-induced immune cell signaling in ventricle-contacting GBM, revealing differential signaling between GBM subtypes. Subregion analysis within a given tumor supported initial findings and revealed intratumor compartmentalization of T cell memory and exhaustion phenotypes within GBM subtypes. Collectively, these results characterize immunotherapeutically targetable features of macrophages and suppressed lymphocytes in GBMs defined by MRI-detectable lateral ventricle contact.

Unsupervised machine learning reveals risk stratifying glioblastoma tumor cells.

A goal of cancer research is to reveal cell subsets linked to continuous clinical outcomes to generate new therapeutic and biomarker hypotheses. We introduce a machine learning algorithm, Risk Assessment Population IDentification (RAPID), that is unsupervised and automated, identifies phenotypically distinct cell populations, and determines whether these populations stratify patient survival. With a pilot mass cytometry dataset of 2 million cells from 28 glioblastomas, RAPID identified tumor cells whose abundance independently and continuously stratified patient survival. Statistical validation within the workflow included repeated runs of stochastic steps and cell subsampling. Biological validation used an orthogonal platform, immunohistochemistry, and a larger cohort of 73 glioblastoma patients to confirm the findings from the pilot cohort. RAPID was also validated to find known risk stratifying cells and features using published data from blood cancer. Thus, RAPID provides an automated, unsupervised approach for finding statistically and biologically significant cells using cytometry data from patient samples.

Mechatronic Design of a Two-Arm Concentric Tube Robot System for Rigid Neuroendoscopy.

Open surgical approaches are still often employed in neurosurgery, despite the availability of neuroendoscopic approaches that reduce invasiveness. The challenge of maneuvering instruments at the tip of the endoscope makes neuroendoscopy demanding for the physician. The only way to aim tools passed through endoscope ports is to tilt the entire endoscope; but, tilting compresses brain tissue through which the endoscope passes and can damage it. Concentric tube robots can provide necessary dexterity without endoscope tilting, while passing through existing ports in the endoscope and carrying surgical tools in their inner lumen. In this paper we describe the mechatronic design of a new concentric tube robot that can deploy two concentric tube manipulators through a standard neuroendoscope. The robot uses a compact differential drive and features embedded motor control electronics and redundant position sensors for safety. In addition to the mechatronic design of this system, this paper contributes experimental validation in the context of colloid cyst removal, comparing our new robotic system to standard manual endoscopy in a brain phantom. The robotic approach essentially eliminated endoscope tilt during the procedure (17.09° for the manual approach vs. 1.16° for the robotic system). The robotic system also enables a single surgeon to perform the procedure - typically in a manual approach one surgeon aims the endoscope and another operates the tools delivered through its ports.

A Retrospective Cohort Study of Longitudinal Audiologic Assessment in Single and Fractionated Stereotactic Radiosurgery for Vestibular Schwannoma.

BACKGROUND: Fractionated stereotactic radiosurgery (SRS) for vestibular schwannomas (VS) has been theorized to allow for tumor control with higher rates of hearing preservation in selected patients with useful hearing. However, there is a paucity of literature with formal audiologic measures of hearing preservation to support the standard use of fractionated SRS in VS. We hypothesized that fractionation would diminish the amount of hearing damage. OBJECTIVE: To evaluate the relationship between audiologic performance and SRS fractionation scheme. METHODS: We performed an IRB-approved retrospective review of patients treated with 1, 3, or 5 fraction SRS for VS at our institution from 1998 to 2016. Pre- and post-SRS audiograms with speech awareness threshold (SAT) in treated and contralateral ears were obtained. Contralateral ear measurements were used for hearing normalization to account for presbycusis. RESULTS: Fifty-six patients with median audiologic follow-up 2.0 yr (mean 2.66 yr, min-max 0.50-9.45 yr) were included. Patients treated with single fractionation had a significantly worsened SAT (dB) compared to patients treated with 5 fractions (P = .008) and compared to all multifraction patients (P = .009) at 12 to 24 mo follow-up. CONCLUSION: This retrospective analysis supports the use of fractionated SRS to preserve hearing in patients with VS. SAT can be used as an objective metric of hearing response to radiosurgery.

MR-conditional steerable needle robot for intracerebral hemorrhage removal.

