Search

Motivation In many biomedical applications, we are confronted with paired groups of samples, such as treated versus control. The aim is to detect discriminating features, i.e. biomarkers, based on high-dimensional (omics-) data. This problem can be phrased more generally as a two-sample problem requiring statistical significance testing to establish differences , and interpretations to identify distinguishing features. The multivariate maximum mean discrepancy (MMD) test quantifies group-level differences, whereas statistically significantly associated features are usually found by univariate feature selection. Currently, few general-purpose methods simultaneously perform multivariate feature selection and two-sample testing. Results We introduce a sparse, interpretable, and optimized MMD test (SpInOpt-MMD) that enables two-sample testing and feature selection in the same experiment. SpInOpt-MMD is a versatile method and we demonstrate its application to a variety of synthetic and real-world data types including images, gene expression measurements, and text data. SpInOpt-MMD is effective in identifying relevant features in small sample sizes and outperforms other feature selection methods such as SHapley Additive exPlanations and univariate association analysis in several experiments. Availability and implementation The code and links to our public data are available at https://github.com/BorgwardtLab/spinoptmmd. [ABSTRACT FROM AUTHOR]

Mechanistic learning refers to the synergistic combination of mechanistic mathematical modeling and data-driven machine or deep learning. This emerging field finds increasing applications in (mathematical) oncology. This review aims to capture the current state of the field and provides a perspective on how mechanistic learning may progress in the oncology domain. We highlight the synergistic potential of mechanistic learning and point out similarities and differences between purely data-driven and mechanistic approaches concerning model complexity, data requirements, outputs generated, and interpretability of the algorithms and their results. Four categories of mechanistic learning (sequential, parallel, extrinsic, intrinsic) of mechanistic learning are presented with specific examples. We discuss a range of techniques including physics-informed neural networks, surrogate model learning, and digital twins. Example applications address complex problems predominantly from the domain of oncology research such as longitudinal tumor response predictions or time-to-event modeling. As the field of mechanistic learning advances, we aim for this review and proposed categorization framework to foster additional collaboration between the data- and knowledge-driven modeling fields. Further collaboration will help address difficult issues in oncology such as limited data availability, requirements of model transparency, and complex input data which are embraced in a mechanistic learning framework. [ABSTRACT FROM AUTHOR]

Background: In the last decades, medical research fields studying rare conditions such as spinal cord injury (SCI) have made extensive efforts to collect large-scale data. However, most analysis methods rely on complete data. This is particularly troublesome when studying clinical data as they are prone to missingness. Often, researchers mitigate this problem by removing patients with missing data from the analyses. Less commonly, imputation methods to infer likely values are applied. Objective: Our objective was to study how handling missing data influences the results reported, taking the example of SCI registries. We aimed to raise awareness on the effects of missing data and provide guidelines to be applied for future research projects, in SCI research and beyond. Methods: Using the Sygen clinical trial data (n = 797), we analyzed the impact of the type of variable in which data is missing, the pattern according to which data is missing, and the imputation strategy (e.g. mean imputation, last observation carried forward, multiple imputation). Results: Our simulations show that mean imputation may lead to results strongly deviating from the underlying expected results. For repeated measures missing at late stages (> = 6 months after injury in this simulation study), carrying the last observation forward seems the preferable option for the imputation. This simulation study could show that a one-size-fit-all imputation strategy falls short in SCI data sets. Conclusions: Data-tailored imputation strategies are required (e.g., characterisation of the missingness pattern, last observation carried forward for repeated measures evolving to a plateau over time). Therefore, systematically reporting the extent, kind and decisions made regarding missing data will be essential to improve the interpretation, transparency, and reproducibility of the research presented. [ABSTRACT FROM AUTHOR]

Purpose: Whilst survival in paediatric critical care has improved, clinicians lack tools capable of predicting long-term outcomes. We developed a machine learning model to predict poor school outcomes in children surviving intensive care unit (ICU). Methods: Population-based study of children < 16 years requiring ICU admission in Queensland, Australia, between 1997 and 2019. Failure to meet the National Minimum Standard (NMS) in the National Assessment Program-Literacy and Numeracy (NAPLAN) assessment during primary and secondary school was the primary outcome. Routine ICU information was used to train machine learning classifiers. Models were trained, validated and tested using stratified nested cross-validation. Results: 13,957 childhood ICU survivors with 37,200 corresponding NAPLAN tests after a median follow-up duration of 6 years were included. 14.7%, 17%, 15.6% and 16.6% failed to meet NMS in school grades 3, 5, 7 and 9. The model demonstrated an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.8 (standard deviation SD, 0.01), with 51% specificity to reach 85% sensitivity [relative Area Under the Precision Recall Curve (rel-AUPRC) 3.42, SD 0.06]. Socio-economic status, illness severity, and neurological, congenital, and genetic disorders contributed most to the predictions. In children with no comorbidities admitted between 2009 and 2019, the model achieved a AUROC of 0.77 (SD 0.03) and a rel-AUPRC of 3.31 (SD 0.42). Conclusions: A machine learning model using data available at time of ICU discharge predicted failure to meet minimum educational requirements at school age. Implementation of this prediction tool could assist in prioritizing patients for follow-up and targeting of rehabilitative measures. [ABSTRACT FROM AUTHOR]

