Rushin H. Gindra, Giovanni Palla, Mathias Nguyen
et al.
Spatial transcriptomics enables simultaneous measurement of gene expression and tissue morphology, offering unprecedented insights into cellular organization and disease mechanisms. However, the field lacks comprehensive benchmarks for evaluating multimodal learning methods that leverage both histology images and gene expression data. Here, we present HESCAPE, a large-scale benchmark for cross-modal contrastive pretraining in spatial transcriptomics, built on a curated pan-organ dataset spanning 6 different gene panels and 54 donors. We systematically evaluated state-of-the-art image and gene expression encoders across multiple pretraining strategies and assessed their effectiveness on two downstream tasks: gene mutation classification and gene expression prediction. Our benchmark demonstrates that gene expression encoders are the primary determinant of strong representational alignment, and that gene models pretrained on spatial transcriptomics data outperform both those trained without spatial data and simple baseline approaches. However, downstream task evaluation reveals a striking contradiction: while contrastive pretraining consistently improves gene mutation classification performance, it degrades direct gene expression prediction compared to baseline encoders trained without cross-modal objectives. We identify batch effects as a key factor that interferes with effective cross-modal alignment. Our findings highlight the critical need for batch-robust multimodal learning approaches in spatial transcriptomics. To accelerate progress in this direction, we release HESCAPE, providing standardized datasets, evaluation protocols, and benchmarking tools for the community
Gianluca Carloni, Biagio Brattoli, Seongho Keum
et al.
Computational pathology (CPath) has shown great potential in mining actionable insights from Whole Slide Images (WSIs). Deep Learning (DL) has been at the center of modern CPath, and while it delivers unprecedented performance, it is also known that DL may be affected by irrelevant details, such as those introduced during scanning by different commercially available scanners. This may lead to scanner bias, where the model outputs for the same tissue acquired by different scanners may vary. In turn, it hinders the trust of clinicians in CPath-based tools and their deployment in real-world clinical practices. Recent pathology Foundation Models (FMs) promise to provide better domain generalization capabilities. In this paper, we benchmark FMs using a multi-scanner dataset and show that FMs still suffer from scanner bias. Following this observation, we propose ScanGen, a contrastive loss function applied during task-specific fine-tuning that mitigates scanner bias, thereby enhancing the models' robustness to scanner variations. Our approach is applied to the Multiple Instance Learning task of Epidermal Growth Factor Receptor (EGFR) mutation prediction from H\&E-stained WSIs in lung cancer. We observe that ScanGen notably enhances the ability to generalize across scanners, while retaining or improving the performance of EGFR mutation prediction.
Colorectal cancer (CRC) poses a major public health challenge due to its increasing prevalence, particularly among younger populations. Microsatellite instability-high (MSI-H) CRC and deficient mismatch repair (dMMR) CRC constitute 15% of all CRC and exhibit remarkable responsiveness to immunotherapy, especially with PD-1 inhibitors. Despite this, there is a significant need to optimise immunotherapeutic regimens to maximise clinical efficacy and patient quality of life whilst minimising monetary costs. To address this, we employ a novel framework driven by delay integro-differential equations to model the interactions among cancer cells, immune cells, and immune checkpoints. Several of these components are being modelled deterministically for the first time in cancer, paving the way for a deeper understanding of the complex underlying immune dynamics. We consider two compartments: the tumour site and the tumour-draining lymph node, incorporating phenomena such as dendritic cell (DC) migration, T cell proliferation, and CD8+ T cell exhaustion and reinvigoration. Parameter values and initial conditions are derived from experimental data, integrating various pharmacokinetic, bioanalytical, and radiographic studies, along with deconvolution of bulk RNA-sequencing data from the TCGA COADREAD and GSE26571 datasets. We finally optimised neoadjuvant treatment with pembrolizumab, a widely used PD-1 inhibitor, to balance efficacy, efficiency, and toxicity in locally advanced MSI-H/dMMR CRC patients. We mechanistically analysed factors influencing treatment success and improved upon currently FDA-approved therapeutic regimens for metastatic MSI-H/dMMR CRC, demonstrating that a single medium-to-high dose of pembrolizumab may be sufficient for effective tumour eradication while being efficient, safe and practical.