BACKGROUND: Intracerebral hemorrhage (ICH) is one of the deadliest forms of stroke in the USA. Conventional surgical techniques such as craniotomy or stereotactic aspiration disrupt a large volume of healthy brain tissue in their attempts to reach the surgical site. Consequently, the surviving patients suffer from debilitating complications. METHODS: We fabricated a novel MR-conditional steerable needle robot for ICH treatment. The robot system is powered by a custom-designed high power and low-cost pneumatic motor. We tested the robot's targeting accuracy and MR-conditionality performance, and performed phantom evacuation experiment under MR image guidance. RESULTS: Experiments demonstrate that the robotic hardware is MR-conditional; the robot has the targeting accuracy of 1.26 ± 1.22 mm in bench-top tests. With real-time MRI guidance, the robot successfully reached the desired target and evacuated an 11.3 ml phantom hematoma in 9 min. CONCLUSION: MRI-guided steerable needle robotic system is a potentially feasible approach for ICH treatment by providing accurate needle guidance and intraoperative surgical outcome evaluation.

Using a Guided Machine Learning Ensemble Model to Predict Discharge Disposition following Meningioma Resection.

Objective Machine learning (ML) algorithms are powerful tools for predicting patient outcomes. This study pilots a novel approach to algorithm selection and model creation using prediction of discharge disposition following meningioma resection as a proof of concept. Materials and Methods A diversity of ML algorithms were trained on a single-institution database of meningioma patients to predict discharge disposition. Algorithms were ranked by predictive power and top performers were combined to create an ensemble model. The final ensemble was internally validated on never-before-seen data to demonstrate generalizability. The predictive power of the ensemble was compared with a logistic regression. Further analyses were performed to identify how important variables impact the ensemble. Results Our ensemble model predicted disposition significantly better than a logistic regression (area under the curve of 0.78 and 0.71, respectively, p = 0.01). Tumor size, presentation at the emergency department, body mass index, convexity location, and preoperative motor deficit most strongly influence the model, though the independent impact of individual variables is nuanced. Conclusion Using a novel ML technique, we built a guided ML ensemble model that predicts discharge destination following meningioma resection with greater predictive power than a logistic regression, and that provides greater clinical insight than a univariate analysis. These techniques can be extended to predict many other patient outcomes of interest.

Risk Factors for Intraoperative and Postoperative Cerebrospinal Fluid Leaks in Endoscopic Transsphenoidal Sellar Surgery.

Objective To determine the factors associated with intra- and postoperative cerebrospinal fluid (CSF) leaks in setting of endoscopic transsphenoidal sellar surgery. Study Design Retrospective cohort. Setting Tertiary referral center. Subjects and Methods This study included 806 patients who underwent endoscopic transsphenoidal sellar surgery between 2004 and 2016. The associations between CSF leaks (intra- and postoperative) and patient demographics, medical history, tumor characteristics, and intraoperative repair techniques were analyzed. Results In sum, 205 (25.4%) patients had a CSF leak: 188 (23.3%) intraoperative leaks and 38 (4.7%) postoperative leaks. Twenty-one (2.6%) patients had postoperative leaks after having repair of an intraoperative leak; 55% of patients with a postoperative leak had an intraoperative leak repaired. On multivariate analysis, body mass index (BMI), hydrocephalus, suprasellar extension, and craniopharyngioma significantly predicted intraoperative CSF leaks, while only BMI and hydrocephalus predicted postoperative CSF leaks. Patients having septal flap repairs of CSF leaks had a higher postoperative leak rate relative to other repair techniques (odds ratio, 6.37; P = .013). Rigid reconstruction did not correlate with leaks. Conclusion For this large cohort of patients undergoing endoscopic transsphenoidal sellar surgery, BMI and hydrocephalus were identified as predictors of postoperative CSF leaks, including those occurring after repair of intraoperative leak. These variables may put stress on the surgical repair of sellar defects, and consideration of these risk factors may help counsel patients and guide perioperative decision making in regard to repair strategies and CSF diversion techniques.

Tumor Progression Following Petroclival Meningioma Subtotal Resection: A Volumetric Study.