Motivation With the steadily increasing abundance of omics data produced all over the world under vastly different experimental conditions residing in public databases, a crucial step in many data-driven bioinformatics applications is that of data integration. The challenge of batch-effect removal for entire databases lies in the large number of batches and biological variation, which can result in design matrix singularity. This problem can currently not be solved satisfactorily by any common batch-correction algorithm. Results We present reComBat , a regularized version of the empirical Bayes method to overcome this limitation and benchmark it against popular approaches for the harmonization of public gene-expression data (both microarray and bulkRNAsq) of the human opportunistic pathogen Pseudomonas aeruginosa. Batch-effects are successfully mitigated while biologically meaningful gene-expression variation is retained. reComBat fills the gap in batch-correction approaches applicable to large-scale, public omics databases and opens up new avenues for data-driven analysis of complex biological processes beyond the scope of a single study. Availability and implementation The code is available at https://github.com/BorgwardtLab/reComBat , all data and evaluation code can be found at https://github.com/BorgwardtLab/batchCorrectionPublicData. Supplementary information Supplementary data are available at Bioinformatics Advances online. [ABSTRACT FROM AUTHOR]

Transmission chains within small urban areas (accommodating ∼30 per cent of the European population) greatly contribute to case burden and economic impact during the ongoing coronavirus pandemic and should be a focus for preventive measures to achieve containment. Here, at very high spatio-temporal resolution, we analysed determinants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) transmission in a European urban area, Basel-City (Switzerland). We combined detailed epidemiological, intra-city mobility and socio-economic data sets with whole-genome sequencing during the first SARS-CoV-2 wave. For this, we succeeded in sequencing 44 per cent of all reported cases from Basel-City and performed phylogenetic clustering and compartmental modelling based on the dominating viral variant (B.1-C15324T; 60 per cent of cases) to identify drivers and patterns of transmission. Based on these results we simulated vaccination scenarios and corresponding healthcare system burden (intensive care unit (ICU) occupancy). Transmissions were driven by socio-economically weaker and highly mobile population groups with mostly cryptic transmissions which lacked genetic and identifiable epidemiological links. Amongst more senior population transmission was clustered. Simulated vaccination scenarios assuming 60–90 per cent transmission reduction and 70–90 per cent reduction of severe cases showed that prioritising mobile, socio-economically weaker populations for vaccination would effectively reduce case numbers. However, long-term ICU occupation would also be effectively reduced if senior population groups were prioritised, provided there were no changes in testing and prevention strategies. Reducing SARS-CoV-2 transmission through vaccination strongly depends on the efficacy of the deployed vaccine. A combined strategy of protecting risk groups by extensive testing coupled with vaccination of the drivers of transmission (i.e. highly mobile groups) would be most effective at reducing the spread of SARS-CoV-2 within an urban area. [ABSTRACT FROM AUTHOR]

Recurrent high grade glioma patients face a poor prognosis for which no curative treatment option currently exists. In contrast to prescribing high dose hypofractionated stereotactic radiotherapy (HFSRT, ≥ 6 Gy × 5 in daily fractions) with debulking intent, we suggest a personalized treatment strategy to improve tumor control by delivering high dose intermittent radiation treatment (iRT, ≥ 6 Gy × 1 every 6 weeks). We performed a simulation analysis to compare HFSRT, iRT and iRT plus boost ( ≥ 6 Gy × 3 in daily fractions at time of progression) based on a mathematical model of tumor growth, radiation response and patient-specific evolution of resistance to additional treatments (pembrolizumab and bevacizumab). Model parameters were fitted from tumor growth curves of 16 patients enrolled in the phase 1 NCT02313272 trial that combined HFSRT with bevacizumab and pembrolizumab. Then, iRT +/− boost treatments were simulated and compared to HFSRT based on time to tumor regrowth. The modeling results demonstrated that iRT + boost(− boost) treatment was equal or superior to HFSRT in 15(11) out of 16 cases and that patients that remained responsive to pembrolizumab and bevacizumab would benefit most from iRT. Time to progression could be prolonged through the application of additional, intermittently delivered fractions. iRT hence provides a promising treatment option for recurrent high grade glioma patients for prospective clinical evaluation. [ABSTRACT FROM AUTHOR]