The surface morphology of the developing mammalian brain is crucial for understanding brain function and dysfunction. Computational modeling offers valuable insights into the underlying mechanisms for early brain folding. Recent findings indicate significant regional variations in brain tissue growth, while the role of these variations in cortical development remains unclear. In this study, we unprecedently explored how regional cortical growth affects brain folding patterns using computational simulation. We first developed growth models for typical cortical regions using machine learning (ML)-assisted symbolic regression, based on longitudinal real surface expansion and cortical thickness data from prenatal and infant brains derived from over 1,000 MRI scans of 735 pediatric subjects with ages ranging from 29 post-menstrual weeks to 24 months. These models were subsequently integrated into computational software to simulate cortical development with anatomically realistic geometric models. We comprehensively quantified the resulting folding patterns using multiple metrics such as mean curvature, sulcal depth, and gyrification index. Our results demonstrate that regional growth models generate complex brain folding patterns that more closely match actual brains structures, both quantitatively and qualitatively, compared to conventional uniform growth models. Growth magnitude plays a dominant role in shaping folding patterns, while growth trajectory has a minor influence. Moreover, multi-region models better capture the intricacies of brain folding than single-region models. Our results underscore the necessity and importance of incorporating regional growth heterogeneity into brain folding simulations, which could enhance early diagnosis and treatment of cortical malformations and neurodevelopmental disorders such as cerebral palsy and autism.
Purpose: Microaneurysms (MAs) have distinct, oval-shaped, hyperreflective walls on structural OCT, and inconsistent flow signal in the lumen with OCT angiography (OCTA). Their relationship to regional macular edema in diabetic retinopathy (DR) has not been quantitatively explored. Participants: A total of 99 participants, including 23 with mild, NPDR, 25 with moderate NPDR, 34 with severe NPDR, and 17 with proliferative DR. Methods: We obtained 3 x 3-mm scans with a commercial device (Solix, Visionix/Optovue) in 99 patients with DR. Trained graders manually identified MAs and their location relative to the anatomic layers from cross-sectional OCT. Microaneurysms were first classified as perfused if flow signal was present in the OCTA channel. Then, perfused MAs were further classified into fully and partially perfused MAs based on the flow characteristics in en face OCTA. The presence of retinal fluid based on OCT near MAs was compared between perfused and nonperfused types. We also compared OCT-based MA detection to fundus photography (FP)- and fluorescein angiography (FA)-based detection. Results: We identified 308 MAs (166 fully perfused, 88 partially perfused, 54 nonperfused) in 42 eyes using OCT and OCTA. Nearly half of the MAs identified in this study straddle the inner nuclear layer and outer plexiform layer. Compared with partially perfused and nonperfused MAs, fully perfused MAs were more likely to be associated with local retinal fluid. The associated fluid volumes were larger with fully perfused MAs compared with other types. OCT/OCTA detected all MAs found on FP. Although not all MAs seen with FA were identified with OCT, some MAs seen with OCT were not visible with FA or FP. Conclusions: OCT-identified MAs with colocalized flow on OCTA are more likely to be associated with DME than those without flow.
Leo Pio-Lopez, Johanna Bischof, Jennifer V. LaPalme
et al.
All cognitive agents are composite beings. Specifically, complex living agents consist of cells, which are themselves competent sub-agents navigating physiological and metabolic spaces. Behavior science, evolutionary developmental biology, and the field of machine intelligence all seek an answer to the scaling of biological cognition: what evolutionary dynamics enable individual cells to integrate their activities to result in the emergence of a novel, higher-level intelligence that has goals and competencies that belong to it and not to its parts? Here, we report the results of simulations based on the TAME framework, which proposes that evolution pivoted the collective intelligence of cells during morphogenesis of the body into traditional behavioral intelligence by scaling up the goal states at the center of homeostatic processes. We tested the hypothesis that a minimal evolutionary framework is sufficient for small, low-level setpoints of metabolic homeostasis in cells to scale up into collectives (tissues) which solve a problem in morphospace: the organization of a body-wide positional information axis (the classic French Flag problem). We found that these emergent morphogenetic agents exhibit a number of predicted features, including the use of stress propagation dynamics to achieve its target morphology as well as the ability to recover from perturbation (robustness) and long-term stability (even though neither of these was directly selected for). Moreover we observed unexpected behavior of sudden remodeling long after the system stabilizes. We tested this prediction in a biological system - regenerating planaria - and observed a very similar phenomenon. We propose that this system is a first step toward a quantitative understanding of how evolution scales minimal goal-directed behavior (homeostatic loops) into higher-level problem-solving agents in morphogenetic and other spaces.
