Association of high ctDNA tumor fraction (TF) and detection of clinically significant copy number (CN) losses in liquid biopsies (LBx) from patients with advanced prostate cancer (PCa).

Abstract

53 Background: Detection of somatic genomic alterations (GA) in LBx is dependent upon the amount of tumor DNA shed, inferred by ctDNA TF. Many clinically actionable GA in PCa are homozygous copy number (CN) losses, which present distinct detection challenges relative to short variant (SV) alterations. BRCA2 CN loss in PCa has been associated with prolonged benefit from PARP inhibitors (PARPi), possibly due to the inability to develop resistance-associated BRCA2 reversion mutations. Methods: Tissue-based FoundationOne CDx (TBx) or blood-based FoundationOneLiquid CDx (LBx) comprehensive genomic profiling (CGP) data from PCa patients was interrogated. We compared GA detection rates in TBx, all LBx (any ctDNA TF), and high-TF (ctDNA TF≥20%) LBx for pathogenic/likely pathogenic SVs and CN losses in 5 clinically relevant genes: BRCA1 , BRCA2 , PTEN , RB1 , and TP53 . We used logistic regression on data from the nationwide de-identified Flatiron Health-Foundation Medicine Clinico-Genomic Database (~280 US cancer clinics, ~800 sites of care) to identify clinical factors predicting high ctDNA TF for optimal CN loss detection. Results: 24,888 TBx and 13,604 LBx were available from patients with predominantly advanced/recurrent PCa. The prevalence of CN losses and SVs in the 5 genes, as detected by TBx, are shown in the Table. >30% of all GA in BRCA2 (36%), PTEN (72%) and RB1 (48%) were CN losses, while losses were rare for BRCA1 (6%) and TP53 (7%). Across all LBx, CN loss rates were lower than in TBx ( BRCA1 0.01 vs 0.05%, BRCA2 1.0 vs 3.0%, PTEN 7.4 vs 24%, RB1 1.4 vs 2.7%, TP53 0.7 vs 2.6%), while SV rates were similar to TBx (not shown). However, high-TF LBx and TBx had similar or higher CN loss rates ( BRCA1 0.06 vs 0.05%, BRCA2 3.3 vs 3.0%, PTEN 28 vs 24%, RB1 5.4 vs 2.7%, TP53 2.9 vs 2.6%), suggesting that high ctDNA TF enables optimal CN detection. Logistic regression linked high alkaline phosphatase, high PSA, and blood draw within 60 days prior to a new therapy with greater likelihood of ctDNA TF≥20% (all P <0.01). Conclusions: Detection of SVs requires only modest ctDNA TF, while CN losses require higher ctDNA TF. High alkaline phosphatase and PSA levels are associated with greater likelihood of ctDNA TF sufficient for CN loss detection. About one-third of men with PCa harboring BRCA2 GA have CN loss, which confers prolonged benefit from PARPi. A validated ctDNA TF is a critical quality indicator, and when low, should prompt consideration of confirmatory TBx to ensure that actionable CN losses are not missed. Gene Rate of homozygous CN loss detection (TBx, %, 95CI) Rate of SV detection (TBx, %) Proportion of all GA due to CN loss (TBx, %) BRCA1 0.05 [0.03-0.09] 0.76 [0.65-0.87] 6.5 [3.5-11] BRCA2 3.0 [2.8-3.2] 5.2 [5.0-5.5] 36 [34-39] PTEN 24 [23-25] 9.0 [8.6-9.4] 72 [71-73] RB1 2.7 [2.5-2.9] 2.9 [2.7-3.1] 48 [45-51] TP53 2.6 [2.4-2.8] 36 [35-37] 6.7 [6.2-7.2]