Curation of myeloma observational study MALIMAR using XNAT: solving the challenges posed by real-world data.

OBJECTIVES: MAchine Learning In MyelomA Response (MALIMAR) is an observational clinical study combining "real-world" and clinical trial data, both retrospective and prospective. Images were acquired on three MRI scanners over a 10-year window at two institutions, leading to a need for extensive curation. METHODS: Curation involved image aggregation, pseudonymisation, allocation between project phases, data cleaning, upload to an XNAT repository visible from multiple sites, annotation, incorporation of machine learning research outputs and quality assurance using programmatic methods. RESULTS: A total of 796 whole-body MR imaging sessions from 462 subjects were curated. A major change in scan protocol part way through the retrospective window meant that approximately 30% of available imaging sessions had properties that differed significantly from the remainder of the data. Issues were found with a vendor-supplied clinical algorithm for "composing" whole-body images from multiple imaging stations. Historic weaknesses in a digital video disk (DVD) research archive (already addressed by the mid-2010s) were highlighted by incomplete datasets, some of which could not be completely recovered. The final dataset contained 736 imaging sessions for 432 subjects. Software was written to clean and harmonise data. Implications for the subsequent machine learning activity are considered. CONCLUSIONS: MALIMAR exemplifies the vital role that curation plays in machine learning studies that use real-world data. A research repository such as XNAT facilitates day-to-day management, ensures robustness and consistency and enhances the value of the final dataset. The types of process described here will be vital for future large-scale multi-institutional and multi-national imaging projects. CRITICAL RELEVANCE STATEMENT: This article showcases innovative data curation methods using a state-of-the-art image repository platform; such tools will be vital for managing the large multi-institutional datasets required to train and validate generalisable ML algorithms and future foundation models in medical imaging. KEY POINTS: • Heterogeneous data in the MALIMAR study required the development of novel curation strategies. • Correction of multiple problems affecting the real-world data was successful, but implications for machine learning are still being evaluated. • Modern image repositories have rich application programming interfaces enabling data enrichment and programmatic QA, making them much more than simple "image marts".

MicroRNAs as biomarkers for trastuzumab-based therapy in HER2-positive advanced oesophago-gastric cancer patients.

Background: This study aimed to identify microRNAs (miRs) as circulating biomarkers of resistance to first-line trastuzumab-based therapy in advanced HER2-positive oesophago-gastric cancer patients. Methods: A high-throughput 1015 Exiqon miRCURY LNA™ microRNA inhibitor library screen was performed in trastuzumab-treated HER2-positive NCI-N87 and HER2-negative FLO-1 oesophago-gastric cancer cell lines. NanoString nCounter® miR analysis was performed in NCI-N87, FLO-1, and MAGIC trial (ISRCTN93793971) formalin-fixed paraffin-embedded (FFPE) oesophago-gastric cancer patient samples. MiR-148a-3p copies in plasma samples were quantified using digital droplet polymerase chain reaction (ddPCR) from HER2-positive oesophago-gastric cancer patients treated with standard-of-care trastuzumab-based therapy within the FOrMAT (NCT02112357) and PLATFORM (NCT02678182) clinical trials. The primary endpoints were overall survival (OS) for plasma miR-148a-3p HIGH (>median) versus LOW (≤median). The secondary endpoints were progression-free survival (PFS) and 3-month progression-free rates (PFRs) miR-148a-3p HIGH versus LOW. PLATFORM sensitivity analysis normalised miR-148a-3p (NmiR-148a-3p). Results: The inhibition of miR-148a-3p reduced NCI-N87 relative cell viability (<0.6) and expression was high (>242) in NCI-N87 and HER2-positive MAGIC trial patients (n=5). Normalised-miR-148a-3p (NmiR-148a-3p) LOW versus HIGH demonstrated a statistically significant difference in 3-month PFRs (n=23; OR, 0.11 [0.02-0.78]; p=0.027; aOR, 0.03 [0.001-0.71], p=0.029) but no difference in OS or PFS. There was no statistically significant relationship between miR-148-3p LOW versus HIGH for OS (PLATFORM, n=62; hazard ratio [HR], 0.98 [0.57-1.66]; p=0.933; FOrMAT, n=8; HR, 0.54 [0.13-2.31]; p=0.322), PFS (n=62; HR, 1.08 [0.65-1.81]; p=0.759; FOrMAT, n=8; HR, 1.26 [0.31-5.07]; p=0.714), or PFRs (PLATFORM, n=31; odds ratio [OR], 0.67 [0.2-2.8]; p=0.577). Conclusion: Normalised miR-148a-3p may be a relevant biomarker for trastuzumab-based therapy in advanced HER2-positive oesophago-gastric cancer patients.

