213P - Constructing a high-definition patient-digital twin (PDT) in treatment-naïve women with advanced cancer

Background

Lifestyle, environmental or mood factors may influence cancer outcomes. To harness these complex data for clinical decision-making, we propose building a PDT using a multidimensional and longitudinal data capture approach in prevalent cancers. We report the feasibility at the 10% planned accrual (N=30/300).

Methods

Prospective, multicentric (n=9) observational study. Inclusion criteria: treatment-naïve women >18 y.o. with metastatic lung or colorectal cancer eligible for any first-line treatment or hormone-positive breast cancer eligible for first-line endocrine + CDK4/6 inhibitor. Patients are profiled at baseline (V1), 1 month (m; V2), 3m (V3), and every 3m (V4…Vn), including an e-CRF with 24 clinical, 18 demographic, and 96 bloodwork fields; body CT; plasma (Pl) metabolomics and proteomics, germline methylome, stool (St) metabolomics and metagenomics; and V1 tumor (T) and germline genomics. Continuous physiological monitoring is captured via smartwatch (HR, pO2, sleep, activity). EORTCQ30, GH28, PRO-CTCAE and diet PDAQ questionnaires V1 to Vn, spontaneous reporting of emotional states (from a list of 20), and pain (VAS, daily) are captured via App. Coverage is computed as % of obtained vs. planned data. Sample quality is assessed based on pre-established QC parameters.

Results

N=30 pts (14 breast, 12 lung, 4 colon). Median(range) time on study: 202 days (2-350 d): 4 pts for > 9 m, 25 for >6 m and 29 for >4m. Clinical and analytical fields completed at 86%, and 93% (V1); 89% and 96% (V2); 81% and 93% (V3); 74% and 78% (V4); 100% and 100% (V5). Samples: V1: 93% T, 100% Pl and 80% St; V2: 100% Pl, 82% St; V3: 92% Pl, 78% St; V4: 87% Pl, 61% St; V5: 100% Pl, 100% St samples were obtained. 100% T, Pl and St samples met DNA QC criteria, yielding 11.9 (0.29-70), 10.2 (0.4-59) and 9.0 (1.5-27) ug of DNA, respectively. 100% samples for methylomics and metabolomics met 5 and 4 QC criteria respectively. Physiologic monitoring captured 94%, 83%, 69% and 63% of the maximum expected activity, HR, pO2 and sleep quality data. 90% and 93% of ptes entered >1 emotion (median 0.56/day) and pain data (median 0.53/day). 250 of 328 (76%) planned questionnaires were filled.

Conclusions

High patient engagement and data/sample quality affirm the feasibility of constructing a PDT.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Centro Nacional de Investigaciones Oncologicas (CNIO), Madrid, Spain.

Funding

Centro Nacional de Investigaciones Oncologicas (CNIO), Madrid, Spain.

Disclosure

All authors have declared no conflicts of interest.