PACS-Integrated Tools for Peritumoral Edema Volumetrics Provide Additional Information to RANO-BM-Based Assessment of Lung Cancer Brain Metastases after Stereotactic Radiotherapy: A Pilot Study.

Stereotactic radiotherapy (SRT) is the standard of care treatment for brain metastases (METS) today. Nevertheless, there is limited understanding of how posttreatment lesional volumetric changes may assist prediction of lesional outcome. This is partly due to the paucity of volumetric segmentation tools. Edema alone can cause significant clinical symptoms and, therefore, needs independent study along with standard measurements of contrast-enhancing tumors. In this study, we aimed to compare volumetric changes of edema to RANO-BM-based measurements of contrast-enhancing lesion size. Patients with NSCLC METS ≥10 mm on post-contrast T1-weighted image and treated with SRT had measurements for up to seven follow-up scans using a PACS-integrated tool segmenting the peritumoral FLAIR hyperintense volume. Two-dimensional contrast-enhancing and volumetric edema changes were compared by creating treatment response curves. Fifty NSCLC METS were included in the study. The initial median peritumoral edema volume post-SRT relative to pre-SRT baseline was 37% (IQR 8-114%). Most of the lesions with edema volume reduction post-SRT experienced no increase in edema during the study. In over 50% of METS, the pattern of edema volume change was different than the pattern of contrast-enhancing lesion change at different timepoints, which was defined as incongruent. Lesions demonstrating incongruence at the first follow-up were more likely to progress subsequently. Therefore, edema assessment of METS post-SRT provides critical additional information to RANO-BM.

Systematic Literature Review of Machine Learning Algorithms Using Pretherapy Radiologic Imaging for Glioma Molecular Subtype Prediction.

BACKGROUND: The molecular profile of gliomas is a prognostic indicator for survival, driving clinical decision-making for treatment. Pathology-based molecular diagnosis is challenging because of the invasiveness of the procedure, exclusion from neoadjuvant therapy options, and the heterogeneous nature of the tumor. PURPOSE: We performed a systematic review of algorithms that predict molecular subtypes of gliomas from MR Imaging. DATA SOURCES: Data sources were Ovid Embase, Ovid MEDLINE, Cochrane Central Register of Controlled Trials, Web of Science. STUDY SELECTION: Per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, 12,318 abstracts were screened and 1323 underwent full-text review, with 85 articles meeting the inclusion criteria. DATA ANALYSIS: We compared prediction results from different machine learning approaches for predicting molecular subtypes of gliomas. Bias analysis was conducted for each study, following the Prediction model Risk Of Bias Assessment Tool (PROBAST) guidelines. DATA SYNTHESIS: Isocitrate dehydrogenase mutation status was reported with an area under the curve and accuracy of 0.88 and 85% in internal validation and 0.86 and 87% in limited external validation data sets, respectively. For the prediction of O6-methylguanine-DNA methyltransferase promoter methylation, the area under the curve and accuracy in internal validation data sets were 0.79 and 77%, and in limited external validation, 0.89 and 83%, respectively. PROBAST scoring demonstrated high bias in all articles. LIMITATIONS: The low number of external validation and studies with incomplete data resulted in unequal data analysis. Comparing the best prediction pipelines of each study may introduce bias. CONCLUSIONS: While the high area under the curve and accuracy for the prediction of molecular subtypes of gliomas are reported in internal and external validation data sets, limited use of external validation and the increased risk of bias in all articles may present obstacles for clinical translation of these techniques.

Long-term Outcomes in Primary CNS Lymphoma After R-MVP and High-Dose Chemotherapy With Autologous Hematopoietic Stem Cell Transplant.

