Brain Tumor Response Research Articles

Federated Learning Approach for Measuring the Response of Brain Tumors to Chemotherapy

Brain tumor is a fatal disease and one of the major causes of rising death rates in adults. Predicting methylation of the O6-Methylguanine-DNA Methyltransferase (MGMT) gene status utilizing Magnetic resonance imaging (MRI) imaging is highly important since it is a predictor of brain tumor responses to chemotherapy, which reduces the number of needed surgeries. Deep Learning (DL) approaches became powerful in extracting meaningful relationships and making accurate predictions. DL-based models require a large database and accessing or transferring patient data to train the model. Federated machine learning has recently gained popularity, as it offers practical solutions for data privacy, centralized computation, and high computing power. This study aims to investigate the feasibility of federated learning (FL) by developing a FL-based approach to predict MGMT promoter methylation status using the BraTs2021 dataset for the four sequence types, (Fluid Attenuated Inversion Recovery (FLAIR), T1-weighted (T1w), T1-weighted Gadolinium Post Contrast (T1wCE/T1Gd), and T2-weighted (T2w)) MRI images. The FL model compared to the DL-based and the experimental results show that even with imbalanced and heterogeneous datasets, the FL approach reached the training model to 99.99% of the model quality achieved with centralized data after 300 communication rounds between 10 institutions using OpenFl framework and the improved EfficentNet-B3 neural network architecture.

NIMG-40. MRI-BASED ESTIMATION OF THE ABUNDANCE OF IMMUNOHISTOCHEMISTRY MARKERS IN GBM BRAIN USING DEEP LEARNING

Abstract Glioblastoma (GBM) is a devastating primary brain tumor known for its heterogeneity with a median survival of 15 months. Clinical imaging remains the primary modality to assess brain tumor response, but it is nearly impossible to distinguish between tumor growth and treatment response. Ki67 is a marker of active cell proliferation that shows inter- and intra-patient heterogeneity and should change under many therapies. In this work, we assessed the utility of a semi-supervised deep learning approach for regionally predicting high-vs-low Ki67 in GBM patients based on MRI. We used both labeled and unlabeled datasets to train the model. Labeled data included 114 MRI-localized biopsies from 43 unique GBM patients with available immunohistochemistry Ki67 labels. Unlabeled data included nine repeat routine pretreatment paired scans of newly-diagnosed GBM patients acquired within three days. Data augmentation techniques were utilized to enhance the size of our data and increase generalizability. Data was split between training, validation, and testing sets using 65-15-20 percent ratios. Model inputs were 16x16x3 patches around biopsies on T1Gd and T2 MRIs for labeled data, and around randomly selected patches inside the T2 abnormal region for unlabeled data. The network was a 4-conv layered VGG-inspired architecture. Training objective was accurate prediction of Ki67 in labeled patches and consistency in predictions across repeat unlabeled patches. We measured final model accuracy on held-out test samples. Our promising preliminary results suggest potential for deep learning in deconvolving the spatial heterogeneity of proliferative GBM subpopulations. If successful, this model can provide a non-invasive readout of cell proliferation and reveal the effectiveness of a given cytotoxic therapy dynamically during the patient's routine follow up. Further, the spatial resolution of our approach provides insights into the intra-tumoral heterogeneity of response which can be related to heterogeneity in localization of therapies (e.g. radiation therapy, drug dose heterogeneity).

Current emerging MRI tools for radionecrosis and pseudoprogression diagnosis.

Purpose of reviewThis review aims to cover current MRI techniques for assessing treatment response in brain tumors, with a focus on radio-induced lesions.Recent findingsPseudoprogression and radionecrosis are common radiological entities after brain tumor irradiation and are difficult to distinguish from real progression, with major consequences on daily patient care. To date, shortcomings of conventional MRI have been largely recognized but morphological sequences are still used in official response assessment criteria. Several complementary advanced techniques have been proposed but none of them have been validated, hampering their clinical use. Among advanced MRI, brain perfusion measures increase diagnostic accuracy, especially when added with spectroscopy and susceptibility-weighted imaging. However, lack of reproducibility, because of several hard-to-control variables, is still a major limitation for their standardization in routine protocols. Amide Proton Transfer is an emerging molecular imaging technique that promises to offer new metrics by indirectly quantifying intracellular mobile proteins and peptide concentration. Preliminary studies suggest that this noncontrast sequence may add key biomarkers in tumor evaluation, especially in posttherapeutic settings.SummaryBenefits and pitfalls of conventional and advanced imaging on posttreatment assessment are discussed and the potential added value of APT in this clinicoradiological evolving scenario is introduced.

