Clinical relevance of tumor fraction assessment from circulating tumor DNA in metastatic colorectal cancer.

237 Background: Both tissue (TBx) and liquid (LBx) biopsies are indicated for detecting actionable alterations in metastatic colorectal cancer (mCRC). LBx is easier to obtain and often has faster turnaround time, but its informative results depend on the ctDNA tumor fraction (TF) content. We aimed to (1) evaluate the concordance between LBx and TBx in CRC; and (2) determine baseline lab and clinical factors associated with TF. Methods: CRC patients (pts) with both tissue and liquid Foundation Medicine comprehensive genomic profiling within an interval of up to 90 days between samples collection were analyzed. Positive percent agreement (PPA) and negative predictive value (NPV) for RAS (KRAS/NRAS) and BRAFV600E detection were calculated with tissue as reference. Clinical data was obtained by the US-wide de-identified Flatiron Health-Foundation Medicine real-world clinicogenomic CRC database, from ~280 cancer clinics (~800 sites of care) between 1/2011-12/2023. mCRC pts with LBx collected up to 60 days prior to therapy (Tx) start were analyzed for association between TF and baseline factors using logistic regression with TF≥1% as the binary outcome. Baseline factors include age, ECOG, race/ethnicity, line of therapy, practice type, metachronous vs. synchronous disease, tumor side, albumin, alkaline phosphatase (AP), serum creatinine, hemoglobin, lactate dehydrogenase (LDH), neutrophil-to-lymphocyte ratio, opioid and steroid pre-Tx. Foundation Medicine’s ctDNA TF is a composite algorithm prioritizing aneuploidy at higher levels to avoid germline signal and prioritizing variant allele frequency of canonical alterations at lower levels to maximize dynamic range. Results: A total of 402 pts had TBx and LBx (268 [67%] TF≥1% and 134 [33%] TF<1%). Without accounting for TF, the overall PPA for RAS and BRAFV600E was 77% and 83%, and the NPV was 65% and 99%. In LBx with TF≥1%, the PPA for RAS and BRAFV600E was 99% and 100%, and NPV was 98% and 100%. In contrast, for TF<1% samples, the PPA for RAS and BRAFV600E was 37% and 50%, and NPV was 57% and 97%. Of the 633 pts with LBx collected prior to Tx, 474 (75%) had TF≥1%. TF≥1% was independently associated with ECOG 1+ (odds ratio [OR]: 1.58, p=0.038), community oncology care site (OR: 2.58, p=0.002), elevated levels of AP (OR: 3.00, p<0.001) and LDH (OR: 3.97,p=0.002). TF≥1% was less likely to occur in right-side tumors (OR: 0.53, p=0.012) and metachronous disease (OR: 0.48, p<0.001). Conclusions: Approximately 70% of LBx samples from CRC pts have TF≥1%, with near-perfect concordance for RAS/BRAFV600E. For LBx with TF<1%, interpretation of negative results can be challenging due to low tumor shedding, especially for RAS mutations, where 43% may be false negatives and a subsequent TBx will provide more confidence in the negative results. Clinical factors such as right-sided tumors and metachronous disease may help anticipate low TF and guide decisions to pursue TBx.