BACKGROUND AND OBJECTIVES: Primary CNS lymphoma (PCNSL), a rare CNS malignancy, is usually treated with high-dose methotrexate in the first-line setting, typically followed by consolidation therapy. Due to the broad range of currently available treatments for PCNSL, comparability in long-term follow-up studies is limited, and data are scattered across small studies. METHODS: In this study, we report the long-term survival of patients with newly diagnosed immunocompetent PCNSL, enrolled in a phase II trial from June 2005 to September 2011. Patients were treated using rituximab, methotrexate, vincristine, and procarbazine (R-MVP) chemotherapy followed by high-dose chemotherapy (HDC) and autologous stem cell transplant (ASCT) in those with partial or complete response to R-MVP. In a post hoc analysis, clinical and imaging features were evaluated in those still alive. RESULTS: 26 of 32 patients underwent HDC-ASCT consolidation. Of them, 3 patients died of treatment-related toxicity and 2 due to disease progression within 1 year of ASCT. None of the remaining 21 patients had disease progression with a median follow-up of 12.1 years and were included in the analysis. Compared with the post-HDC-ASCT assessment, at the last follow-up, there was no significant difference in the median Karnofsky Performance Status (80 [range: 60-100] vs 90 [range: 70-100]), the median Neurologic Assessment in Neuro-Oncology score (1 [range: 0-4] vs 1 [range: 0-5]), and leukoencephalopathy score (1 [range: 0-3] vs 1 [range: 1-4]). DISCUSSION: Long-term follow-up demonstrated that treatment was well tolerated in most patients enrolled in this study, with stable leukoencephalopathy on imaging and stable clinical performance status. Disease recurrence was not observed beyond 2 years after HDC-ASCT consolidation.

Multicenter Phase II Trial of the PARP Inhibitor Olaparib in Recurrent IDH1- and IDH2-mutant Glioma.

Purpose: Isocitrate dehydrogenase (IDH) 1 and IDH2 mutations (IDH1/2mt) are frequent in glioma. Preclinical studies suggest IDH1/2mts confer "BRCAness" phenotype, a vulnerability that can be targeted through PARP inhibition. To test this hypothesis, we conducted a multicenter study of olaparib monotherapy in patients with IDH1/2mt gliomas. Methods: Patients with recurrent, contrast-enhancing IDH1/2mt gliomas were enrolled in a two-step phase II trial; the primary endpoint was overall response rate per Response Assessment in Neuro-Oncology (RANO) criteria. Olaparib 300 mg orally twice daily was given. Results: A total of 15 evaluable patients were enrolled. Histology was astrocytoma (N = 12) and oligodendroglioma (N = 3). Most toxicities were grade 1 or 2. Best response was stable disease (SD) in 9 (60%) patients. Median progression-free survival (PFS) was 3.63 months and median overall survival was 20.7 months. For patients with SD, median PFS was 5.53 months; 4 patients had SD for >6 months. Among patients with best response progressive disease (N = 6), 5 had grade 4 tumor and 4 had known CDKN2A alteration. PFS was 5.23 months for grades 2 or 3 tumors (N = 10) versus 1.8 months for grade 4 (N = 5; P = 0.0013). Conclusion: The study did not meet the prespecified response-based activity threshold for moving to step 2. However, prolonged SD was observed in patients with grades 2 and 3 histologies, suggesting olaparib monotherapy could be of clinical benefit in select populations. Grade 4 tumors per 2021 World Health Organization classification defined by histology or CDKN2A alteration derived no benefit from this drug, highlighting the usefulness of this classification for future patient stratification and trial design. Significance: A single-arm phase II trial of olaparib in IDH-mutant glioma demonstrated clinically significant prolonged SD for select patients with grade 2/3 disease, suggesting potential benefit of olaparib in IDH-mutant gliomas.

Comparing Detection Schemes for Adversarial Images against Deep Learning Models for Cancer Imaging.

Deep learning (DL) models have demonstrated state-of-the-art performance in the classification of diagnostic imaging in oncology. However, DL models for medical images can be compromised by adversarial images, where pixel values of input images are manipulated to deceive the DL model. To address this limitation, our study investigates the detectability of adversarial images in oncology using multiple detection schemes. Experiments were conducted on thoracic computed tomography (CT) scans, mammography, and brain magnetic resonance imaging (MRI). For each dataset we trained a convolutional neural network to classify the presence or absence of malignancy. We trained five DL and machine learning (ML)-based detection models and tested their performance in detecting adversarial images. Adversarial images generated using projected gradient descent (PGD) with a perturbation size of 0.004 were detected by the ResNet detection model with an accuracy of 100% for CT, 100% for mammogram, and 90.0% for MRI. Overall, adversarial images were detected with high accuracy in settings where adversarial perturbation was above set thresholds. Adversarial detection should be considered alongside adversarial training as a defense technique to protect DL models for cancer imaging classification from the threat of adversarial images.

