Genetic profiles of cervical tumors by high-throughput sequencing for personalized medical care

Abstract

Cancer treatment is facing major evolution since the advent of targeted therapies. Building genetic profiles could predict sensitivity or resistance to these therapies and highlight disease-specific abnormalities, supporting personalized patient care. In the context of biomedical research and clinical diagnosis, our laboratory has developed an oncogenic panel comprised of 226 genes and a dedicated bioinformatic pipeline to explore somatic mutations in cervical carcinomas, using high-throughput sequencing. Twenty-nine tumors were sequenced for exons within 226 genes. The automated pipeline used includes a database and a filtration system dedicated to identifying mutations of interest and excluding false positive and germline mutations. One-hundred and seventy-six total mutational events were found among the 29 tumors. Our cervical tumor mutational landscape shows that most mutations are found in PIK3CA (E545K, E542K) and KRAS (G12D, G13D) and others in FBXW7 (R465C, R505G, R479Q). Mutations have also been found in ALK (V1149L, A1266T) and EGFR (T259M). These results showed that 48% of patients display at least one deleterious mutation in genes that have been already targeted by the Food and Drug Administration approved therapies. Considering deleterious mutations, 59% of patients could be eligible for clinical trials. Sequencing hundreds of genes in a clinical context has become feasible, in terms of time and cost. In the near future, such an analysis could be a part of a battery of examinations along the diagnosis and treatment of cancer, helping to detect sensitivity or resistance to targeted therapies and allow advancements towards personalized oncology.

Keywords: Cervix; NGS; diagnosis; panel; targeted therapie.

Figure 1

Sequencing workflow's major steps. After DNA extraction (Day 1), DNA molecules are sheared by sonification, and DNA fragments are ligated to adapters containing a patient-specific index (Day 2). DNA from 2 patients and 1 control are pooled equimolarly together during the multiplexing step (Day 3). Regions of interest (exons from 226 genes) are retrieved by a targeted enrichment system with biotinylated baits (Day 4–5). Then DNA is sequenced in an Illumina MiSeq (Day 6) and sequencing raw data are processed by the bioinformatic pipeline (Day 7–8), to extract the most likely somatic variations.

Figure 2

Representation of missense mutations found by each variant caller. (A) Mutations extracted directly from CANDID database (total: 2746). (B) mutations remaining after passing through filtration system (total: 221). (C) Proportion of mutations detected by at least 2 variant-callers classified as deleterious by SIFT or POLYPHEN.

Figure 3

Distribution of deleterious mutations among the 29 tumors. A gene is considered actionable if linked to targeted therapies approved by Food and Drug Administration. A mutation is classified as deleterious if considered as such by SIFT algorithm or POLYPHEN algorithm. (A) Proportion of tumors with a deleterious mutation on gene considered actionable (blue) and proportion of patients for whom their deleterious mutations could allow inclusion in a clinical trial (purple). (B) Representation of most mutated genes across tumors.