Genotyping of circulating tumor DNA in biliary tract cancer to reveal diagnostic and prognostic information.

291 Background: Biliary tract cancer (BTC) shows increasing incidence and is associated with a high mortality. Diagnosis is difficult due to the frequently occurring inaccessibility of the tumor for biopsy. Noninvasive approaches for (i) assessing and (ii) monitoring the tumor-specific molecular setup are desirable. Characterization of circulating tumor DNA (ctDNA) may help to achieve this goal. Methods: Blood and tumor tissue samples from patients with locally advanced or metastatic BTC prior to and during palliative treatment were collected. Tumor tissue and corresponding ctDNA samples underwent targeted genotyping of the 15 most frequently mutated genes in BTC. Findings were correlated with clinical and imaging data. Results: 24 therapy naive patients with histologically confirmed BTC were included for analyses. The overall mutational concordance (blood/tissue) was 74% and 92% for intrahepatic tumors. The mean variant allele frequency (VAF) in tumor tissue of therapy naïve patients was significantly higher compared to the respective ctDNA (p = 0.0291). In turn, the sequencing depth for ctDNA was significantly deeper (p = 0.0001), enabling us to detect also rare variants. Mean ctDNA VAF at baseline significantly correlated with tumor load (Spearman, r = 0.4073, p = 0.0433) and, exclusively for intrahepatic tumors, also with progression-free survival (Spearman, r = -0.5878, p = 0.0386). During 1st line palliative treatment, we detected a change in the mutational landscape in 36% of cases, Moreover, we had access to ctDNA samples of 5 pretreated patients. While ctDNA samples of therapy naïve patients (n = 23) showed a mean of 0.78 mutations per patient, ctDNA samples of pretreated patients (n = 5) exhibited a mean of 0.4 mutations (p = 0.5519). Conclusions: The molecular landscape of BTC is depicted in ctDNA which may enable to adapt diagnostic and therapeutic strategies to the specific molecular setup present at a certain time. In contrast, the use of targeted resequencing is likely to be insufficient for a comprehensive assessment of treatment induced BTC evolution. For this purpose, we suggest a more extensive analysis of ctDNA by broader sequencing applications and the incorporation of epigenetics.