BACKGROUND: One study has investigated postoperative growth rates following subtotal resection of petroclival meningiomas utilizing linear measurements, which are insensitive to the multidimensional complex growth of meningiomas, to estimate tumor volume. OBJECTIVE: To describe petroclival meningioma growth patterns following less-than-complete resection utilizing volumetric analysis and to identify variables associated with tumor progression. METHODS: Patients with surgically resected WHO grade I petroclival meningiomas were retrospectively reviewed (1999-2015). Image analysis software was utilized to perform volumetric analyses of tumor size and growth on serial MRI studies. The impact of preoperative and postoperative variables on tumor growth after subtotal resection was analyzed. An increase in tumor volume of at least 20% was defined as "tumor growth." RESULTS: Twenty-three patients had immediate preoperative and serial postoperative MRI studies available for review. The mean preoperative tumor volume was 20.9 cm3 (range 0.4-54.6). The mean extent of resection was 75.5% (range 31.5%-100.0%). At a mean follow-up of 24.8 mo, 12 tumors (66.7%) exhibited radiological tumor growth, while 6 tumors did not change in size. The median annual volumetric growth rate was 2.82 cm3/yr (range -0.34 to 10.1). Extent of resection and immediate postoperative tumor volume were significantly correlated with the annual volumetric growth rate following resection. At last follow-up, 3 (13%) patients required further intervention. CONCLUSION: The majority of petroclival meningiomas exhibit growth following subtotal resection. Extent of resection is strongly associated with risk for disease progression following surgery.

Decreased survival in glioblastomas is specific to contact with the ventricular-subventricular zone, not subgranular zone or corpus callosum.

The clinical effect of radiographic contact of glioblastoma (GBM) with neurogenic zones (NZ)-the ventricular-subventricular (VSVZ) and subgranular (SGZ) zones-and the corpus callosum (CC) remains unclear and, in the case of the SGZ, unexplored. We investigated (1) if GBM contact with a NZ correlates with decreased survival; (2) if so, whether this effect is associated with a specific NZ; and (3) if radiographic contact with or invasion of the CC by GBM is associated with decreased survival. We retrospectively identified 207 adult patients who underwent cytoreductive surgery for GBM followed by chemotherapy and/or radiation. Age, preoperative Karnofsky performance status score (KPS), and extent of resection were recorded. Preoperative MRIs were blindly analyzed to calculate tumor volume and assess its contact with VSVZ, SGZ, CC, and cortex. Overall (OS) and progression free (PFS) survivals were calculated and analyzed with multivariate Cox analyses. Among the 207 patients, 111 had GBM contacting VSVZ (VSVZ+GBMs), 23 had SGZ+GBMs, 52 had CC+GBMs, and 164 had cortex+GBMs. VSVZ+, SGZ+, and CC+ GBMs were significantly larger in size relative to their respective non-contacting controls. Multivariate Cox survival analyses revealed GBM contact with the VSVZ, but not SGZ, CC, or cortex, as an independent predictor of lower OS, PFS, and early recurrence. We hypothesize that the VSVZ niche has unique properties that contribute to GBM pathobiology in adults.

The Natural History of Petroclival Meningiomas: A Volumetric Study.

OBJECTIVES: This study characterizes primary petroclival meningioma growth rates, before intervention, using volumetric analysis. In addition, predictors of growth are analyzed. METHODS: Patients with previously untreated petroclival meningiomas were retrospectively reviewed (1999-2015). Image analysis software was used to perform volumetric analyses of tumor size and growth. Three-dimensional segmentation volumetric analyses were compared with volumes estimated utilizing three orthogonal dimensions. Tumor growth was defined as a 15% increase in volume. RESULTS: Thirty-four patients who underwent at least two magnetic resonance imaging (MRI) studies before intervention were included. The mean age was 55.2 years, and 65.7% were women. The mean tumor volume at presentation was 5.6 cm (range, 0.1-25.8 cm) as determined from segmentation volumetric analysis. At a mean follow-up of 44.5 months (range, 3.7-125.1 mo), 88.2% of tumors grew. The mean annual volumetric growth rate was 2.38 cm/yr (-0.63 to 25.9 cm/yr). Tumor volume, T2 hyperintensity within the tumor, peritumoral edema, and ataxia and/or cerebellar symptoms at presentation were all significantly associated with greater rates of tumor growth. Ultimately, 10 (29.4%) patients underwent treatment during the follow-up period. CONCLUSION: Our experience demonstrates that the vast majority (88%) of untreated petroclival meningiomas grow; the mean volumetric growth rate was noted to be 2.38 cm/yr. We found a significant association between increased growth rate and larger tumor size at diagnosis, T2 hyperintensity within the tumor, peritumoral edema, and the presence of ataxia and/or cerebellar symptoms.

