Prognostic significance of blood-based multi-cancer detection in circulating tumor DNA (ctDNA): 5-year outcomes analysis.

Abstract

101 Background: In the case-control Circulating Cell-free Genome Atlas (CCGA) study (NCT02889978), a multi-cancer early detection (MCED) test was developed that uses next-generation sequencing to detect a cancer signal shared across > 50 cancer types using ctDNA in blood. The concentration of ctDNA in blood is associated with cancer aggressiveness and prognosis. Previous analysis of participant outcomes in the second (cross-validation) CCGA substudy evaluated the prognostic value of cancer signal detection by an early version of the MCED test with 3-year follow-up. Participants with confirmed cancer and no cancer signal detected (NCSD) MCED test result had better 3-year survival than those with a cancer signal detected (CSD). In the present analysis, we evaluated the prognostic value of cancer signal detection by a refined version of the MCED test in the third (validation) CCGA substudy (CCGA3) using an updated statistical methodology with 5-year follow-up, a typical timeframe for cancer-survivor status. Methods: Participant blood samples collected during CCGA3 were analyzed using the MCED test. Participants with confirmed cancer were followed for up to 5 years and their overall survival stratified by cancer signal detection (CSD/NCSD). Observed survival was compared to the expected survival of a reference population calculated using Surveillance, Epidemiology, and End Results (SEER) data matched to the distribution of age, sex, cancer type, and stage in each signal detection group. A one-sample proportional hazard model was used to assess differences between observed and expected survival based on cancer signal detection status. Results: Follow-up data were available for 2475/2513 (99%) of participants with stageable, invasive cancer. Of these, 792 (32%) died during follow-up, 673/792 (85%) of whom had a CSD; of the 1683 (67%) participants alive at follow-up, 579/1683 (35%) had a CSD. Overall observed survival rates of both groups were higher than the expected survival rates based on SEER data matched for known clinical factors (43% vs 40% [CSD]; 88% vs 81% [NCSD]). Observed vs expected survival rates for participants with: stage I cancer were 66% vs 71% (CSD) and 90% vs 85% (NCSD); stage II cancer were 64% vs 67% (CSD) and 92% vs 83% (NCSD); stage III cancer were 48% vs 42% (CSD) and 79% vs 66% (NCSD); stage IV cancer were 22% vs 16% (CSD) and 56% vs 32% (NCSD). Overall HR for NCSD vs CSD across all stages was 0.60 (95% CI: 0.50-0.72; P = 6.18e-09). HRs for signal detection group vs SEER reference populations were < 1 at all stages with NCSD; with CSD, HRs were < 1 at stages III and IV and ≥1 at stages I and II. Conclusions: In CCGA3, 5-year follow-up confirmed that while CSD was associated with hazard of death, early-stage cases had long-term survival close to expected rates. These results suggest that a CSD MCED test result may inform prognosis and urgency of treatment. Clinical trial information: NCT02889978 .