The peritoneal cancer index as a predictor of complete cytoreduction at primary and interval cytoreductive surgery in advanced ovarian cancer.

OBJECTIVE: The peritoneal cancer index quantitatively assesses cancer distribution and tumor burden in the peritoneal cavity. The aim of this study is to evaluate the association between the peritoneal cancer index and completeness of surgical cytoreduction for ovarian cancer and to identify a cut-off above which complete cytoreduction is unlikely. METHODS: This is a single-center prospective cohort observational study. A total of 100 consecutive patients who underwent ovarian cancer surgery were included. Peritoneal cancer index scores prior to and after surgery were calculated, and a cut-off value for incomplete cytoreduction was identified using a receiver operator characteristic (ROC) curve. Surgical complexity, blood loss, length of surgery, and complications were analyzed and associations with the peritoneal cancer index score were evaluated. RESULTS: The overall median peritoneal cancer index score was 9.5 (range 0-36). The median age of the patients was 61 years (range 24-85). The most common stage was III (13% stage II, 53% stage III, 34% stage IV) and the most common histologic sub-type was high-grade serous (76% high-grade serous, 8% low-grade serous, 5% clear cell, 4% serous borderline, 2% endometrioid, 2% adult granulosa cell, 2% adenocarcinoma, 1% carcinosarcoma). Complete cytoreduction was achieved in 82% of patients, with a median score of 9 (range 0-30). The remaining 18% had a median score of 28.5 (range 0-36). The best predictor of incomplete cytoreduction was the peritoneal cancer index score, with an area under the curve (AUC) of 0.928 (95% CI 0.85 to 1.00). ROC curve analysis determined a peritoneal cancer index cut-off score of 20. Major complications occurred in 15% of patients with peritoneal cancer index scores >20 and in 2.5% of patients with scores ≤20, which was statistically significant (p=0.014). CONCLUSIONS: In our study we found that a peritoneal cancer index score of ≤20 was associated with a high likelihood of complete cytoreduction. Incorporating the peritoneal cancer index into routine surgical practice and research may impact treatment plans.

Standard and Hypofractionated Dose Escalation to Intraprostatic Tumor Nodules in Localized Prostate Cancer: 5-Year Efficacy and Toxicity in the DELINEATE Trial.