For multimodality therapies such as the combination of hyperthermia and radiation, quantification of biological effects is key for dose prescription and response prediction. Tumour spheroids have a microenvironment that more closely resembles that of tumours in vivo and may thus be a superior in vitro cancer model than monolayer cultures. Here, the response of tumour spheroids formed from two established human cancer cell lines (HCT116 and CAL27) to single and combination treatments of radiation (0–20 Gy), and hyperthermia at 47 °C (0–780 CEM43) has been evaluated. Response was analysed in terms of spheroid growth, cell viability and the distribution of live/dead cells. Time-lapse imaging was used to evaluate mechanisms of cell death and cell detachment. It was found that sensitivity to heat in spheroids was significantly less than that seen in monolayer cultures. Spheroids showed different patterns of shrinkage and regrowth when exposed to heat or radiation: heated spheroids shed dead cells within four days of heating and displayed faster growth post-exposure than samples that received radiation or no treatment. Irradiated spheroids maintained a dense structure and exhibited a longer growth delay than spheroids receiving hyperthermia or combination treatment at (thermal) doses that yielded equivalent levels of clonogenic cell survival. We suggest that, unlike radiation, which kills dividing cells, hyperthermia-induced cell death affects cells independent of their proliferation status. This induces microenvironmental changes that promote spheroid growth. In conclusion, 3D tumour spheroid growth studies reveal differences in response to heat and/or radiation that were not apparent in 2D clonogenic assays but that may significantly influence treatment efficacy. [ABSTRACT FROM AUTHOR]

Thermo-radiosensitisation is a promising approach for treatment of radio-resistant tumours such as those containing hypoxic subregions. Response prediction and treatment planning should account for tumour response heterogeneity, e.g. due to microenvironmental factors, and quantification of the biological effects induced. 3D tumour spheroids provide a physiological in vitro model of tumour response and a systems oncology framework for simulating spheroid response to radiation and hyperthermia is presented. Using a cellular automaton model, 3D oxygen diffusion, delivery of radiation and/or hyperthermia were simulated for many (1 0 4 − 1 0 7 ) individual cells forming a spheroid. The iterative oxygen diffusion model was compared to an analytical oxygenation model and simulations were calibrated and validated against experimental data for irradiated (0–10 Gy) and/or heated (0–240 CEM43) HCT116 spheroids. Despite comparable clonogenic survival, spheroid growth differed significantly following radiation or hyperthermia. This dynamic response was described well by the simulation ( R 2 > 0.85). Heat-induced cell death was implemented as a fast, proliferation-independent process, allowing reoxygenation and repopulation, whereas radiation was modelled as proliferation-dependent mitotic catastrophe. This framework stands out both through its experimental validation and its novel ability to predict spheroid response to multimodality treatment. It provides a good description of response where biological dose-weighting based on clonogenic survival alone was insufficient. [ABSTRACT FROM AUTHOR]

Combined radiotherapy (RT) and hyperthermia (HT) treatments may improve treatment outcome by heat induced radio-sensitisation. We propose an empirical cell survival model (AlphaR model) to describe this multimodality therapy. The model is motivated by the observation that heat induced radio-sensitisation may be explained by a reduction in the DNA damage repair capacity of heated cells. We assume that this repair is only possible up to a threshold level above which survival will decrease exponentially with dose. Experimental cell survival data from two cell lines (HCT116, Cal27) were considered along with that taken from the literature (baby hamster kidney [BHK] and Chinese hamster ovary cells [CHO]) for HT and combined RT-HT. The AlphaR model was used to study the dependence of clonogenic survival on treatment temperature, and thermal dose R2 ≥ 0.95 for all fits). For HT survival curves (0-80 CEM43 at 43.5-57 °C), the number of free fit AlphaR model parameters could be reduced to two. Both parameters increased exponentially with temperature. We derived the relative biological effectiveness (RBE) or HT treatments at different temperatures, to provide an alternative description of thermal dose, based on our AlphaR model. For combined RT-HT, our analysis is restricted to the linear quadratic arm of the model. We show that, for the range used (20-80 CEM43, 0-12 Gy), thermal dose is a valid indicator of heat induced radio-sensitisation, and that the model parameters can be described as a function thereof. Overall, the proposed model provides a flexible framework for describing cell survival curves, and may contribute to better quantification of heat induced radio-sensitisation, and thermal dose in general. [ABSTRACT FROM AUTHOR]

