A systematic review of biologically-informed deep learning models for cancer: fundamental trends for encoding and interpreting oncology data.

BACKGROUND: There is an increasing interest in the use of Deep Learning (DL) based methods as a supporting analytical framework in oncology. However, most direct applications of DL will deliver models with limited transparency and explainability, which constrain their deployment in biomedical settings. METHODS: This systematic review discusses DL models used to support inference in cancer biology with a particular emphasis on multi-omics analysis. It focuses on how existing models address the need for better dialogue with prior knowledge, biological plausibility and interpretability, fundamental properties in the biomedical domain. For this, we retrieved and analyzed 42 studies focusing on emerging architectural and methodological advances, the encoding of biological domain knowledge and the integration of explainability methods. RESULTS: We discuss the recent evolutionary arch of DL models in the direction of integrating prior biological relational and network knowledge to support better generalisation (e.g. pathways or Protein-Protein-Interaction networks) and interpretability. This represents a fundamental functional shift towards models which can integrate mechanistic and statistical inference aspects. We introduce a concept of bio-centric interpretability and according to its taxonomy, we discuss representational methodologies for the integration of domain prior knowledge in such models. CONCLUSIONS: The paper provides a critical outlook into contemporary methods for explainability and interpretability used in DL for cancer. The analysis points in the direction of a convergence between encoding prior knowledge and improved interpretability. We introduce bio-centric interpretability which is an important step towards formalisation of biological interpretability of DL models and developing methods that are less problem- or application-specific.

Meta-analysis informed machine learning: Supporting cytokine storm detection during CAR-T cell Therapy.

Cytokine release syndrome (CRS), also known as cytokine storm, is one of the most consequential adverse effects of chimeric antigen receptor therapies that have shown otherwise promising results in cancer treatment. When emerging, CRS could be identified by the analysis of specific cytokine and chemokine profiles that tend to exhibit similarities across patients. In this paper, we exploit these similarities using machine learning algorithms and set out to pioneer a meta-review informed method for the identification of CRS based on specific cytokine peak concentrations and evidence from previous clinical studies. To this end we also address a widespread challenge of the applicability of machine learning in general: reduced training data availability. We do so by augmenting available (but often insufficient) patient cytokine concentrations with statistical knowledge extracted from domain literature. We argue that such methods could support clinicians in analyzing suspect cytokine profiles by matching them against the said CRS knowledge from past clinical studies, with the ultimate aim of swift CRS diagnosis. We evaluate our proposed methods under several design choices, achieving performance of more than 90% in terms of CRS identification accuracy, and showing that many of our choices outperform a purely data-driven alternative. During evaluation with real-world CRS clinical data, we emphasize the potential of our proposed method of producing interpretable results, in addition to being effective in identifying the onset of cytokine storm.

Patient attrition in Molecular Tumour Boards: a systematic review.

BACKGROUND: Molecular Tumour Boards (MTBs) were created with the purpose of supporting clinical decision-making within precision medicine. Though in use globally, reporting on these meetings often focuses on the small percentages of patients that receive treatment via this process and are less likely to report on, and assess, patients who do not receive treatment. METHODS: A literature review was performed to understand patient attrition within MTBs and barriers to patients receiving treatment. A total of 51 papers were reviewed spanning a 6-year period from 11 different countries. RESULTS: In total, 20% of patients received treatment through the MTB process. Of those that did not receive treatment, the main reasons were no mutations identified (27%), no actionable mutations (22%) and clinical deterioration (15%). However, data were often incomplete due to inconsistent reporting of MTBs with only 55% reporting on patients having no mutations, 55% reporting on the presence of actionable mutations with no treatment options and 59% reporting on clinical deterioration. DISCUSSION: As patient attrition in MTBs is an issue which is very rarely alluded to in reporting, more transparent reporting is needed to understand barriers to treatment and integration of new technologies is required to process increasing omic and treatment data.

Safety, tolerability and pharmacokinetics of the fibroblast growth factor receptor inhibitor AZD4547 in Japanese patients with advanced solid tumours: a Phase I study.