Predicting the Local Response of Metastatic Brain Tumor to Gamma Knife Radiosurgery by Radiomics With a Machine Learning Method.

PurposeThe current study proposed a model to predict the response of brain metastases (BMs) treated by Gamma knife radiosurgery (GKRS) using a machine learning (ML) method with radiomics features. The model can be used as a decision tool by clinicians for the most desirable treatment outcome.Methods and MaterialUsing MR image data taken by a FLASH (3D fast, low-angle shot) scanning protocol with gadolinium (Gd) contrast-enhanced T1-weighting, the local response (LR) of 157 metastatic brain tumors was categorized into two groups (Group I: responder and Group II: non-responder). We performed a radiomics analysis of those tumors, resulting in more than 700 features. To build a machine learning model, first, we used the least absolute shrinkage and selection operator (LASSO) regression to reduce the number of radiomics features to the minimum number of features useful for the prediction. Then, a prediction model was constructed by using a neural network (NN) classifier with 10 hidden layers and rectified linear unit activation. The training model was evaluated with five-fold cross-validation. For the final evaluation, the NN model was applied to a set of data not used for model creation. The accuracy and sensitivity and the area under the receiver operating characteristic curve (AUC) of the prediction model of LR were analyzed. The performance of the ML model was compared with a visual evaluation method, for which the LR of tumors was predicted by examining the image enhancement pattern of the tumor on MR images.ResultsBy the LASSO analysis of the training data, we found seven radiomics features useful for the classification. The accuracy and sensitivity of the visual evaluation method were 44 and 54%. On the other hand, the accuracy and sensitivity of the proposed NN model were 78 and 87%, and the AUC was 0.87.ConclusionsThe proposed NN model using the radiomics features can help physicians to gain a more realistic expectation of the treatment outcome than the traditional method.

TAMI-07. THE IMMUNE MICROENVIRONMENT IN LOWER GRADE GLIOMAS

Abstract The determinants of the tumor-associated immune response in brain tumors are poorly understood. Using tumor samples from two molecularly distinct subtypes of lower grade glioma, MAPK-driven glioma with biallelic inactivation of CDKN2A (n=30) and IDH-mutant, 1p/19q-intact astrocytoma (n=29), we demonstrate qualitative and quantitative differences in the tumor-associated immune response and we investigate the molecular mechanisms involved. Histologically the MAPK-driven gliomas were comprised of pleomorphic xanthoastrocytoma (PXA) (n=11) and anaplastic PXA (n=19). Seven patients had paired samples from two sequential surgeries. Immune cell populations and their activity were determined by quantitative multiplex immunostaining and Digital Spatial Profiling and gene expression was analyzed by Nanostring. Functional studies were performed using established cell lines and two new patient-derived lines from MAPK-driven LGGs. MAPK-driven tumors exhibited an increased number of CD8+ T cells and tumor-associated microglial/macrophage (TAMs), including CD163+ TAMs, as compared to IDH-mutant astrocytoma. In contrast, IDH-mutant tumors had increased FOXP3+ immunosuppressive T regulatory cells. Transcriptional and protein level analyses in MAPK-driven tumors suggested an active cytotoxic T cell response with robust expression of granzyme B, present on 27% of CD8+ T cells, increased MHC class I expression, and altered cytokine profiles. Interestingly, MAPK-driven tumors also had increased expression of immunosuppressive molecules, including CXCR4, PD-L1, and VEGFA. Expression differences for cell surface and secreted proteins were confirmed in patient-derived tumor lines and functional relationships between altered chemokine expression and immune cell infiltration was investigated. Our data provide novel insights into the immune contexture of MAPK driven LGGs and suggest MAPK driven gliomas with biallelic inactivation of CDKN2A may be particularly vulnerable to immunotherapeutic modulation