Xianghao Zhan, Yuzhe Liu, Nicholas J. Cecchi
et al.
Strain and strain rate are effective traumatic brain injury predictors. Kinematics-based models estimating these metrics suffer from significant different distributions of both kinematics and the injury metrics across head impact types. To address this, previous studies focus on the kinematics but not the injury metrics. We have previously shown the kinematic features vary largely across head impact types, resulting in different patterns of brain deformation. This study analyzes the spatial distribution of brain deformation and applies principal component analysis (PCA) to extract the representative patterns of injury metrics (maximum principal strain (MPS), MPS rate (MPSR) and MPSXMPSR) in four impact types (simulation, football, mixed martial arts and car crashes). We apply PCA to decompose the patterns of the injury metrics for all impacts in each impact type, and investigate the distributions among brain regions using the first principal component (PC1). Furthermore, we developed a deep learning head model (DLHM) to predict PC1 and then inverse-transform to predict for all brain elements. PC1 explained >80% variance on the datasets. Based on PC1 coefficients, the corpus callosum and midbrain exhibit high variance on all datasets. We found MPSXMPSR the most sensitive metric on which the top 5% of severe impacts further deviates from the mean and there is a higher variance among the severe impacts. Finally, the DLHM reached mean absolute errors of <0.018 for MPS, <3.7 (1/s) for MPSR and <1.1 (1/s) for MPSXMPSR, much smaller than the injury thresholds. The brain injury metric in a dataset can be decomposed into mean components and PC1 with high explained variance. The brain dynamics decomposition enables better interpretation of the patterns in brain injury metrics and the sensitivity of brain injury metrics across impact types. The decomposition also reduces the dimensionality of DLHM.
Marta Félez-Sánchez, Marleny Vergara, Silvia de Sanjosé
et al.
Human Papillomaviruses (HPVs) are involved in the etiology of anogenital and head and neck cancers. The HPV DNA prevalence greatly differs by anatomical site. Indeed, the high rates of viral DNA prevalence in anal and cer-vical carcinomas contrast with the lower fraction of cancer cases attributable to HPVs in other anatomical sites, chiefly the vulva, the penis and head and neck. Here we analyzed 2635 Formalin Fixed Paraffin Embedded surgical samples that had previously tested negative for the presence of HPVs DNA using the SPF10/DEIA procedure, in order to identify the presence of other PVs not explicitly targeted by standard molecular epidemiologic approaches. All samples were reanalyzed using five broad-PV PCR primer sets (CP1/2, FAP6064/FAP64, SKF/SKR, MY9/MY11, MFI/MFII) targeting the main PV main clades. In head and neck carcinoma samples (n = 1141), we recovered DNA from two BetaHPVs, namely HPV20 and HPV21, and from three cutaneous AlphaPVs, namely HPV2, HPV57 and HPV61. In vulvar squamous cell carcinoma samples (n = 902), we found one of the samples containing DNA of one cutaneous HPV, namely HPV2, and 29 samples contained DNA from essentially mucosal HPVs. In penile squamous cell carcinoma samples (n = 592), we retrieved the DNA of HPV16 in 16 samples. Our results show first that the SPF10/DEIA is very sensitive, as we recovered only 2.1% (55/2635) false negative results; second, that although the DNA of cutaneous HPVs can be detected in cancer samples, their relative contribution remains anyway minor (0.23%; 6/2635) and may be neglected for screening and vaccination purposes; and third, their contribution to malignancy is not necessarily warranted and needs to be elucidated.
Aude Castel, Patrick M. Burns, Javier Benito
et al.