Top advances of the year: Neuro-oncology.

Management of brain tumors has been challenging given the limited therapeutic options and disabling morbidities associated with central nervous system (CNS) dysfunction. This review focuses on recent developments in the field, with an emphasis on clinical management. The growing clinical trials landscape reflects advanced insights into cancer immunology and genomics and the need to address molecular and clinical heterogeneity. Recent phase 3 trials investigating anti-PD-1 immunotherapies, particularly nivolumab, have failed to demonstrate improved survival in glioblastoma, underscoring the need to better understand the complexity of CNS immunologic surveillance. Conversely, targeted therapies have accounted for several US Food and Drug Administration approvals extended to brain tumors, particularly therapies directed to BRAF V600E mutations and TRAK fusions, underscoring a need to routinely screen patients for these rare molecular abnormalities. In primary CNS lymphoma, attention has turned to long-term outcomes of consolidation therapies, and recent studies have highlighted the excellent disease control afforded by high-dose chemotherapy and stem cell transplantation. Meningiomas remain a focus of investigations, with preliminary promising results observed with octreotide combined with mTOR inhibition, and immunotherapy with single-agent pembrolizumab. Finally, proton radiotherapy has emerged as a novel alternative for leptomeningeal metastases from solid tumors, which can now be treated more safely with craniospinal irradiation and monitored by the enumeration of circulating tumor cells in the cerebrospinal fluid as a biomarker. Taken together, these incremental advances have improved outcomes in select brain tumor patient populations, whereas ongoing clinical trials hold the promise of meaningful advances and breakthroughs for larger proportions of patients with brain tumors.

Radiotherapy combined with nivolumab or temozolomide for newly diagnosed glioblastoma with unmethylated MGMT promoter: An international randomized phase III trial.

BACKGROUND: Addition of temozolomide (TMZ) to radiotherapy (RT) improves overall survival (OS) in patients with glioblastoma (GBM), but previous studies suggest that patients with tumors harboring an unmethylated MGMT promoter derive minimal benefit. The aim of this open-label, phase III CheckMate 498 study was to evaluate the efficacy of nivolumab (NIVO) + RT compared with TMZ + RT in newly diagnosed GBM with unmethylated MGMT promoter. METHODS: Patients were randomized 1:1 to standard RT (60 Gy) + NIVO (240 mg every 2 weeks for eight cycles, then 480 mg every 4 weeks) or RT + TMZ (75 mg/m2 daily during RT and 150-200 mg/m2/day 5/28 days during maintenance). The primary endpoint was OS. RESULTS: A total of 560 patients were randomized, 280 to each arm. Median OS (mOS) was 13.4 months (95% CI, 12.6 to 14.3) with NIVO + RT and 14.9 months (95% CI, 13.3 to 16.1) with TMZ + RT (hazard ratio [HR], 1.31; 95% CI, 1.09 to 1.58; P = .0037). Median progression-free survival was 6.0 months (95% CI, 5.7 to 6.2) with NIVO + RT and 6.2 months (95% CI, 5.9 to 6.7) with TMZ + RT (HR, 1.38; 95% CI, 1.15 to 1.65). Response rates were 7.8% (9/116) with NIVO + RT and 7.2% (8/111) with TMZ + RT; grade 3/4 treatment-related adverse event (TRAE) rates were 21.9% and 25.1%, and any-grade serious TRAE rates were 17.3% and 7.6%, respectively. CONCLUSIONS: The study did not meet the primary endpoint of improved OS; TMZ + RT demonstrated a longer mOS than NIVO + RT. No new safety signals were detected with NIVO in this study. The difference between the study treatment arms is consistent with the use of TMZ + RT as the standard of care for GBM.ClinicalTrials.gov NCT02617589.