IMMU-22. EARLY RELEASE OF TYPE I INTERFERON DRIVES BRAIN CANCER IMMUNOGENICITY AND RESPONSE TO IMMUNOTHERAPY

Abstract BACKGROUND Molecular drivers of cancer immunogenicity in brain tumors are still being unraveled. While BATF3 expression, STING, and interferon response factors (IRFs) are necessary for cancer immunogenicity, the presence of type I interferon (IFN-I) is contextual having been reported to elicit both anti-tumoral and pro-tumoral effects. A better understanding of IFN-I signaling mechanisms is necessary to elucidate drivers of brain cancer immunogenicity and resistance. OBJECTIVE We sought to assess the role of IFN-I signaling in brain tumor immunogenicity and response to immune checkpoint inhibitors (ICIs) in ICI sensitive brain tumor models (i.e. GL261). We then sought to develop strategies to reset IFN-I signaling in ICI resistant brain tumor models (i.e. KR158b). METHODS To reset IFN-I signaling in immunologically ‘cold’ tumors unresponsive to ICIs, we developed lipid-nanoparticles (NPs) to deliver mRNA payloads to the brain tumor microenvironment (TME). RESULTS In immune-sensitive GL261 tumors, we showed that early release of IFN-I unlocks cancer immunogenicity and ICI response. Blockade of IFN-I during tumorigenesis (within 24h, but not days later) increases tumorigenicity and abrogates ICI activity in sensitive tumors. In ICI resistant KR158b tumors, we show that systemic administration of tumor-derived RNA-NPs localize to myeloid cells within the TME for simultaneous activation of multiple innate pathways including BATF3 (necessary for effector DCs), IRF5 (necessary for M2 to M1 macrophage reprograming), and IRF7 (necessary for IFN-I production). These RNA-NPs induce near-immediate release of IFN-I (within hours), reprogram the brain TME in an IFNAR1 (IFN-I receptor) dependent manner, and elicit significant anti-KR158b efficacy as a monotherapy. Following IFNAR1 blockade, RNA-NP mediated anti-tumor efficacy was abrogated. We demonstrated safety of tumor-specific RNA-NPs (derived from KR158b) in acute/chronic GLP toxicity studies without normal-brain cross-reactivity, and confirmed feasibility/safety and immunologic activity in large-animal studies. FUTURE DIRECTIONS We have since received FDA-IND approval for first-in-human trials (IND#BB-19304) in glioblastoma patients.

Immune checkpoint inhibitor response in mismatch repair-deficient colorectal cancer and other solid tumors: is it truly disease-agnostic?

Immune checkpoint inhibitor response in mismatch repair-deficient colorectal cancer and other solid tumors: is it truly disease-agnostic?

Long-term apparent diffusion coefficient value changes in patients undergoing radiosurgical treatment of meningiomas

A noninvasive method to predict the progress or treatment response of meningiomas is desirable to improve the tumor management. Studies showed that apparent diffusion coefficient (ADC) pretreatment values can predict treatment response in brain tumors. The aim of this study was to analyze changes of intratumoral ADC values in patients with meningiomas undergoing conservative or radiosurgery. MR images of 51 patients with diagnose of meningiomas were retrospectively reviewed. Twenty-five patients undergoing conservative or radiosurgery treatment, respectively, were included in the study. The follow-up data ranged between 1 and 10 years. Based on ROI analysis, the mean ADC values, ADC10%min, and ADC90%max were evaluated at different time points during follow-up. Baseline ADC values in between both groups were similar. The ADCmean values, ADC10%min, and ADC90%max within the different groups did not show any significant changes during the follow-up times in the untreated (ADCmean over 10 years period: 0.87 ± 0.05 × 10-3 mm2/s) and radiosurgically treated (ADCmean over 4 years period: 1.02 ± 0.12 × 10-3 mm2/s) group. However, statistically significant difference was observed when comparing the ADCmean and ADC90%max values of untreated with radiosurgically treated (p < 0.0001) meningiomas. Also, ADC10%min revealed statistically significant difference between the untreated and the radiosurgery group (p < 0.05). ADC values in conservatively managed meningiomas remain stable during the follow-up. However, meningiomas undergoing radiosurgery reveal significant change of the mean ADC values over time, suggesting that ADC may reflect a change in the biological behavior of the tumor. These observations might suggest the value of ADC changes as an indicator of treatment response.

Abstract 2256: Low dose brain radiotherapy enhances the efficacy of an extracranial in situ vaccine regimen against melanoma brain metastases in a pre-clinical murine model