Background: The chronically instrumented pregnant sheep has been used as a model of human fetal development and responses to pathophysiologic stimuli. This is due to the unique amenability of the unanesthetized fetal sheep to the surgical placement and maintenance of catheters and electrodes, allowing repetitive blood sampling, substance injection, recording of bioelectrical activity, application of electric stimulation and in vivo organ imaging. Recently, there has been growing interest in pleiotropic effects of vagus nerve stimulation (VNS) on various organ systems such as innate immunity, metabolism, and appetite control. There is no approach to study this in utero and corresponding physiological understanding is scarce. New Method: Based on our previous presentation of a stable chronically instrumented unanesthetized fetal sheep model, here we describe the surgical instrumentation procedure allowing successful implantation of a cervical uni- or bilateral VNS probe with or without vagotomy. Results: In a cohort of 53 animals, we present the changes in blood gas, metabolic, and inflammatory markers during the postoperative period. We detail the design of a VNS probe which also allows recording from the nerve. We also present an example of vagus electroneurogram (VENG) recorded from the VNS probe and an analytical approach to the data. Comparison with Existing Methods: This method represents the first implementation of VENG/VNS in a large pregnant mammalian organism. Conclusions: This study describes a new surgical procedure allowing to record and manipulate chronically the vagus nerve activity in an animal model of human pregnancy.
Nora Bloise, Loredana Petecchia, Gabriele Ceccarelli
et al.
Human bone marrow-derived mesenchymal stem cells (hBM-MSCs) are considered a great promise in the repair and regeneration of bone. Considerable efforts have been oriented towards uncovering the best strategy to promote stem cells osteogenic differentiation. In previous studies, hBM-MSCs exposed to physical stimuli such as pulsed electromagnetic fields (PEMFs) or directly seeded on nanostructured titanium surfaces ($TiO_{2}$) were shown to improve their differentiation to osteoblasts in osteogenic condition. In the present study, the effect of a daily PEMF-exposure on osteogenic differentiation of hBM-MSCs seeded onto nanostructured $TiO_{2}$ (with clusters under 100 nm of dimension) was investigated. $TiO_{2}$-seeded cells were exposed to PEMF (magnetic field intensity: 2 mT; intensity of induced electric field: 5 mV; frequency: 75 Hz) and examined in terms of cell physiology modifications and osteogenic differentiation. Results showed that PEMF exposure affected $TiO_{2}$-seeded cells osteogenesis by interfering with selective calcium-related osteogenic pathways, and greatly enhanced hBM-MSCs osteogenic features such as the expression of early/late osteogenic genes and protein production (e.g., ALP, COL-I, osteocalcin and osteopontin) and ALP activity. Finally, PEMF-treated cells resulted to secrete into conditioned media higher amounts of BMP-2, DCN and COL-I than untreated cell cultures. These findings confirm once more the osteoinductive potential of PEMF, suggesting that its combination with $TiO_{2}$ nanostructured surface might be a great option in bone tissue engineering applications.
The set of T cells that express the same T cell receptor (TCR) sequence represents a T cell clone. The number of different naive T cell clones in an organism reflects the number of different T cell receptors (TCRs) arising from recombination of the V(D)J gene segments during T cell development in the thymus. TCR diversity and more specifically, the clone abundance distribution is an important factor in immune function. Specific recombination patterns occur more frequently than others while subsequent interactions between TCRs and self-antigens are known to trigger proliferation and sustain naive T cell survival. These processes are TCR-dependent, leading to clone-dependent thymic export and naive T cell proliferation rates. Using a mean-field approximation to the solution of a regulated birth-death-immigration model and a modification arising from sampling, we systematically quantify how TCR-dependent heterogeneities in immigration and proliferation rates affect the shape of clone abundance distributions (the number of different clones that are represented by a specific number of cells, or "clone counts"). By comparing predicted clone counts derived from our heterogeneous birth-death-immigration model with experimentally sampled clone abundances, we show that although heterogeneity in immigration rates causes very little change to predicted clone-counts, significant heterogeneity in proliferation rates is necessary to generate the observed abundances with reasonable physiological parameter values. Our analysis provides constraints among physiological parameters that are necessary to yield predictions that qualitatively match the data. Assumptions of the model and potentially other important mechanistic factors are discussed.
Okyaz Eminaga, Mahmoud Abbas, Christian Kunder
et al.