Identifying clinically applicable machine learning algorithms for glioma segmentation: recent advances and discoveries.

Background: While there are innumerable machine learning (ML) research algorithms used for segmentation of gliomas, there is yet to be a US FDA cleared product. The aim of this study is to explore the systemic limitations of research algorithms that have prevented translation from concept to product by a review of the current research literature. Methods: We performed a systematic literature review on 4 databases. Of 11 727 articles, 58 articles met the inclusion criteria and were used for data extraction and screening using TRIPOD. Results: We found that while many articles were published on ML-based glioma segmentation and report high accuracy results, there were substantial limitations in the methods and results portions of the papers that result in difficulty reproducing the methods and translation into clinical practice. Conclusions: In addition, we identified that more than a third of the articles used the same publicly available BRaTS and TCIA datasets and are responsible for the majority of patient data on which ML algorithms were trained, which leads to limited generalizability and potential for overfitting and bias.

Real-time PACS-integrated longitudinal brain metastasis tracking tool provides comprehensive assessment of treatment response to radiosurgery.

Background: Treatment of brain metastases can be tailored to individual lesions with treatments such as stereotactic radiosurgery. Accurate surveillance of lesions is a prerequisite but challenging in patients with multiple lesions and prior imaging studies, in a process that is laborious and time consuming. We aimed to longitudinally track several lesions using a PACS-integrated lesion tracking tool (LTT) to evaluate the efficiency of a PACS-integrated lesion tracking workflow, and characterize the prevalence of heterogenous response (HeR) to treatment after Gamma Knife (GK). Methods: We selected a group of brain metastases patients treated with GK at our institution. We used a PACS-integrated LTT to track the treatment response of each lesion after first GK intervention to maximally seven diagnostic follow-up scans. We evaluated the efficiency of this tool by comparing the number of clicks necessary to complete this task with and without the tool and examined the prevalence of HeR in treatment. Results: A cohort of eighty patients was selected and 494 lesions were measured and tracked longitudinally for a mean follow-up time of 374 days after first GK. Use of LTT significantly decreased number of necessary clicks. 81.7% of patients had HeR to treatment at the end of follow-up. The prevalence increased with increasing number of lesions. Conclusions: Lesions in a single patient often differ in their response to treatment, highlighting the importance of individual lesion size assessments for further treatment planning. PACS-integrated lesion tracking enables efficient lesion surveillance workflow and specific and objective result reports to treating clinicians.

Immune-checkpoint inhibitors for glioblastoma: what have we learned?

BACKGROUND: Glioblastoma, the most common malignant primary brain tumor, remains a lethal disease with few therapeutic options. Immunotherapies, particularly immune checkpoint inhibitors (ICPi), have revolutionized cancer treatment, but their role in glioblastoma is uncertain. OBJECTIVE: To review the state of immunotherapies in glioblastoma, with an emphasis on recently published ICPi clinical trials. METHODS: In this editorial/opinion article, we critically review results of the first generation of trials of ipilimumab, nivolumab and pembrolizumab in glioblastoma, as well as future directions. RESULTS: Expression of PD-L1 is frequent in glioblastoma, ranging from 60-70% of patients. Phase 1 studies of nivolumab with and without ipilimumab, as well as pembrolizumab, showed no new safety concerns in brain tumors, and no neurotoxicity. However, randomized phase 3 trials of nivolumab showed no survival improvements over bevacizumab in recurrent glioblastoma; no role in newly diagnosed disease as a replacement for temozolomide in unmethylated MGMT promoter tumors; and no benefit as an addition to temozolomide in methylated MGMT tumors. However, studies examining post treatment tumor samples have shown signs of increased immunologic response, and occasional long lasting radiographic responses have been seen. A small study of pembrolizumab suggested a potential role as a "neoadjuvant" treatment in resectable recurrent glioblastoma, while other studies are investigating selection of patients with higher mutational burden and novel agents and combinatorial strategies. CONCLUSION: Despite initial negative trials, immunotherapy remains of high interest in glioblastoma, and many trials are still ongoing. Improving our mechanistic understanding of the immunosuppression and T cell dysfunction induced by both tumor and the CNS microenvironment remains however crucial for the development of successful immunotherapeutic approaches in this disease.