Abstract Brain metastases develop in &amp;gt;60% of advanced melanoma patients, resulting in considerable morbidity and mortality. We previously showed that a regimen combining radiation (RT) and intra-tumoral (IT) injection of immunocytokine (IC: αGD2 antibody fused to IL2) resulted in enhanced response to immune checkpoint inhibition (ICI: αCTLA4 Ab) in mice bearing immunologically cold murine melanoma tumors. Here, we used the immunologically cold B78 murine melanoma model to investigate the efficacy of this in situ vaccine (ISV) regimen (RT + IT-IC + ICI) in mice bearing melanoma tumors in the brain and at extracranial sites. B78 (GD2+) cells were implanted into the right and left flanks and the right striatum of the brain in syngeneic C57BL/6 mice. Right flank tumors (150-200 mm3) were treated following a previously optimized ISV regimen [RT (12Gy, day 1), IT-IC (days 6-10) and αCTLA4 (days 3, 6, 9)]. ISV eliminated both flank tumors [22/22 complete response (CR)] compared to αCTLA-4 alone (0/8 CR). However, ISV had only modest effect in prolonging survival in mice with a B78 brain melanoma tumor and all mice succumbed to progressive brain tumor burden. Comparing tumors from the left flank and brain of mice treated with ISV to a right flank tumor, qPCR demonstrated no difference in the expression of Mhc-1 or of various checkpoint receptors/ligands (including Ctla4, Tigit, Pd-L1, Lag3). Immunohistochemistry and flow cytometry at day 15 after ISV demonstrated decreased regulatory T cells (Tregs; CD4+, FOXP3+, CD25+) and a higher ratio of CD8+ T cells:Tregs in the left flank tumor compared to the brain tumor. In contrast, these levels did not differ between left flank and brain tumors in untreated mice. Following ISV, CD8+ T cell:Treg ratio at both the left flank and brain tumor sites significantly increased compared to these respective tumor sites in untreated mice. Multiplex profiling of 32 cytokines demonstrated indistinguishable cytokine expression between untreated tumors in the left flank and brain but showed a distinct pattern of response at these locations following right flank ISV with reduced activation of pro-inflammatory cytokines in the brain melanoma tumor. Given the capacity of low dose RT to temporarily deplete Tregs and increase production of pro-inflammatory cytokines, we hypothesized that low dose whole brain (WB) RT (4 Gy x 1) would enhance ISV response of melanoma brain tumors. Addition of WBRT at day 15 after ISV significantly increased survival of mice with a melanoma tumor in the brain compared to ISV or WBRT alone with 2/11 exhibiting durable CR compared to 0/10 with either ISV or WBRT alone. Interestingly, WBRT administered concurrently on day 1 with ISV had no survival benefit compared to ISV alone - suggesting a critical immunomodulatory role for WBRT at day 15. With WBRT + ISV, we observed an increased ratio of CD8+:FOXP3+ T cells in melanoma brain tumors compared to ISV alone. These results suggest that low dose brain RT may enhance the response to an extracranial ISV at immunologically cold brain metastases. Citation Format: Raghava N. Sriramaneni, Paul A. Clark, Amber M. Bates, Bryce R. Anderson, Wonjong Jin, Justin C. Jagodinsky, Alexander L. Rakhmilevich, Zachary S. Morris. Low dose brain radiotherapy enhances the efficacy of an extracranial in situ vaccine regimen against melanoma brain metastases in a pre-clinical murine model [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2256.

Abstract 6560: Sann-Jhong-Kuey-Jian-Tang could inhibit glioblastoma cells through increasing TNF-α, Fas, Bax but decreasing Bcl-2, Mcl-1 expression

Abstract Glioblastoma is a highly malignant brain tumor and poor response to current chemotherapeutic medicine (such as Temozolomide). Glioblastoma remains a challenging disease and there is a need to identify new treatments. Sann-Jhong-Kuey-Jian-Tang (SJKJT), a traditional Chinese medicine prescription, has been prescribed as complementary medicine for patients with solid tumors in Taiwan. However, the anticancer effects of SJKJT on human brain tumor have not yet been elucidated. The present study focused on the anticancer effects and molecular mechanisms of SJKJT in human Glioblastoma by using DBTRG cells. In the present study, we evaluated the cytotoxic effects of SJKJT on DBTRG cells by MTT assay. The protein expression levels of tumor necrosis factor-α (TNF-α), Fas, Fas-associated Death Domain (FADD), Caspase-8, Caspase-9, Caspase-3, Bcl-2-associated X protein (Bax), Bcl-2 and myeloid cell leukemia 1 protein (Mcl-1) in the DBTRG cells were measured by western blotting. The cell cycles were analyzed by fluorescence-activated cell sorting (FACS). The protein expression of caspase-3 in DBTRG cells was analyzed by immunocytochemical (ICC) method. The results revealed that SJKJT inhibit the proliferation of DBTRG cells in a time- and dose-dependent manner. The protein expression levels of TNF-α, Fas, FADD, Caspase-8, Caspase-9, Caspase-3, Bcl-2-associated X protein (Bax) increased in the DBTRG cells treated with SJKJT; however, the level of Bcl-2 and Mcl-1 were decreased. These results indicated that SJKJT can inhibit DBTRG cells through both extrinsic and intrinsic pathway. The use of traditional Chinese medicine prescription SJKJT may become a feasible therapy option. Further studies are warranted to elucidate its mechanisms of action. Citation Format: Chin Cheng Su. Sann-Jhong-Kuey-Jian-Tang could inhibit glioblastoma cells through increasing TNF-α, Fas, Bax but decreasing Bcl-2, Mcl-1 expression [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6560.