PURPOSE: Our purpose was to report 5-year efficacy and toxicity of intraprostatic lesion boosting using standard and hypofractionated radiation therapy. METHODS AND MATERIALS: DELINEATE (ISRCTN 04483921) is a single center phase 2 multicohort study including standardly fractionated (cohort A: 74 Gy/37F to prostate and seminal vesicles [PSV]; cohort C 74 Gy/37F to PSV plus 60 Gy/37F to pelvic lymph nodes) and moderately hypofractionated (cohort B: 60 Gy/20F to PSV) prostate intensity-modulated radiation therapy patients with National Comprehensive Cancer Network intermediate/high-risk disease. Patients received an integrated boost of 82 Gy (cohorts A and C) or 67 Gy (cohort B) to multiparametric magnetic resonance imaging identified lesion(s). Primary endpoint was late Radiation Therapy Oncology Group (RTOG) gastrointestinal (GI) toxicity at 1 year. Secondary endpoints were acute and late toxicity (clinician and patient reported) and freedom from biochemical/clinical failure at 5 years. RESULTS: Two hundred and sixty-five men were recruited and 256 were treated (55 cohort A, 153 cohort B, and 48 cohort C). Median follow-up for each cohort was >5 years. Cumulative late RTOG grade 2+ GI toxicity at 1 year was 3.6% (95% confidence interval [CI], 0.9%-13.8%) (cohort A), 7.2% (95% CI, 4%-12.6%) (cohort B), and 8.4% (95% CI, 3.2%-20.8%) (cohort C). Cumulative late RTOG grade 2+ GI toxicity to 5 years was 12.8% (95% CI, 6.3%-25.1%) (cohort A), 14.6% (95% CI, 9.9%-21.4%) (cohort B), and 20.7% (95% CI, 11.2%-36.2%) (cohort C). Cumulative RTOG grade 2+ genitourinary toxicity to 5 years was 12.9% (95% CI, 6.4%-25.2%) (cohort A), 18.2% (95% CI, 12.8%-25.4%) (cohort B), and 18.2% (95% CI, 9.5%-33.2%) (cohort C). Five-year freedom from biochemical/clinical failure was 98.2% (95% CI, 87.8%-99.7%) (cohort A), 96.7% (95% CI, 91.3%- 98.8%) (cohort B), and 95.1% (95% CI, 81.6-98.7%) (cohort C). CONCLUSIONS: The DELINEATE trial has shown safety, tolerability, and feasibility of focal boosting in 20 or 37 fractions. Efficacy results indicate a low chance of prostate cancer recurrence 5 years after radiation therapy. Evidence from ongoing phase 3 randomized trials is awaited.

Development of machine learning support for reading whole body diffusion-weighted MRI (WB-MRI) in myeloma for the detection and quantification of the extent of disease before and after treatment (MALIMAR): protocol for a cross-sectional diagnostic test accuracy study.

INTRODUCTION: Whole-body MRI (WB-MRI) is recommended by the National Institute of Clinical Excellence as the first-line imaging tool for diagnosis of multiple myeloma. Reporting WB-MRI scans requires expertise to interpret and can be challenging for radiologists who need to meet rapid turn-around requirements. Automated computational tools based on machine learning (ML) could assist the radiologist in terms of sensitivity and reading speed and would facilitate improved accuracy, productivity and cost-effectiveness. The MALIMAR study aims to develop and validate a ML algorithm to increase the diagnostic accuracy and reading speed of radiological interpretation of WB-MRI compared with standard methods. METHODS AND ANALYSIS: This phase II/III imaging trial will perform retrospective analysis of previously obtained clinical radiology MRI scans and scans from healthy volunteers obtained prospectively to implement training and validation of an ML algorithm. The study will comprise three project phases using approximately 633 scans to (1) train the ML algorithm to identify active disease, (2) clinically validate the ML algorithm and (3) determine change in disease status following treatment via a quantification of burden of disease in patients with myeloma. Phase 1 will primarily train the ML algorithm to detect active myeloma against an expert assessment ('reference standard'). Phase 2 will use the ML output in the setting of radiology reader study to assess the difference in sensitivity when using ML-assisted reading or human-alone reading. Phase 3 will assess the agreement between experienced readers (with and without ML) and the reference standard in scoring both overall burden of disease before and after treatment, and response. ETHICS AND DISSEMINATION: MALIMAR has ethical approval from South Central-Oxford C Research Ethics Committee (REC Reference: 17/SC/0630). IRAS Project ID: 233501. CPMS Portfolio adoption (CPMS ID: 36766). Participants gave informed consent to participate in the study before taking part. MALIMAR is funded by National Institute for Healthcare Research Efficacy and Mechanism Evaluation funding (NIHR EME Project ID: 16/68/34). Findings will be made available through peer-reviewed publications and conference dissemination. TRIAL REGISTRATION NUMBER: NCT03574454.