Phase I/randomized phase II trial of TRC105 plus bevacizumab versus bevacizumab in recurrent glioblastoma: North Central Cancer Treatment Group N1174 (Alliance).

Background: Patients with glioblastoma (GBM) have a poor prognosis and limited effective treatment options. Bevacizumab has been approved for treatment of recurrent GBM, but there is questionable survival benefit. Based on preclinical and early clinical data indicating that CD105 upregulation may represent a mechanism of resistance to bevacizumab, we hypothesized that combining bevacizumab with the anti-CD105 antibody TRC105 may improve efficacy in recurrent GBM. Methods: Phase I dose-escalation/comparative randomized phase II trial in patients with GBM. During phase I, the maximum tolerated dose (MTD) of TRC105 in combination with bevacizumab was determined. In phase II, patients were randomized 1:1 to TRC105 and bevacizumab or bevacizumab monotherapy. Patients received TRC105 (10 mg/kg) weekly and bevacizumab (10 mg/kg) every 2 weeks. Efficacy, as assessed by progression-free survival (PFS), was the primary endpoint; safety, quality of life, and correlative outcomes were also evaluated. Results: In total, 15 patients were enrolled in phase I and 101 in phase II; 52 patients were randomized to TRC105 with bevacizumab and 49 to bevacizumab monotherapy. The MTD was determined to be 10 mg/kg TRC105 weekly plus bevacizumab 10 mg/kg every 2 weeks. An increased occurrence of grade ≥3 adverse events was seen in the combination arm, including higher incidences of anemia. Median PFS was similar in both treatment arms: 2.9 months for combination versus 3.2 months for bevacizumab monotherapy (HR = 1.16, 95% CI = 0.75-1.78, P = .51). Quality of life scores were similar for both treatment arms. Conclusions: TRC105 in combination with bevacizumab was well tolerated in patients with recurrent GBM, but no difference in efficacy was observed compared to bevacizumab monotherapy.

Phase III trial of chemoradiotherapy with temozolomide plus nivolumab or placebo for newly diagnosed glioblastoma with methylated MGMT promoter.

BACKGROUND: Nearly all patients with newly diagnosed glioblastoma experience recurrence following standard-of-care radiotherapy (RT) + temozolomide (TMZ). The purpose of the phase III randomized CheckMate 548 study was to evaluate RT + TMZ combined with the immune checkpoint inhibitor nivolumab (NIVO) or placebo (PBO) in patients with newly diagnosed glioblastoma with methylated MGMT promoter (NCT02667587). METHODS: Patients (N = 716) were randomized 1:1 to NIVO [(240 mg every 2 weeks × 8, then 480 mg every 4 weeks) + RT (60 Gy over 6 weeks) + TMZ (75 mg/m2 once daily during RT, then 150-200 mg/m2 once daily on days 1-5 of every 28-day cycle × 6)] or PBO + RT + TMZ following the same regimen. The primary endpoints were progression-free survival (PFS) and overall survival (OS) in patients without baseline corticosteroids and in all randomized patients. RESULTS: As of December 22, 2020, median (m)PFS (blinded independent central review) was 10.6 months (95% CI, 8.9-11.8) with NIVO + RT + TMZ vs 10.3 months (95% CI, 9.7-12.5) with PBO + RT + TMZ (HR, 1.1; 95% CI, 0.9-1.3) and mOS was 28.9 months (95% CI, 24.4-31.6) vs 32.1 months (95% CI, 29.4-33.8), respectively (HR, 1.1; 95% CI, 0.9-1.3). In patients without baseline corticosteroids, mOS was 31.3 months (95% CI, 28.6-34.8) with NIVO + RT + TMZ vs 33.0 months (95% CI, 31.0-35.1) with PBO + RT + TMZ (HR, 1.1; 95% CI, 0.9-1.4). Grade 3/4 treatment-related adverse event rates were 52.4% vs 33.6%, respectively. CONCLUSIONS: NIVO added to RT + TMZ did not improve survival in patients with newly diagnosed glioblastoma with methylated or indeterminate MGMT promoter. No new safety signals were observed.