Radial basis function-generated finite difference scheme for simulating the brain cancer growth model under radiotherapy in various types of computational domains

Background and ObjectivesWe extend the original mathematical model, i.e., Swanson’s reaction-diffusion equation to the surfaces with no boundary, and we find a new numerical method based on a meshless approach for solving numerically Swanson’s reaction-diffusion model in the square and on the sphere. MethodsTo solve numerically the Swanson’s reaction-diffusion model and its extension version, a collocation meshless technique, namely radial basis function-generated finite difference (RBF-FD) scheme is employed for approximating the spatial variables in the square domain and on the sphere, respectively. Also, to approximate the time variable of the studied models, a first-order semi-implicit backward Euler scheme is used. The resulting fully discrete scheme is a linear system of algebraic equations per time step that is solved via the biconjugate gradient stabilized (BiCGSTAB) iterative algorithm with a zero-fill incomplete lower-upper (ILU) preconditioner. ResultsThe numerical simulations show the growth of untreated and treated brain tumors with radiotherapy using estimated and clinical data (given from magnetic resonance imaging (MRI) scans of patients). Moreover, the results reported here can be used for improving the treatment strategies of the invasive brain tumor. ConclusionsUsing the developed numerical scheme in this paper, we can simulate the behavior of the invasive form of brain tumor response to radiotherapy. Also, we can see the effects of radiation response on the brain tumor cell concentration of individual patients. The proposed meshless technique, which is applied for solving numerically the studied model, does not depend on any background mesh or triangulation for approximation in comparison with mesh-dependent methods. Moreover, we apply this technique to the sphere via any set of distributed points easily.

3D extracellular matrix microenvironment in bioengineered tissue models of primary pediatric and adult brain tumors

Dynamic alterations in the unique brain extracellular matrix (ECM) are involved in malignant brain tumors. Yet studies of brain ECM roles in tumor cell behavior have been difficult due to lack of access to the human brain. We present a tunable 3D bioengineered brain tissue platform by integrating microenvironmental cues of native brain-derived ECMs and live imaging to systematically evaluate patient-derived brain tumor responses. Using pediatric ependymoma and adult glioblastoma as examples, the 3D brain ECM-containing microenvironment with a balance of cell-cell and cell-matrix interactions supports distinctive phenotypes associated with tumor type-specific and ECM-dependent patterns in the tumor cells’ transcriptomic and release profiles. Label-free metabolic imaging of the composite model structure identifies metabolically distinct sub-populations within a tumor type and captures extracellular lipid-containing droplets with potential implications in drug response. The versatile bioengineered 3D tumor tissue system sets the stage for mechanistic studies deciphering microenvironmental role in brain tumor progression.

Dependency of the blood oxygen level dependent-response to hyperoxic challenges on the order of gas administration in intracranial malignancies.

Literature reports contradicting results on the response of brain tumors to vascular stimuli measured in T2*-weighted MRI. Here, we analyzed the potential dependency of the MRI-response to (hypercapnic) hyperoxia on the order of the gas administration. T2* values were quantified at 3 Tesla in eight consenting patients at rest and during inhalation of hyperoxic/hypercapnic gas mixtures. Patients were randomly divided into two groups undergoing different gas administration protocols (group A: medical air-pure oxygen-carbogen; group B: medical air-carbogen-pure oxygen). Mann-Whitney U test and Wilcoxon signed rank test have been used to proof differences in T2* regarding respiratory challenge or different groups, respectively. T2* values at rest for gray and white matter were 50.3 ± 2.6ms and 46.1 ± 2.0ms, respectively, and slightly increased during challenge. In tumor areas, T2* at rest were: necrosis = 74.1 ± 10.1ms; edema = 60.3 ± 17.6ms; contrast-enhancing lesions = 48.6 ± 20.7ms; and solid T2-hyperintense lesions = 45.0 ± 3.0ms. Contrast-enhancing lesions strongly responded to oxygen (+ 20.7%) regardless on the gas protocol (p = 0.482). However, the response to carbogen significantly depended on the order of gas administration (group A, + 18.6%; group B, - 6.4%, p = 0.042). In edemas, a different trend between group was found when breathing oxygen (group A, - 9.9%; group B, + 19.5%, p = 0.057). Preliminary results show a dependency of the T2* response of contrast-enhancing brain tumor lesions on the order of the gas administration. The gas administration protocol is an important factor in the interpretation of the T2*-response in areas of abnormal vascular growth.