Different convolutional neural network (CNN) models have been tested for their application in histological image analyses. However, these models are prone to overfitting due to their large parameter capacity, requiring more data or valuable computational resources for model training. Given these limitations, we introduced a novel architecture (termed PlexusNet). We utilized 310 Hematoxylin and Eosin stained (H&E) annotated histological images of prostate cancer cases from TCGA-PRAD and Stanford University and 398 H&E whole slides images from the Camelyon 2016 challenge. PlexusNet-architecture -derived models were compared to models derived from several existing "state of the art" architectures. We measured discrimination accuracy, calibration, and clinical utility. An ablation study was conducted to study the effect of each component of PlexusNet on model performance. A well-fitted PlexusNet-based model delivered comparable classification performance (AUC: 0.963) in distinguishing prostate cancer from healthy tissues, although it was at least 23 times smaller, had a better model calibration and clinical utility than the comparison models. A separate smaller PlexusNet model accurately detected slides with breast cancer metastases (AUC: 0.978); it helped reduce the slide number to examine by 43.8% without consequences, although its parameter capacity was 200 times smaller than ResNet18. We found that the partitioning of the development set influences the model calibration for all models. However, with PlexusNet architecture, we could achieve comparable well-calibrated models trained on different partitions. In conclusion, PlexusNet represents a novel model architecture for histological image analysis that achieves classification performance comparable to other models while providing orders-of-magnitude parameter reduction.
Julie A Cass, C. Dave Williams, Tom C Irving
et al.
During muscle contraction, myosin motors anchored to thick filaments bind to and slide actin thin filaments. These motors rely on energy derived from ATP, supplied, in part, by diffusion from the sarcoplasm to the interior of the lattice of actin and myosin filaments. The radial spacing of filaments in this lattice may change or remain constant during contraction. If the lattice is isovolumetric, it must expand when the muscle shortens. If, however, the spacing is constant or has a different pattern of axial and radial motion, then the lattice changes volume during contraction, driving fluid motion and assisting in the transport of molecules between the contractile lattice and the surrounding intracellular space. We first create an advective-diffusive-reaction flow model and show that the flow into and out of the sarcomere lattice would be significant in the absence of lattice expansion. Advective transport coupled to diffusion has the potential to substantially enhance metabolite exchange within the crowded sarcomere. Using time-resolved x-ray diffraction of contracting muscle, we next show that the contractile lattice is neither isovolumetric nor constant in spacing. Instead, lattice spacing is time-varying, depends on activation, and can manifest as an effective time-varying Poisson ratio. The resulting fluid flow in the sarcomere lattice of synchronous insect flight muscles is greater than expected for constant lattice spacing conditions. Lattice spacing depends on a variety of factors that produce radial force, including crossbridges, titin-like molecules, and other structural proteins. Volume change and advective transport varies with the phase of muscle stimulation but remains significant at all conditions. Akin to "breathing," advective-diffusive transport in sarcomeres is sufficient to promote metabolite exchange and may play a role in the regulation of contraction itself.
Olufemi E. Kadri, Cortes Williams, Vassilios Sikavitsas
et al.
Flow-induced shear stresses have been found to be a stimulatory factor in pre-osteoblastic cells seeded in 3D porous scaffolds and cultured under continuous flow perfusion. However, due to the complex internal structure of the scaffolds, whole scaffold calculations of the local shear forces are computationally-intensive. Instead, representative volume elements (RVEs), which are obtained by extracting smaller portions of the scaffold, are commonly used in literature without a numerical accuracy standard. Hence, the goal of this study is to examine how closely the whole scaffold simulations are approximated by the two types of boundary conditions used to enable the RVEs: "wall boundary condition" (WBC) and "periodic boundary condition" (PBC). To that end, Lattice-Boltzmann Method fluid dynamics simulations were used to model the surface shear stresses in 3D scaffold reconstructions, obtained from high resolution microcomputed tomography images. It was found that despite the RVEs being sufficiently larger than 6 times the scaffold pore size (which is the only accuracy guideline found in literature), the stresses were still significantly under-predicted by both types of boundary conditions: between 20 and 80% average error, depending on the scaffold's porosity. Moreover, it was found that the error grew with higher porosity. This is likely due to the small pores dominating the flow field, and thereby negating the effects of the unrealistic boundary conditions, when the scaffold porosity is small. Finally, it was found that the PBC was always more accurate and computationally efficient than the WBC. Therefore, it is the recommended type of RVE. Overall, this work provides a previously-unavailable guidance to researchers regarding the best choice of boundary conditions for RVE simulations.