Influence of glioblastoma contact with the lateral ventricle on survival: a meta-analysis.

The ventricular-subventricular zone (V-SVZ), which lies in the walls of the lateral ventricles (LV), is the largest neurogenic niche within the adult brain. Whether radiographic contact with the LV influences survival in glioblastoma (GBM) patients remains unclear. We assimilated and analyzed published data comparing survival in GBM patients with (LV+GBM) and without (LV-GBM) radiographic LV contact. PubMed, EMBASE, and Cochrane electronic databases were searched. Fifteen studies with survival data on LV+GBM and LV-GBM patients were identified. Their Kaplan-Meier survival curves were digitized and pooled for generation of median overall (OS) and progression free (PFS) survivals and log-rank hazard ratios (HRs). The log-rank and reported multivariate HRs after accounting for the common predictors of GBM survival were analyzed separately by meta-analyses. The calculated median survivals (months) from pooled data were 12.95 and 16.58 (OS), and 4.54 and 6.25 (PFS) for LV+GBMs and LV-GBMs, respectively, with an overall log-rank HRs of 1.335 [1.204-1.513] (OS) and 1.387 [1.225-1.602] (PFS). Meta-analysis of log-rank HRs resulted in summary HRs of 1.58 [1.35-1.85] (OS, 10 studies) and 1.41 [1.22-1.64] (PFS, 5 studies). Meta-analysis of multivariate HRs resulted in summary HRs of 1.35 [1.14-1.58] (OS, 6 studies) and 1.64 [0.88-3.05] (PFS, 3 studies). Patients with GBM contacting the LV have lower survival. This effect may be independent of the common predictors of GBM survival, suggesting a clinical influence of V-SVZ contact on GBM biology.

Single cell analysis of human tissues and solid tumors with mass cytometry.

BACKGROUND: Mass cytometry measures 36 or more markers per cell and is an appealing platform for comprehensive phenotyping of cells in human tissue and tumor biopsies. While tissue disaggregation and fluorescence cytometry protocols were pioneered decades ago, it is not known whether established protocols will be effective for mass cytometry and maintain cancer and stromal cell diversity. METHODS: Tissue preparation techniques were systematically compared for gliomas and melanomas, patient derived xenografts of small cell lung cancer, and tonsil tissue as a control. Enzymes assessed included DNase, HyQTase, TrypLE, collagenase (Col) II, Col IV, Col V, and Col XI. Fluorescence and mass cytometry were used to track cell subset abundance following different enzyme combinations and treatment times. RESULTS: Mechanical disaggregation paired with enzymatic dissociation by Col II, Col IV, Col V, or Col XI plus DNase for 1 h produced the highest yield of viable cells per gram of tissue. Longer dissociation times led to increasing cell death and disproportionate loss of cell subsets. Key markers for establishing cell identity included CD45, CD3, CD4, CD8, CD19, CD64, HLA-DR, CD11c, CD56, CD44, GFAP, S100B, SOX2, nestin, vimentin, cytokeratin, and CD31. Mass and fluorescence cytometry identified comparable frequencies of cancer cell subsets, leukocytes, and endothelial cells in glioma (R = 0.97), and tonsil (R = 0.98). CONCLUSIONS: This investigation establishes standard procedures for preparing viable single cell suspensions that preserve the cellular diversity of human tissue microenvironments. © 2016 International Clinical Cytometry Society.

Type 2 diabetes is an independent negative prognostic factor in patients undergoing surgical resection of a WHO grade I meningioma.