Background AZD4547 is a potent, oral, highly selective fibroblast growth factor receptor (FGFR) inhibitor in clinical development for treating tumours with a range of FGFR aberrations, including FGFR mutations, amplifications and fusions. Methods This open-label, Phase I, multicentre study (NCT01213160) evaluated the safety, pharmacokinetics, and preliminary antitumour efficacy (RECIST v1.1) of AZD4547 monotherapy in Japanese patients with advanced solid tumours. Part A was a dose-escalation part; Part B was a dose-expansion part in patients with FGFR-amplified tumours, confirmed by fluorescence in situ hybridization. Results Thirty patients enrolled in Part A (dose range: 40 mg twice daily [bid] to 120 mg bid; 160 mg once daily [qd]), four in Part B (80 mg bid). No dose-limiting toxicities were observed and maximum tolerated dose was not determined. Most common adverse events (AEs; any grade) were: dysgeusia (50% of patients); stomatitis (41%); diarrhoea (38%); hyperphosphataemia (38%); dry mouth (35%). Common grade ≥3 AEs were nausea (12% of patients) and neutropenia (9%). No complete or partial responses were observed: 21/30 patients had stable disease ≥4 weeks in Part A, and 1/4 patients had stable disease ≥10 weeks in Part B. Following single and multiple dosing, absorption rate appeared moderate; peak plasma concentrations generally occurred 3-4 h post-dose, then declined biphasically with terminal half-life ~30 h. Steady state was reached by day 8. Compared with single dosing, plasma concentrations were, on average, 2.4- and 3.3- to 5.4-fold higher after qd and bid dosing, respectively. Conclusions AZD4547 was well tolerated in Japanese patients, with best response of stable disease ≥4 weeks.

Digital ECMT Cancer Trial Matching Tool: an Open Source Research Application to Support Oncologists in the Identification of Precision Medicine Clinical Trials.

PURPOSE: Matching patients with cancer to precision medicine clinical trials on the basis of their tumor genotype has the potential to improve outcomes for patients who have exhausted standard-of-care treatment options. However, the matching process presents a substantial challenge because of the number of clinical trials available. We describe a free, open source research tool designed to extract relevant trial information to support oncologists in the matching process, and we illustrate its utility with recent case studies of patients who were matched to trials using this tool. METHODS: Trial records are sourced from ClinicalTrials.gov and indexed using natural language processing techniques, including named entity recognition, term normalization, and relationship extraction. Relationships between trials and genetic alterations are assigned scores on the basis of a rule-based system. All data are updated daily. A user interface is provided via R Shiny app. RESULTS: An instance of the trial match tool, configured for UK clinical trials, is hosted by the digital Experimental Cancer Medicine Team (see link in Data Sharing Statement). Users select the relevant cancer type and genetic alteration(s). Matching studies are ranked according to the score assigned for the selected genetic alterations. Results may be downloaded and attached to the patient's health record if desired. The tool is currently being used to support the ongoing TARGET National study, which aims to match up to 6,000 patients to early phase clinical trials. We present three case studies that exemplify relationships between genetic alterations and studies. CONCLUSION: With increasing numbers of precision medicine treatments and as comprehensive molecular profiling of tumor samples becomes more common, decision support tools are likely to become increasingly important. This work represents an important step toward the development and wider implementation of such systems.

Biomarker identification using dynamic time warping analysis: a longitudinal cohort study of patients with COVID-19 in a UK tertiary hospital.

OBJECTIVES: COVID-19 is a heterogeneous disease, and many reports have described variations in demographic, biochemical and clinical features at presentation influencing overall hospital mortality. However, there is little information regarding longitudinal changes in laboratory prognostic variables in relation to disease progression in hospitalised patients with COVID-19. DESIGN AND SETTING: This retrospective observational report describes disease progression from symptom onset, to admission to hospital, clinical response and discharge/death among patients with COVID-19 at a tertiary centre in South East England. PARTICIPANTS: Six hundred and fifty-one patients treated for SARS-CoV-2 between March and September 2020 were included in this analysis. Ethical approval was obtained from the HRA Specific Review Board (REC 20/HRA/2986) for waiver of informed consent. RESULTS: The majority of patients presented within 1 week of symptom onset. The lowest risk patients had low mortality (1/45, 2%), and most were discharged within 1 week after admission (30/45, 67%). The highest risk patients, as determined by the 4C mortality score predictor, had high mortality (27/29, 93%), with most dying within 1 week after admission (22/29, 76%). Consistent with previous reports, most patients presented with high levels of C reactive protein (CRP) (67% of patients >50 mg/L), D-dimer (98%>upper limit of normal (ULN)), ferritin (65%>ULN), lactate dehydrogenase (90%>ULN) and low lymphocyte counts (81%

Wave comparisons of clinical characteristics and outcomes of COVID-19 admissions - Exploring the impact of treatment and strain dynamics.