Immune-checkpoint inhibitors for glioblastoma: what have we learned? / Inibidores de checkpoint imunológico para glioblastoma: o que aprendemos?

ABSTRACT Background: Glioblastoma, the most common malignant primary brain tumor, remains a lethal disease with few therapeutic options. Immunotherapies, particularly immune checkpoint inhibitors (ICPi), have revolutionized cancer treatment, but their role in glioblastoma is uncertain. Objective: To review the state of immunotherapies in glioblastoma, with an emphasis on recently published ICPi clinical trials. Methods: In this editorial/opinion article, we critically review results of the first generation of trials of ipilimumab, nivolumab and pembrolizumab in glioblastoma, as well as future directions. Results: Expression of PD-L1 is frequent in glioblastoma, ranging from 60-70% of patients. Phase 1 studies of nivolumab with and without ipilimumab, as well as pembrolizumab, showed no new safety concerns in brain tumors, and no neurotoxicity. However, randomized phase 3 trials of nivolumab showed no survival improvements over bevacizumab in recurrent glioblastoma; no role in newly diagnosed disease as a replacement for temozolomide in unmethylated MGMT promoter tumors; and no benefit as an addition to temozolomide in methylated MGMT tumors. However, studies examining post treatment tumor samples have shown signs of increased immunologic response, and occasional long lasting radiographic responses have been seen. A small study of pembrolizumab suggested a potential role as a "neoadjuvant" treatment in resectable recurrent glioblastoma, while other studies are investigating selection of patients with higher mutational burden and novel agents and combinatorial strategies. Conclusion: Despite initial negative trials, immunotherapy remains of high interest in glioblastoma, and many trials are still ongoing. Improving our mechanistic understanding of the immunosuppression and T cell dysfunction induced by both tumor and the CNS microenvironment remains however crucial for the development of successful immunotherapeutic approaches in this disease.

RESUMO Antecedentes: Glioblastoma é o tumor cerebral primário maligno mais comum e continua sendo uma doença letal com poucas opções terapêuticas. As imunoterapias, em especial, os inibidores de checkpoint imunológico (ICPi), revolucionaram o tratamento do câncer, mas seu papel no glioblastoma é incerto. Objetivo: Revisar o estado atual do papel das imunoterapias no glioblastoma, com ênfase nos ensaios clínicos ICPi publicados recentemente. Métodos: Neste artigo de revisão, analisamos criticamente os resultados da primeira geração de estudos de ipilimumab, nivolumab e pembrolizumab em glioblastoma, bem como as perspectivas futuras. Resultados: A expressão de PD-L1 é frequente no glioblastoma, variando de 60-70% dos pacientes. Estudos de fase 1 de nivolumab com e sem ipilimumab, bem como pembrolizumab, não revelaram novas questões de tolerabilidade nem neurotoxicidade. No entanto, ensaios randomizados de fase 3 de nivolumab não apontaram melhorias na sobrevida em relação ao bevacizumab em glioblastoma recorrente, nenhum papel na doença recém-diagnosticada como substituto da temozolomida em tumores promotores de MGMT não metilados e nenhum benefício como adição à temozolomida em tumores MGMT metilados. No entanto, estudos que examinaram amostras de tumores pós-tratamento mostraram sinais de aumento da resposta imunológica, e respostas radiográficas ocasionais de longa duração foram observadas. Um pequeno estudo de pembrolizumab sugeriu um papel potencial como tratamento "neoadjuvante" no glioblastoma recorrente ressecável, ao passo que outros estudos estão investigando a seleção de pacientes com maior carga mutacional e novos agentes e estratégias combinatórias. Conclusão: Apesar dos ensaios iniciais negativos, a imunoterapia continua sendo de grande interesse no glioblastoma, e muitos ensaios ainda estão em andamento. No entanto, melhorar a nossa compreensão dos mecanismos de imunossupressão e disfunção das células T induzidas tanto pelo tumor quanto pelo microambiente do SNC continua sendo crucial para o desenvolvimento de abordagens imunoterapêuticas bem-sucedidas nesta doença.