OBJECTIVES: In recent years, there has been increased recognition of the relationship between type 2 diabetes mellitus (DM) and poor outcomes following a variety of surgical procedures. We sought to study the role of type 2 DM as a prognostic factor affecting the long-term survival of patients undergoing surgical resection of a WHO Grade I meningioma. METHODS: We conducted a retrospective cohort study on 196 patients who had a WHO Grade I meningioma resected at our institution between 2001 and 2013. The medical record was reviewed to identify a pre-existing diagnosis of type 2 DM. Patient mortality was reviewed by medical record and Social Security Death Index (SSDI). Variables associated with survival in a univariate analysis were included in the multivariate Cox model if P<0.10. Variables with probability values >0.05 were then removed from the multivariate model in a step-wise fashion. RESULTS: 33 (17%) patients had pre-existing diagnoses of type 2 DM prior to clinical presentation. Mean survival time in diabetic patients was 52.1 months compared to 160.9 months in non-diabetics. The decreased survival rate and time in patients with type 2 DM were found to be statistically significant (p=0.008 and p<0.0001, respectively). In a multivariate Cox analysis, a pre-existing history of type 2 DM was independently associated with decreased survival following the resection of a WHO Grade I meningioma (HR=2.6, p=0.045). CONCLUSIONS: A pre-existing diagnosis of type 2 DM is an independent negative prognostic indicator following the resection of a WHO Grade I meningioma.

Factors Associated with Low Socioeconomic Status Predict Poor Postoperative Follow-up after Meningioma Resection.

Objectives To quantify the rates of loss of follow-up after meningioma resection and to identify any key demographical associations. Design Retrospective cohort. Setting Vanderbilt University Medical Center, 2001-2013. Participants A total of 281 patients surgically treated for an intracranial meningioma at a single institution between 2001 and 2013. Main Outcome Measures Patient clinical follow-up within the first postoperative year. Results A history of tobacco use (p < 0.0001), ongoing alcohol abuse at time of presentation (p = 0.0014), Medicaid coverage (p < 0.0001), and lack of a college degree (p < 0.0001) were all found to be predictors of loss of follow-up at a statistically significant level. Conclusions Several factors associated with low socioeconomic status are predictors of poor clinical follow-up after meningioma resection. The health risk of poor follow-up in this patient population is significant, and increased measures are needed to ensure regular appointment attendance.

Risk of post-operative pneumocephalus in patients with obstructive sleep apnea undergoing transsphenoidal surgery.

Patients undergoing transsphenoidal surgery (TSS) have an anterior skull base defect that limits the use of positive pressure ventilation post-operatively. Obstructive sleep apnea (OSA) can be seen in these patients and is treated with continuous positive airway pressure (CPAP). In our study we documented the incidence of pre-existing OSA and reported the incidence of diagnosed pneumocephalus and its relationship to OSA. A retrospective review was conducted from a surgical outcomes database. Electronic medical records were reviewed, with an emphasis on diagnosis of OSA and documented symptomatic pneumocephalus. A total of 324 patients underwent 349 TSS for sellar mass resection. The average body mass index of the study cohort was 32.5kg/m(2). Sixty-nine patients (21%) had documented OSA. Only 25 out of 69 (36%) had a documented post-operative CPAP plan. Out of all 349 procedures, there were two incidents of pneumocephalus diagnosed. Neither of the patients had pre-existing OSA. One in five patients in our study had pre-existing OSA. Most patients returned to CPAP use within several weeks of TSS for resection of a sellar mass. Neither of the patients with pneumocephalus had pre-existing OSA and none of the patients with early re-initiation of CPAP developed this complication. This study provides preliminary evidence that resuming CPAP early in the post-operative period might be less dangerous than previously assumed.

Initial Experiments with the Leap Motion as a User Interface in Robotic Endonasal Surgery.

The Leap Motion controller is a low-cost, optically-based hand tracking system that has recently been introduced on the consumer market. Prior studies have investigated its precision and accuracy, toward evaluating its usefulness as a surgical robot master interface. Yet due to the diversity of potential slave robots and surgical procedures, as well as the dynamic nature of surgery, it is challenging to make general conclusions from published accuracy and precision data. Thus, our goal in this paper is to explore the use of the Leap in the specific scenario of endonasal pituitary surgery. We use it to control a concentric tube continuum robot in a phantom study, and compare user performance using the Leap to previously published results using the Phantom Omni. We find that the users were able to achieve nearly identical average resection percentage and overall surgical duration with the Leap.