Improved detection of circulating tumor DNA in patients with leiomyosarcoma with fragment size restriction.

Abstract

3046 Background: Prior work has shown that detection of circulating tumor DNA (ctDNA) at time of diagnosis of leiomyosarcoma (LMS) is associated with lower likelihood of objective response and patients with detectable ctDNA after two cycles of chemotherapy have worse survival. However, because ctDNA exists at much lower concentrations in plasma compared to cell-free DNA of non-tumor origin, the sensitivity of these prognostic measures to tumor signals remains unclear. With increasing evidence of ctDNA fragments being shorter than the background non-tumor cell-free DNA, we sought to test whether restricting our analysis to smaller fragment sizes would improve detection of ctDNA in LMS. Methods: Plasma was serially collected from patients with LMS undergoing chemotherapy. Cell-free DNA extracted from these samples was profiled by ultra-low-passage whole-genome sequencing (ULPWGS). Copy number alterations were identified and used to detect ctDNA using the ichorCNA algorithm before and after restricting the dataset to fragments of 90-150bp (short). We compared detectability of ctDNA via ichorCNA between analyses using all sequencing data and those using data restricted to short fragments. Results: From 28 patients, 126 plasma samples were profiled. The median fragment length of cell-free DNA was 240bp (IQR 142-349) for patients with LMS, compared to 307bp (IQR 171-465, p < 0.001) in samples collected from healthy controls. Short fragments made up 19.48% of LMS libraries at diagnosis when ctDNA levels were highest, 15.27% of all LMS samples, and 12.96% of libraries from healthy controls. While ctDNA was detectable by ULPWGS in 17% of all samples, detection increased to 40% when analyzed using only short cell-free DNA fragments (p < 0.0001). The proportion of diagnostic samples with detectable ctDNA was nominally higher when analyzed by short fragments (39% vs. 64% with 90-150bp size restriction, p = 0.1078, n = 28) and was significantly higher in samples collected after two cycles of chemotherapy (5% vs. 40% with 90-150bp size restriction, p = 0.0197, n = 20). Increases in ctDNA detectability with fragment size restriction were also observed in each of localized and metastatic subgroups (metastatic: 16% without restriction, 38% with restriction, n = 97, p = 0.0012; localized: 17% vs. 45%, n = 39, p = 0.04). Conclusions: Our results demonstrate that detection of ctDNA is improved by analyzing short fragments of cell-free DNA in samples collected from patients with LMS. These findings represent a potential to increase the sensitivity of an affordable, low-coverage liquid biopsy assay and may enhance the prognostic value of ctDNA detection in these patients. Further study of the association between ctDNA detection and outcome is needed to fully validate the impact of fragment size restricted analysis of cell-free DNA samples in patients with LMS and is currently ongoing in a prospective study.