Integrating Circulating Tumor DNA Features and Plasma Protein Markers to Detected Early Lymphoid Neoplasm

Introduction

The disease burden of lymphoid neoplasm has been rising in China over the last decade. But most patients manifest with advanced stage disease at initial diagnosis, and the prognosis is poor with a 5-year survival rate of 38.3%. Here we reported a novel multivariate cancer risk score (CRS) model which is used to detect early lymphoid neoplasm from the peripheral blood. It incorporates three cancer hallmarks, copy number aberrations (CNA) and fragment size (FS) via shallow whole genome sequencing (sWGS) from cell-free DNA (cfDNA), and a panel of seven tumor protein markers in a single blood draw (10ml).

Methods

44 newly diagnosed and untreated stage I-IV lymphoid neoplasm patients and 247 healthy individuals with no cancer diagnosis were enrolled in this study. 10ml peripheral blood was collected from each participant after enrollment. cfDNA was extracted and subjected to sWGS whereas plasma was subjected to measure the levels of 7 PTMs. The cancer risk score (CRS) of a subject was calculated via an established CRS model ^[1].

Results

Firstly, genomic and epigenetic features were explored from cfDNA sWGS results. CNA is a ubiquitous genomic hallmark in a wide spectrum of cancers. In this study, 26 of the 44 (59.1%) lymphoid neoplasm patients had CNA in at least one genomic segment (>5Mb). FS feature of cfDNA bears the correspondence of the epigenetic landscapes of cells that give rise to those cfDNA fragments. When CNA was combined with FS, 29 (65.9%) patients were able to be detected by the CNA+FS classifier. On the other hand, 13 (29.5%) patients were tested positive by PTMs alone, indicating non-DNA molecular surrogates can also serve as cancer biomarkers with acceptable performance. When CNA, FS and PTM were incorporated into a multidimensional and multivariate CRS model, it achieved the best performance allowing 31 (70.0%) lymphoid neoplasm cases to be identified with a positive predictive value (PPV) of 86.1% at 98.0% specificity. The sensitivity of CRS model increases with the advances of disease with a sensitivity of 50.0% in early stage (stage I -Ⅱ) and 90.0% in late stage (stage Ⅲ-Ⅳ).

Conclusion

In summary, this study provides an efficient and non-invasive method to detect lymphoid neoplasm. Instead of relying only on one dimension of cancer markers, the multidimensional approach which incorporating CNA, fragment size and protein markers is plausible in early detection of lymphoid neoplasm with sufficient accuracy and robustness.