Article
Artificial intelligence matches cancer genotypes to patient phenotypes
The advent of liquid biopsies targeting genetic mutations in solid tumors is a major milestone in the field of precision oncology.1 Conventional methods of obtaining tissue for molecular studies are limited by sample size and often do not represent the entire bulk of the tumor.2 This newer minimally invasive, revolutionary technique analyzes circulating cell-free DNA carrying tumor-specific alterations (circulating tumor DNA [ctDNA]) in peripheral blood and detects signature genomic alterations.1 Tp53 mutations have been reported in 25 to 40% of prostatic cancers and > 50% of non-small cell lung cancers (NSCLC), being more common in late-stage and hormone refractory prostate cancers.3,4 Tp53 mutation has been found to be associated with poor prognosis and increased germline mutations.5
The veteran patient population has distinct demographic characteristics that make veterans more vulnerable to genetic mutations and malignancies, including risk of exposure to Agent Orange, smoking, substance abuse, and asbestos. This area is understudied and extremely sparse in the literature for frequency of genetic mutations, risk factors in solid malignancies occurring in the veteran patient population, and the clinical impact of these risk factors. We herein present a quality assurance study for the utility of liquid biopsies regarding the frequency of DNA damage repair (DDR) gene, Tp53, and androgen receptor (AR) mutations. The clinical impact in advanced lung and prostate cancers in the veteran patient population and frequency are the quality assurance observations that are the study endpoints.
Methods
We reviewed for quality assurance documentation from the Foundation Medicine (www.foundationmedicine.com) cancer biomarker tests on liquid biopsies performed at the Corporal Michael J. Crescenz Veteran Affairs Medical Center in Philadelphia, Pennsylvania from May 2019 to April 15, 2020. All biopsies were performed on cancers with biochemical, imaging or tissue evidence of advanced tumor progression. The testing was performed on advanced solid malignancies, including NSCLC, prostate adenocarcinoma, and metastatic colon cancer. Statistical data for adequacy; cases with notable mutations; frequency; and type of mutations of AR, DDR, and Tp53 were noted. General and specific risk factors associated with the veteran patient population were studied and matched with the type of mutations (Table 1).
Results
Thirty-one liquid biopsies were performed over this period—23 for prostate cancer, 7 for patients with lung cancer patients, and 1 for a patient with colon cancer. Of 31 cases, sensitivity/adequacy of liquid biopsy for genetic mutation was detected in 29 (93.5%) cases (Figure 1). Two inadequate biopsies (both from patients with prostate cancer) were excluded from the study, leaving 29 liquid biopsies with adequate ctDNA for analysis that were considered for further statistical purpose—21 prostate, 7 lung, and 1 colon cancer.
Multiple (common and different) genetic mutations were identified; however, our study subcategorized the mutations into the those that were related to prostate cancer, lung cancer, and some common mutations that occur in both cancers. Only the significant ones will be discussed in this review and equivocal result for AR is excluded from this study. Of the 21 prostate cancers, 4 (19.0%) had directed the targeted therapy to driver mutation (AR being most common in prostate cancer), while KRAS mutation, which was more common in lung cancer, was detected in 2/7 (28.6%) lung cancers. Mutations common to both cancer types were DDR gene mutations, which is a broad name for numerous genes including CDK12, ATM, and CHEK2.
Of all cases irrespective of the cancer type, 23/29 (79.3%) showed notable mutations. DDR gene mutations were found in 6 of 21 (28.5%) patients with prostate cancer and 8 of 23 (34.7%) patients with advanced prostate and lung cancers, indicating poor outcome and possible resistance to the current therapy. Of 23 patients showing mutations irrespective of the cancer type, 15 (65.2%) harbored Tp53 mutations, which is much more frequent in veteran patient population when compared with the literature. Fifteen of the 31 (48.4%) total patients were Vietnam War-era veterans who were potentially exposed to Agent Orange and 20 (64.5%) patients who were not Vietnam War-era veterans had a history that included smoking (Figure 2).