First report of tumor-informed circulating tumor DNA in fibrolamellar carcinoma in adolescents and young adults: 51 patients.

Abstract

3034 Background: Fibrolamellar Carcinoma (FLC) is an extremely rare form of primary liver cancer (50-200 cases/year), unrelated to hepatocellular carcinoma. The only established standard of care is surgery. However, recurrence is at least 80%, even with R0 surgery, suggesting that FLC is best thought of as a systemic disease instead of a surgical disease. Systemic therapies in retrospective studies of more than 25 show immunotherapy and chemotherapy combinations have promise. Circulating “tumor-informed” tumor DNA (ctDNA) is a tool for tracking microscopic systemic disease, serving as a marker of systemic involvement, has shown success in management of other cancers. FLC has never been studied using this technology. Our retrospective study of 51 patients over a 2.5-year period is the first attempt to understand if ctDNA can be helpful in the management of FLC. Methods: All patients consented to blood draws on a 1-3 monthly basis, according to the company (Signatera) protocol. All patients had regular imaging reviewed by the FibroFighters Foundation National Tumor Board. Complete response (CR), Partial response (PR) (> 30% decrease), Progressive Disease (PD) (> 20% increase), and Stable Disease (SD) (neither PD/PR/CR) were recorded according RECIST 1.1. The imaging closest in time to the blood draw was used for correlation purposes. “Negative” is defined as a value = 0.00. “Positive” is defined as a value > 0.00. Results: Fifty-one FLC patients (26M, 25F, median age 23, range 10-56), 92% stage (III/IV), consented to ctDNA “blood biopsies,” (n=226 samples, mean 4.4/patient, range 1-11). Negative values correlated with no evidence of disease (NED) on imaging 78% of the time, positive values with disease 94% [PPV/NPV/sens/spec of 0.78/0.94/0.93/0.81]. NPV and sens increased to almost 100% with a ctDNA threshold of > 1.0 at the expense of a lower PPV and spec. However, consecutive negative ctDNA values correlated with radiologic NED 86% [PPV/NPV/sens/spec = 0.86/0.84/0.77/0.91] without a single PD in this cohort (n=64). A “downward trend or slope” on consecutive values predicted Objective Response (CR+PR) and Disease Control (CR+PR+SD) at 68% and 96% [PPV/NPV/sens/spec of 0.68/0.49/0.36/0.78 and 0.96/0.26/0.36/0.94, respectively], while an “upward trend” performed less well to predict relapse [PPV/NPV/sens/spec = 0.54/0.98/0.94/0.81]. Conclusions: In this first exploration of 51 young FLC patients, and 226 tissue-informed cell free ctDNA data points, “Positive” and “Negative” values overall performed well for predicting RECIST 1.1 “disease” or “NED” respectively, especially for consecutive values of zero, where no progressions were observed in 64 distinct FLC scenarios. A downward trend or “slope” performed better than an upward trend to predict objective response (PR+CR) versus relapse (PD).