OBJECTIVES: Dexamethasone has now been incorporated into the standard of care for COVID-19 hospital patients. However, larger intensive care unit studies have failed to show discernible improvements in mortality in the recent wave. We aimed to investigate the impacts of these factors on disease outcomes in a UK hospital study. METHODS: This retrospective observational study reports patient characteristics, interventions and outcomes in COVID-19 patients from a UK teaching hospital; cohort 1, pre 16th June-2020 (pre-dexamethasone); cohort 2, 17th June to 30th November-2020 (post-dexamethasone, pre-VOC 202,012/01 as dominant strain); cohort 3, 1st December-2020 to 3rd March-2021 (during establishment of VOC202012/01 as the dominant strain). RESULTS: Dexamethasone treatment was more common in cohorts 2 and 3 (42.7% and 51.6%) compared with cohort 1 (2.5%). After adjusting for risk, odds of death within 28 days were 2-fold lower in cohort 2 vs 1 (OR:0.47,[0.27,0.79],p = 0.006). Mortality was higher cohort 3 vs 2 (20% vs 14%); but not significantly different to cohort 1 (OR: 0.86,[0.64, 1.15],p = 0.308). CONCLUSIONS: The real world finding of lower mortality following dexamethasone supports the published trial evidence and highlights ongoing need for research with introduction of new treatments and ongoing concern over new COVID-19 variants.

Technology clinical trials: Turning innovation into patient benefit.

Health care needs to continuously evolve and innovate to maintain the health of populations. Technology has the potential to enable better patient engagement and ownership, as well as optimise therapeutic interventions and data-science approaches to facilitate improved health care decisions. Yet, to date, technological innovation has not resulted in the rate of change that could have been predicted from other sectors. This article discusses multiple reasons for this and proposes a newly tested and deployed solution: the technology clinical trial. The technology clinical trial methodology has been developed through working directly with patients, clinical and medical devicetrial experts. This approach enables researchers to use the complex environment of health care as an opportunity to transform the pace of innovation and create new care pathways. Instead of testing a single innovation, researchers can 'step back' and systematically review all areas of the patient's journey for potential optimization. Then integrate novel data science, technological advances, process updates, behavioural science, and patient engagement to co-create a streamlined multidisciplinary solution. As a result, this research has the potential for larger advances due to the emergent benefits that can arise when the individual elements work together as a whole. These potential benefits are then robustly tested, characterised and measured in the trial environment to ensure that future application of the innovative pathway is supported by the robust empirical data health care requires.

Research Evaluation Alongside Clinical Treatment in COVID-19 (REACT COVID-19): an observational and biobanking study.

INTRODUCTION: The COVID-19 pandemic caused by SARS-CoV-2 places immense worldwide demand on healthcare services. Earlier identification of patients at risk of severe disease may allow intervention with experimental targeted treatments, mitigating the course of their disease and reducing critical care service demand. METHODS AND ANALYSIS: This prospective observational study of patients tested or treated for SARS-CoV-2, who are under the care of the tertiary University Hospital Southampton NHS Foundation Trust (UHSFT), captured data from admission to discharge; data collection commenced on 7 March 2020. Core demographic and clinical information, as well as results of disease-defining characteristics, was captured and recorded electronically from hospital clinical record systems at the point of testing. Manual data were collected and recorded by the clinical research team for assessments which are not part of the structured electronic healthcare record, for example, symptom onset date. Thereafter, participant records were continuously updated during hospital stay and their follow-up period. Participants aged >16 years were given the opportunity to provide consent for excess clinical sample storage with optional further biological sampling. These anonymised samples were linked to the clinical data in the Real-time Analytics for Clinical Trials platform and were stored within a biorepository at UHSFT. ETHICS AND DISSEMINATION: Ethical approval was obtained from the HRA Specific Review Board (REC 20/HRA/2986) for waiver of informed consent for the database-only cohort; the procedures conform with the Declaration of Helsinki. The study design, protocol and patient-facing documentation for the biobanking arm of the study have been approved by North West Research Ethics Committee (REC 17/NW/0632) as an amendment to the National Institute for Health Research Southampton Clinical Research Facility-managed Southampton Research Biorepository. This study will be published as peer-reviewed articles and presented at conferences, presentations and workshops.