Glioma response assessment: Classic pitfalls, novel confounders, and emerging imaging tools.

Neuroimaging plays a pivotal role in the care of patients with infiltrating gliomas, in whom imaging changes are often the first indications of tumor response or progression. Unfortunately, evaluation of glioma response is often not straightforward, even for experienced radiologists. Post-surgical or radiation-related changes may mimic the appearance of disease progression, while medications such as corticosteroids and antiangiogenic agents may mimic tumor response without truly arresting tumor growth or improving patient survival. Immunotherapy response can result in inflammatory changes which manifest as progressively increasing tumor enhancement and edema over months. Many of these pitfalls can be minimized or avoided altogether by the use of modern brain tumor response criteria, while others will require new imaging tools before they can be fully addressed. Advanced MRI methods and novel positron emission tomography (PET) agents are proving important for this purpose, and their role will undoubtedly continue to grow in the future.

Kinetic studies of complex biomedical process by magnetic resonance: Cuban experiences

The potentials of the magnetic resonance (MR) methods in the research of biomedical systems have been demonstrated during the 70 years of its existence. It is presented that the Cuban experience in quantitative magnetic resonance associated with molecular, preclinical and clinical studies of significant diseases and drugs development. MR “in vitro” and “in vivo” studies of sickle cell disease, the diabetic foot ulcer, the brain tumor response and the magnetic nano-particle pharmacokinetics, are presented as example. Furthermore, contributions and restrictions of magnetic resonance to diagnostic and optimization of therapeutic pathways are discussed in some concrete cases.

Whole brain radiotherapy for the treatment of newly diagnosed multiple brain metastases.

This is an update to the review published in the Cochrane Library (2012, Issue 4).It is estimated that 20% to 40% of people with cancer will develop brain metastases during the course of their illness. The burden of brain metastases impacts quality and length of survival. To assess the effectiveness and adverse effects of whole brain radiotherapy (WBRT) given alone or in combination with other therapies to adults with newly diagnosed multiple brain metastases. We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and Embase to May 2017 and the National Cancer Institute Physicians Data Query for ongoing trials. We included phase III randomised controlled trials (RCTs) comparing WBRT versus other treatments for adults with newly diagnosed multiple brain metastases. Two review authors independently assessed trial quality and abstracted information in accordance with Cochrane methods. We added 10 RCTs to this updated review. The review now includes 54 published trials (45 fully published reports, four abstracts, and five subsets of data from previously published RCTs) involving 11,898 participants.Lower biological WBRT doses versus controlThe hazard ratio (HR) for overall survival (OS) with lower biological WBRT doses as compared with control (3000 cGy in 10 daily fractions) was 1.21 (95% confidence interval (CI) 1.04 to 1.40; P = 0.01; moderate-certainty evidence) in favour of control. The HR for neurological function improvement (NFI) was 1.74 (95% CI 1.06 to 2.84; P = 0.03; moderate-certainty evidence) in favour of control fractionation.Higher biological WBRT doses versus controlThe HR for OS with higher biological WBRT doses as compared with control (3000 cGy in 10 daily fractions) was 0.97 (95% CI 0.83 to 1.12; P = 0.65; moderate-certainty evidence). The HR for NFI was 1.14 (95% CI 0.92 to 1.42; P = 0.23; moderate-certainty evidence).WBRT and radiosensitisersThe addition of radiosensitisers to WBRT did not confer additional benefit for OS (HR 1.05, 95% CI 0.99 to 1.12; P = 0.12; moderate-certainty evidence) or for brain tumour response rates (odds ratio (OR) 0.84, 95% CI 0.63 to 1.11; P = 0.22; high-certainty evidence).Radiosurgery and WBRT versus WBRT aloneThe HR for OS with use of WBRT and radiosurgery boost as compared with WBRT alone for selected participants was 0.61 (95% CI 0.27 to 1.39; P = 0.24; moderate-certainty evidence). For overall brain control at one year, the HR was 0.39 (95% CI 0.25 to 0.60; P < 0.0001; high-certainty evidence) favouring the WBRT and radiosurgery boost group.Radiosurgery alone versus radiosurgery and WBRTThe HR for local brain control was 2.73 (95% CI 1.87 to 3.99; P < 0.00001; high-certainty evidence)favouring the addition of WBRT to radiosurgery. The HR for distant brain control was 2.34 (95% CI 1.73 to 3.18; P < 0.00001; high-certainty evidence) favouring WBRT and radiosurgery. The HR for OS was 1.00 (95% CI 0.80 to 1.25; P = 0.99; moderate-certainty evidence). Two trials reported worse neurocognitive outcomes and one trial reported worse quality of life outcomes when WBRT was added to radiosurgery.We could not pool data from trials related to chemotherapy, optimal supportive care (OSC), molecular targeted agents, neurocognitive protective agents, and hippocampal sparing WBRT. However, one trial reported no differences in quality-adjusted life-years for selected participants with brain metastases from non-small-cell lung cancer randomised to OSC and WBRT versus OSC alone. None of the trials with altered higher biological WBRT dose-fractionation schemes reported benefit for OS, NFI, or symptom control compared with standard care. However, OS and NFI were worse for lower biological WBRT dose-fractionation schemes than for standard dose schedules.The addition of WBRT to radiosurgery improved local and distant brain control in selected people with brain metastases, but data show worse neurocognitive outcomes and no differences in OS.Selected people with multiple brain metastases from non-small-cell lung cancer may show no difference in OS when OSC is given and WBRT is omitted.Use of radiosensitisers, chemotherapy, or molecular targeted agents in conjunction with WBRT remains experimental.Further trials are needed to evaluate the use of neurocognitive protective agents and hippocampal sparing with WBRT. As well, future trials should examine homogeneous participants with brain metastases with focus on prognostic features and molecular markers.