Nivolumab plus radiotherapy with or without temozolomide in newly diagnosed glioblastoma: Results from exploratory phase I cohorts of CheckMate 143.

Background: The phase 1 cohorts (1c+1d) of CheckMate 143 (NCT02017717) evaluated the safety/tolerability and efficacy of nivolumab plus radiotherapy (RT) ± temozolomide (TMZ) in newly diagnosed glioblastoma. Methods: In total, 136 patients were enrolled. In part A (safety lead-in), 31 patients (n = 15, methylated/unknown MGMT promoter; n = 16, unmethylated MGMT promoter) received nivolumab and RT+TMZ (NIVO+RT+TMZ) and 30 patients with unmethylated MGMT promoter received NIVO+RT. In part B (expansion), patients with unmethylated MGMT promoter were randomized to NIVO+RT+TMZ (n = 29) or NIVO+RT (n = 30). Primary endpoint was safety/tolerability; secondary endpoint was overall survival (OS). Results: NIVO+RT±TMZ was tolerable; grade 3/4 treatment-related adverse events occurred in 51.6% (NIVO+RT+TMZ) and 30.0% (NIVO+RT) of patients in part A and 46.4% (NIVO+RT+TMZ) and 28.6% (NIVO+RT) in part B. No new safety signals were detected. In part A, median OS (mOS) with NIVO+RT+TMZ was 33.38 months (95% CI, 16.2 to not estimable) in patients with methylated MGMT promoter. In patients with unmethylated MGMT promoter, mOS was 16.49 months (12.94-22.08) with NIVO+RT+TMZ and 14.41 months (12.55-17.31) with NIVO+RT. In part B, mOS was 14.75 months (10.01-18.6) with NIVO+RT+TMZ and 13.96 months (10.81-18.14) with NIVO+RT in patients with unmethylated MGMT promoter. Conclusions: CheckMate 143 was the first trial evaluating immune checkpoint inhibition with first-line treatment of glioblastoma. Results showed that NIVO can be safely combined with RT±TMZ, with no new safety signals. Toxicities, including lymphopenia, were more frequent with NIVO+RT+TMZ. OS was similar with or without TMZ in patients with unmethylated MGMT promoter, and differences by MGMT methylation status were observed.

Machine Learning Applications for Differentiation of Glioma from Brain Metastasis-A Systematic Review.

Glioma and brain metastasis can be difficult to distinguish on conventional magnetic resonance imaging (MRI) due to the similarity of imaging features in specific clinical circumstances. Multiple studies have investigated the use of machine learning (ML) models for non-invasive differentiation of glioma from brain metastasis. Many of the studies report promising classification results, however, to date, none have been implemented into clinical practice. After a screening of 12,470 studies, we included 29 eligible studies in our systematic review. From each study, we aggregated data on model design, development, and best classifiers, as well as quality of reporting according to the TRIPOD statement. In a subset of eligible studies, we conducted a meta-analysis of the reported AUC. It was found that data predominantly originated from single-center institutions (n = 25/29) and only two studies performed external validation. The median TRIPOD adherence was 0.48, indicating insufficient quality of reporting among surveyed studies. Our findings illustrate that despite promising classification results, reliable model assessment is limited by poor reporting of study design and lack of algorithm validation and generalizability. Therefore, adherence to quality guidelines and validation on outside datasets is critical for the clinical translation of ML for the differentiation of glioma and brain metastasis.

Using Adversarial Images to Assess the Robustness of Deep Learning Models Trained on Diagnostic Images in Oncology.