An adaptive, biomarker-directed platform study of durvalumab in combination with targeted therapies in advanced urothelial cancer.

Durvalumab is a programmed death-ligand 1 (PD-L1) inhibitor with clinical activity in advanced urothelial cancer (AUC)1. AUC is characterized by several recurrent targetable genomic alterations2-5. This study ( NCT02546661 , BISCAY) combined durvalumab with relevant targeted therapies in biomarker-selected chemotherapy-refractory AUC populations including: (1) fibroblast growth factor receptor (FGFR) inhibitors in tumors with FGFR DNA alterations (FGFRm); (2) pharmacological inhibitor of the enzyme poly-ADP ribose polymerase (PARP) in tumors with and without DNA homologous recombination repair deficiency (HRRm); and (3) TORC1/2 inhibitors in tumors with DNA alteration to the mTOR/PI3K pathway3-5.This trial adopted a new, biomarker-driven, multiarm adaptive design. Safety, efficacy and relevant biomarkers were evaluated. Overall, 391 patients were screened of whom 135 were allocated to one of six study arms. Response rates (RRs) ranged 9-36% across the study arms, which did not meet efficacy criteria for further development. Overall survival (OS) and progression-free survival (PFS) were similar in the combination arms and durvalumab monotherapy arm. Biomarker analysis showed a correlation between circulating plasma-based DNA (ctDNA) and tissue for FGFRm. Sequential circulating tumor DNA analysis showed that changes to FGFRm correlated with clinical outcome. Our data support the clinical activity of FGFR inhibition and durvalumab monotherapy but do not show increased activity for any of the combinations. These findings question the targeted/immune therapy approach in AUC.

Utility of ctDNA to support patient selection for early phase clinical trials: the TARGET study.

Next-generation sequencing (NGS) of circulating tumor DNA (ctDNA) supports blood-based genomic profiling but is not yet routinely implemented in the setting of a phase I trials clinic. TARGET is a molecular profiling program with the primary aim to match patients with a broad range of advanced cancers to early phase clinical trials on the basis of analysis of both somatic mutations and copy number alterations (CNA) across a 641 cancer-associated-gene panel in a single ctDNA assay. For the first 100 TARGET patients, ctDNA data showed good concordance with matched tumor and results were turned round within a clinically acceptable timeframe for Molecular Tumor Board (MTB) review. When a 2.5% variant allele frequency (VAF) threshold was applied, actionable mutations were identified in 41 of 100 patients, and 11 of these patients received a matched therapy. These data support the application of ctDNA in this early phase trial setting where broad genomic profiling of contemporaneous tumor material enhances patient stratification to novel therapies and provides a practical template for bringing routinely applied blood-based analyses to the clinic.

A Phase Ib Open-Label Multicenter Study of AZD4547 in Patients with Advanced Squamous Cell Lung Cancers.

Purpose: Squamous cell lung cancers (SQCLC) account for 25% of all NSCLCs, yet the prognosis of these patients is poor and treatment options are limited. Amplified FGFR1 is one of the most common oncogenic events in SQCLCs, occurring in approximately 20% of cases. AZD4547 is a potent and selective FGFR1-3 inhibitor with antitumor activity in FGFR1-amplified SQCLC cell lines and patient-derived xenografts.Experimental Design: On the basis of these data, we performed a phase I study of AZD4547 in patients with previously treated stage IV FGFR1-amplified SQCLCs (NCT00979134). FGFR1 amplification (FGFR1:CEP8 ≥ 2) was determined by FISH. The primary endpoint was safety/tolerability. Secondary endpoints included antitumor activity, pharmacokinetics, pharmacodynamics, and molecular analyses.Results: Fifteen FGFR1-amplified patients were treated. The most common related adverse events (AE) were gastrointestinal and dermatologic. Grade ≥3-related AEs occurred in 3 patients (23%). Thirteen patients were evaluable for radiographic response assessment. The overall response rate was 8% (1 PR). Two of 15 patients (13.3%) were progression-free at 12 weeks, and the median overall survival was 4.9 months. Molecular tests, including next-generation sequencing, gene expression analysis, and FGFR1 immunohistochemistry, showed poor correlation between gene amplification and expression, potential genomic modifiers of efficacy, and heterogeneity in 8p11 amplicon.Conclusions: AZD4547 was tolerable at a dosage of 80 mg oral twice a day, with modest antitumor activity. Detailed molecular studies show that these tumors are heterogeneous, with a range of mutational covariates and stark differences in gene expression of the 8p11 amplicon that likely explain the modest efficacy of FGFR inhibition in this disease. Clin Cancer Res; 23(18); 5366-73. ©2017 AACR.