Volumetric response of progressing post-SRS lesions treated with laser interstitial thermal therapy.

We analyzed volumetric response of metastatic brain tumors that progressed despite treatment with stereotactic radiosurgery (SRS) after treatment with laser interstitial thermal therapy (LITT). We retrospectively reviewed consecutive patients treated from 1/2012 to 10/2015 with LITT for metastatic brain tumors demonstrating progression after SRS. Volumes were quantified using MRI with contrast-enhanced T1-weighted (T1W) and fluid-attenuated inversion recovery (FLAIR). Fifty lesions from 36 patients were studied. Lesions were assessed prior to LITT, immediately after LITT, 0-90 days after LITT, 90-180 days after LITT, 180-270 days after LITT, and 270-360 days after LITT. The median T1W volume was 5.05cc (range 0.54-23.31cc) before LITT treatment (n = 50), 7.70cc (range 1.72-38.76cc) 0-90 days after LITT (n = 47), and 3.68cc (range 1.282-48.31cc) 180-270 days after LITT (n = 21). The median FLAIR volume was 43.36cc (range 3.09-233.01cc) before LITT treatment (n = 50), 37.13cc (range 3.48-244.23cc) 0-90 days after LITT (n = 43), 31.68cc (range 1.6-248.75cc) 180-270 days after LITT (n = 18). The 6-month FLAIR volume showed a statistically significant reduction compared to pretreatment (p = 0.04). After selecting for cases where patients had two or more post-operative MRIs, we found that 24 lesions (63%) demonstrated an overall downward trend and 14 lesions (37%) demonstrated an upward trend. The median pre-treatment T1W volume for the patients whose lesions demonstrated volumetric reduction after LITT was 3.54cc (range 0.539-10.06cc) and for those who did not demonstrate volumetric reduction after LITT it was 8.81cc (range 0.926-23.313cc). The pre-treatment tumor volume plays a significant role in determining response to LITT with smaller tumor volumes responding better to LITT than tumors with larger volumes.