PURPOSE: Deep learning (DL) models have rapidly become a popular and cost-effective tool for image classification within oncology. A major limitation of DL models is their vulnerability to adversarial images, manipulated input images designed to cause misclassifications by DL models. The purpose of the study is to investigate the robustness of DL models trained on diagnostic images using adversarial images and explore the utility of an iterative adversarial training approach to improve the robustness of DL models against adversarial images. METHODS: We examined the impact of adversarial images on the classification accuracies of DL models trained to classify cancerous lesions across three common oncologic imaging modalities. The computed tomography (CT) model was trained to classify malignant lung nodules. The mammogram model was trained to classify malignant breast lesions. The magnetic resonance imaging (MRI) model was trained to classify brain metastases. RESULTS: Oncologic images showed instability to small pixel-level changes. A pixel-level perturbation of 0.004 (for pixels normalized to the range between 0 and 1) resulted in most oncologic images to be misclassified (CT 25.6%, mammogram 23.9%, and MRI 6.4% accuracy). Adversarial training improved the stability and robustness of DL models trained on oncologic images compared with naive models ([CT 67.7% v 26.9%], mammogram [63.4% vs 27.7%], and MRI [87.2% vs 24.3%]). CONCLUSION: DL models naively trained on oncologic images exhibited dramatic instability to small pixel-level changes resulting in substantial decreases in accuracy. Adversarial training techniques improved the stability and robustness of DL models to such pixel-level changes. Before clinical implementation, adversarial training should be considered to proposed DL models to improve overall performance and safety.

T cell dysfunction in glioblastoma: a barrier and an opportunity for the development of successful immunotherapies.

PURPOSE OF REVIEW: Immunotherapies such as immune checkpoint blockade have revolutionized cancer treatment, but current approaches have failed to improve outcomes in glioblastoma and other brain tumours. T cell dysfunction has emerged as one of the major barriers for the development of central nervous system (CNS)-directed immunotherapy. Here, we explore the unique requirements that T cells must fulfil to ensure immune surveillance in the CNS, and we analyse T cell dysfunction in glioblastoma (GBM) through the prism of CNS-resident immune responses. RECENT FINDINGS: Using comprehensive and unbiased techniques such as single-cell RNA sequencing, multiple studies have dissected the transcriptional state of CNS-resident T cells that patrol the homeostatic brain. A similar approach has revealed that in GBM, tumour-infiltrating T cells lack the hallmarks of antigen-driven exhaustion typical of melanoma and other solid tumours, suggesting the need for better presentation of tumour-derived antigens. Consistently, in a mouse model of GBM, increasing lymphatic drainage to the cervical lymph node was sufficient to promote tumour rejection. SUMMARY: For the success of future immunotherapy strategies, further work needs to explore the natural history of dysfunction in GBM tumour-infiltrating T cells, establish whether these originate from CNS-resident T cells and how they can be manipulated therapeutically.

Comparison of radiomic feature aggregation methods for patients with multiple tumors.

Radiomic feature analysis has been shown to be effective at analyzing diagnostic images to model cancer outcomes. It has not yet been established how to best combine radiomic features in cancer patients with multifocal tumors. As the number of patients with multifocal metastatic cancer continues to rise, there is a need for improving personalized patient-level prognosis to better inform treatment. We compared six mathematical methods of combining radiomic features of 3,596 tumors in 831 patients with multiple brain metastases and evaluated the performance of these aggregation methods using three survival models: a standard Cox proportional hazards model, a Cox proportional hazards model with LASSO regression, and a random survival forest. Across all three survival models, the weighted average of the largest three metastases had the highest concordance index (95% confidence interval) of 0.627 (0.595-0.661) for the Cox proportional hazards model, 0.628 (0.591-0.666) for the Cox proportional hazards model with LASSO regression, and 0.652 (0.565-0.727) for the random survival forest model. This finding was consistent when evaluating patients with different numbers of brain metastases and different tumor volumes. Radiomic features can be effectively combined to estimate patient-level outcomes in patients with multifocal brain metastases. Future studies are needed to confirm that the volume-weighted average of the largest three tumors is an effective method for combining radiomic features across other imaging modalities and tumor types.