Development and Evaluation of a New Technological Way of Engaging Patients and Enhancing Understanding of Drug Tolerability in Early Clinical Development: PROACT.

INTRODUCTION: During early clinical testing of a new medication, it is critical to understand and characterise patient tolerability. However, in early clinical studies, it is difficult for patients to contribute directly to the sponsors' understanding of a new compound. Patient reported opinions about clinical tolerability (PROACT) provides a new, simple and innovative way in which patients can collaborate using an application downloaded to a mobile computer or smartphone. METHODS: PROACT was designed with special consideration given to patient confidentiality, patient engagement and data security. A pilot study was conducted to investigate patient uptake of PROACT and to characterize clinical trial information it captured. Patients recruited to Phase I oncology trials at a UK center were eligible to participate but were required to have a tablet computer or smartphone. Patients used PROACT to upload audio/video messages that became available instantly to their clinical team, who were able to reply to the patient within PROACT. The patient's message was also analyzed, personally-identifiable information removed and anonymized information then made available to the sponsor in an analytics module for decision-making. In parallel, a patient focus group was engaged to provide feedback on communication needs during early clinical trials and the PROACT concept. RESULTS: Of the 16 patients informed of PROACT, 8 had a smart device and consented to take part. Use of PROACT varied and all messages volunteered were relevant and informative for drug development. Topics disclosed included tolerability impacts, study design, and drug formulation. Alignment with the clinical study data provided a richer understanding of tolerability and treatment consequences. This information was available to be shared among the clinical team and the sponsor, to improve patient support and experience. Patient forum feedback endorsed the concept and provided further information to enhance the application. CONCLUSION: Overall, PROACT achieved proof of concept in this small pilot study and delivered a secure end-to-end system that protected patient privacy and provided preliminary insight into patient experiences beyond the usual clinical trial data set. The use of mobile devices to interact actively with participants in clinical trials may be a new way of engaging and empowering patients. Further validation of this technology in larger patient cohorts is ongoing. FUNDING: AstraZeneca.

Complexity of FGFR signalling in metastatic urothelial cancer.

BACKGROUND: Urothelial cancers (UC) are the fourth most common tumours worldwide after prostate (or breast), lung and colorectal cancer. Despite recent improvements in their management, UC remain an aggressive disease associated with a poor outcome. Following disease progression on first-line platinum-based chemotherapy, very few effective treatment options are available and none of them have shown significant improvement in overall survival. Alterations of the fibroblast growth factor receptor (FGFR) pathway including amplification, mutations and overexpression are common in UC. Pre-clinical data suggest that the presence of such dysregulations may confer sensitivity to FGFR inhibitors. MATERIALS AND METHODS: We present here the case of a patient with a metastatic UC of the renal pelvis with lymph node metastases treated with the selective FGFR inhibitor AZD4547. RESULTS: To date, the patient has been on a study drug for 32 months with acceptable tolerance and maintained radiological partial response as per RECIST 1.1 criteria. Exploratory biomarker analysis showed FGFR3, FGFR1, FGF-ligand and fibroblast growth factor receptor substrate 2 (FRS2) expression in the patient's tumour, together with the presence of a germ-line mutation in the FGFR3 extracellular binding domain. This is not a known hotspot mutation, and the functional significance remains unclear. CONCLUSIONS: The FGFR inhibitor AZD4547 exhibits antitumour activity in a metastatic urothelial cancer displaying FGFR1, FGFR3, FGF-ligand and FRS2 expression. This lends support to the further exploration of FGFR inhibitors in urothelial cancer. Further studies are required to determinate the most effective way to select those patients most likely to respond.