Prediction of Local Response After Stereotactic Radiosurgery for Brain Metastases Using Convolutional Neural Networks based radiomics

A deep learning concept based on artificial convolutional neural networks (CNN) is regarded as an emerging radiomics methodology because it uses minimal amount of image preprocessing. The purpose of this study is to predict radiation response of metastatic brain tumor receiving stereotactic radiosurgery (SRS) using the radiomics model based on CNN. The BM lesions that received stereotactic radiosurgery (SRS) were reviewed and 110 tumors were selected for the study. On the planning CT images of the target tumors, a data-set was made by matching the single axial image which presents tumor center and the tumor label. The random assignment was repeated to generate 50 independent experimental combinations of data-sets. The same group is composed of 10 different data-sets combination methods, but the data-sets of the evaluation group are assigned equally. We trained the image of the training group and the matching label on the CNN we implemented. We used the validation data-sets to select the best-performing learning model. The results of this classification were compared to the matched label. CNN is a kind of artificial neural network in which the connectivity pattern between its neurons is inspired by the organization of visual cortex. The CNN of this study is implemented by TensorFlow which is a machine learning framework. The predicted AUC(area under the ROC curve) of individual CNN models for the 50 data-set combinations ranged from 0.602 [95% CI, 36.5% -83.9%] to a maximum of 0.826 [95% CI, 64.3% -100%]. -97.1%] to a maximum of 0.856 [95% CI, 68.2% - 100%]. CNN based ensemble radiomics models that learn BM 's SRS planning CT images and their early responses were able to predict SRS responses with high accuracy for unlearned BM images. This study showed for the first time the applicability of the CNN-based radiomics model to predict the prognosis of radiation therapy with small data.

Residual Deep Convolutional Neural Network Predicts MGMT Methylation Status.

Predicting methylation of the O6-methylguanine methyltransferase (MGMT) gene status utilizing MRI imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare three different residual deep neural network (ResNet) architectures to evaluate their ability in predicting MGMT methylation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture was the best performing model, achieving an accuracy of 94.90% (+/-3.92%) for the test set (classification of a slice as no tumor, methylated MGMT, or non-methylated). ResNet34 (34 layers) achieved 80.72% (+/-13.61%) while ResNet18 (18 layers) accuracy was 76.75% (+/-20.67%). ResNet50 performance was statistically significantly better than both ResNet18 and ResNet34 architectures (p<0.001). We report a method that alleviates the need of extensive preprocessing and acts as a proof of concept that deep neural architectures can be used to predict molecular biomarkers from routine medical images.

Efficacy of afatinib, an irreversible ErbB family blocker, in the treatment of intracerebral metastases of non-small cell lung cancer in mice.

Few effective therapeutic options are currently available for the treatment of non-small cell lung cancer (NSCLC) with brain metastases (BM). Recent evidence shows that NSCLC patients with BMs respond well to afatinib, but little is known about the underlying mechanisms. In this study, we evaluated the efficacy of afatinib in treatment of BMs in mice and investigated whether afatinib could actively penetrate the brain-blood barrier and bind to its target. NSCLC BM model was established in nude mice by intracerebral injection of PC-9.luc cells. The tumors were measured weekly using in vivo quantitative bioluminescence. The mice are administrated afatinib (15, 30 mg·kg-1·d-1, ig) for 14 d. The antitumor efficacy of afatinib was determined by tumor growth inhibition (TGI), which was calculated as [1-(change of tumor volume in treatment group/control group)×100]. Pharmacokinetic characteristics were measure in mice receiving a single dose of afatinib (30 mg/kg, ig). Pharmacodynamics of afatinib was also assessed by detecting the expression of pEGFR (Tyr1068) in brain tumor foci using immunohistochemistry. Administration of afatinib (15, 30 mg·kg-1·d-1) dose-dependently inhibited PC-9 tumor growth in the brain with a TGI of 90.2% and 105%, respectively, on d 14. After administration of afatinib (30 mg/kg), the plasma concentration of afatinib was 91.4±31.2 nmol/L at 0.5 h, reached a peak (417.1±119.9 nmol/L) at 1 h, and was still detected after 24 h. The cerebrospinal fluid (CSF) concentrations followed a similar pattern. The T1/2 values of afatinib in plasma and CSF were 5.0 and 3.7 h, respectively. The AUC(0-24 h) values for plasma and CSF were 2375.5 and 29.1 nmol/h, respectively. The plasma and CSF concentrations were correlated (r=0.844, P<0.01). Pharmacodynamics study showed that the expression levels of pEGFR were reduced by 90% 1 h after afatinib administration. The Emax was 86.5%, and the EC50 was 0.26 nmol/L. A positive correlation between CSF concentrations and pEGFR modulation was revealed. Afatinib penetrates the BBB in NSCLC BM mice and contributes to the brain tumor response. The CSF exposure level is correlated with the plasma level, which in turn is correlated with the modulation of pEGFR in the tumor tissues. The results support for the potential application of afatinib in NSCLC patients with BMs.