Purpose: Treatment planning for carbon ion therapy requires an accurate modeling of the biological response of each tissue to estimate the clinical outcome of a treatment. The relative biological effectiveness (RBE) accounts for this biological response on a cellular level but does not refer to the actual impact on the organ as a whole. For photon therapy, the concept of equivalent uniform dose (EUD) represents a simple model to take the organ response into account, yet so far no formulation of EUD has been reported that is suitable to carbon ion therapy. The authors introduce the concept of an equivalent uniform effect (EUE) that is directly applicable to both ion and photon therapies and exemplarily implemented it as a basis for biological treatment plan optimization for carbon ion therapy. Methods: In addition to a classical EUD concept, which calculates a generalized mean over the RBE-weighted dose distribution, the authors propose the EUE to simplify the optimization process of carbon ion therapy plans. The EUE is defined as the biologically equivalent uniform effect that yields the same probability of injury as the inhomogeneous effect distribution in an organ. Its mathematical formulation is based on the generalized mean effect using an effect-volume parameter to account for different organ architectures and is thus independent of a reference radiation. For both EUD concepts, quadratic and logistic objective functions are implemented into a research treatment planning system. A flexible implementation allows choosing for each structure between biological effect constraints per voxel and EUD constraints per structure. Exemplary treatment plans are calculated for a head-and-neck patient for multiple combinations of objective functions and optimization parameters. Results: Treatment plans optimized using an EUE-based objective function were comparable to those optimized with an RBE-weighted EUD-based approach. In agreement with previous results from photon therapy, the optimization by biological objective functions resulted in slightly superior treatment plans in terms of final EUD for the organs at risk (OARs) compared to voxel-based optimization approaches. This observation was made independent of the underlying objective function metric. An absolute gain in OAR sparing was observed for quadratic objective functions, whereas intersecting DVHs were found for logistic approaches. Even for considerable under- or overestimations of the used effect- or dose-volume parameters during the optimization, treatment plans were obtained that were of similar quality as the results of a voxel-based optimization. Conclusions: EUD-based optimization with either of the presented concepts can successfully be applied to treatment plan optimization. This makes EUE-based optimization for carbon ion therapy a useful tool to optimize more specifically in the sense of biological outcome while voxel-to-voxel variations of the biological effectiveness are still properly accounted for. This may be advantageous in terms of computational cost during treatment plan optimization but also enables a straight forward comparison of different fractionation schemes or treatment modalities. [ABSTRACT FROM AUTHOR]

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number R21CA263911 (HE, SS, HY). Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number R21CA234787 (HE, SS, HY). [Extracted from the article]

Radiotherapy is involved in 50% of all cancer treatments and 40% of cancer cures. Most of these treatments are delivered in fractions of equal doses of radiation (Fractional Equivalent Dosing (FED)) in days to weeks. This treatment paradigm has remained unchanged in the past century and does not account for the development of radioresistance during treatment. Even if under-optimized, deviating from a century of successful therapy delivered in FED can be difficult. One way of exploring the infinite space of fraction size and scheduling to identify optimal fractionation schedules is through mathematical oncology simulations that allow for in silico evaluation. This review article explores the evidence that current fractionation promotes the development of radioresistance, summarizes mathematical solutions to account for radioresistance, both in the curative and non-curative setting, and reviews current clinical data investigating non-FED fractionated radiotherapy. [ABSTRACT FROM AUTHOR]

Simple Summary: This study compares the capabilities of normal and tumour cells to divide after receiving spatially fractionated radiotherapy. For this treatment, a minority of cells (here ∼20%) receive very large, peak doses, whereas the remaining cells are spared and receive only a small valley dose. We show that tumour and normal cells respond differently to this treatment, and that spatially fractionated radiotherapy may have a greater effect on tumour than normal cells. Our work was conducted using laboratory equipment, rather than specialized synchrotron facilities implying that the observed response is present at conventional dose rates and hence purely an effect of the spatial fractionation of the treatment. Microbeam radiotherapy (MRT) is a preclinical method of delivering spatially-fractionated radiotherapy aiming to improve the therapeutic window between normal tissue complication and tumour control. Previously, MRT was limited to ultra-high dose rate synchrotron facilities. The aim of this study was to investigate in vitro effects of MRT on tumour and normal cells at conventional dose rates produced by a bench-top X-ray source. Two normal and two tumour cell lines were exposed to homogeneous broad beam (BB) radiation, MRT, or were separately irradiated with peak or valley doses before being mixed. Clonogenic survival was assessed and compared to BB-estimated surviving fractions calculated by the linear-quadratic (LQ)-model. All cell lines showed similar BB sensitivity. BB LQ-model predictions exceeded the survival of cell lines following MRT or mixed beam irradiation. This effect was stronger in tumour compared to normal cell lines. Dose mixing experiments could reproduce MRT survival. We observed a differential response of tumour and normal cells to spatially fractionated irradiations in vitro, indicating increased tumour cell sensitivity. Importantly, this was observed at dose rates precluding the presence of FLASH effects. The LQ-model did not predict cell survival when the cell population received split irradiation doses, indicating that factors other than local dose influenced survival after irradiation. [ABSTRACT FROM AUTHOR]