Oxidation-associated malondialdehyde (MDA) modification of proteins can generate immunogenic neo-epitopes that are recognized by autoantibodies. In health, IgM antibodies to MDA-adducts are part of the natural antibody pool, while elevated levels of IgG anti-MDA are associated with inflammatory conditions. Yet, in human autoimmune disease IgG anti-MDA responses have not been well characterized and their potential contribution to disease pathogenesis is not known. Here, we investigate MDA-modifications and anti-MDA-modified protein autoreactivity in rheumatoid arthritis (RA). While RA is primarily associated with autoreactivity to citrullinated antigens, we also observed increases in serum IgG anti-MDA in RA patients compared to controls. IgG anti-MDA levels significantly correlated with disease activity by DAS28-ESR and serum TNF-alpha, IL-6, and CRP. Mass spectrometry analysis of RA synovial tissue identified MDA-modified proteins and revealed shared peptides between MDA-modified and citrullinated actin and vimentin. Furthermore, anti-MDA autoreactivity among synovial B cells was discovered when investigating recombinant monoclonal antibodies (mAbs) cloned from single B cells. Several clones were highly specific for MDA-modification with no cross-reactivity to other antigen modifications. The mAbs recognized MDA-adducts in a variety of proteins. Interestingly, the most reactive clone, originated from an IgG1-bearing memory B cell, was encoded by germline variable genes, and showed similarity to previously reported natural IgM. Other anti-MDA clones display somatic hypermutations and lower reactivity. These anti-MDA antibodies had significant in vitro functional properties and induced enhanced osteoclastogenesis, while the natural antibody related high-reactivity clone did not. We postulate that these may represent distinctly different facets of anti-MDA autoreactive responses.
Holger Perfahl, Barry D Hughes, Tomas Alaracon
et al.
We develop an agent-based model of vasculogenesis, the de novo formation of blood vessels. Endothelial cells in the vessel network are viewed as linearly elastic spheres and are of two types: vessel elements are contained within the network; tip cells are located at endpoints. Tip cells move in response to forces due to interactions with neighbouring vessel elements, the local tissue environment, chemotaxis and a persistence force modeling their tendency to continue moving in the same direction. Vessel elements experience similar forces but not chemotaxis. An angular persistence force representing local tissue interactions stabilises buckling instabilities due to proliferation. Vessel elements proliferate, at rates that depend on their degree of stretch: elongated elements proliferate more rapidly than compressed elements. Following division, new cells are more likely to form new sprouts if the parent vessel is highly compressed and to be incorporated into the parent vessel if it is stretched. Model simulations reproduce key features of vasculogenesis. Parameter sensitivity analyses reveal significant changes in network size and morphology on varying the chemotactic sensitivity of tip cells, and the sensitivities of the proliferation rate and sprouting probability to mechanical stretch. Varying chemotactic sensitivity also affects network directionality. Branching and network density are influenced by the sprouting probability. Glyphs depicting multiple network properties show how network quantities change over time and as model parameters vary. We also show how glyphs constructed from in vivo data could be used to discriminate between normal and tumour vasculature and, ultimately, for model validation. We conclude that our biomechanical hybrid model generates vascular networks similar to those generated from in vitro and in vivo experiments.
Glioma is the most common form of primary brain tumor. Demographically, the risk of occurrence increases until old age. Here we present a novel computational model to reproduce the probability of glioma incidence across the lifespan. Previous mathematical models explaining glioma incidence are framed in a rather abstract way, and do not directly relate to empirical findings. To decrease this gap between theory and experimental observations, we incorporate recent data on cellular and molecular factors underlying gliomagenesis. Since evidence implicates the adult neural stem cell as the likely cell-of-origin of glioma, we have incorporated empirically-determined estimates of neural stem cell number, cell division rate, mutation rate and oncogenic potential into our model. We demonstrate that our model yields results which match actual demographic data in the human population. In particular, this model accounts for the observed peak incidence of glioma at approximately 80 years of age, without the need to assert differential susceptibility throughout the population. Overall, our model supports the hypothesis that glioma is caused by randomly-occurring oncogenic mutations within the neural stem cell population. Based on this model, we assess the influence of the (experimentally indicated) decrease in the number of neural stem cells and increase of cell division rate during aging. Our model provides multiple testable predictions, and suggests that different temporal sequences of oncogenic mutations can lead to tumorigenesis. Finally, we conclude that four or five oncogenic mutations are sufficient for the formation of glioma.