Inactivation of phosphatase and tensin homolog (PTEN) is caused by multiple mechanisms, and loss of PTEN activity is related to the progression of various cancers. In gastric cancer (GC), the relationship between the loss of PTEN protein expression and various genetic alterations remains unclear. The effects of microsatellite instability (MSI), Epstein–Barr virus (EBV), HER2 overexpression, and PD-L1 expression on PTEN mutation have not been fully explored. We performed comprehensive cancer panel tests with a cohort of 322 tumor samples from patients with advanced GC. Immunohistochemistry for PTEN protein was performed in all cases, and the loss of protein expression was defined as a complete absence of nuclear staining. In total, 34 cases (10.6%) had pathogenic PTEN mutations, of which 19 (55.9%) showed PTEN protein loss. The most common PTEN variants associated with protein loss were p.R130 (n = 4) followed by p.R335, p.L265fs, and deletions (n = 2). All the ten nonsense mutations identified in the samples resulted in PTEN inactivation. In the remaining 288 GC cases with wild-type PTEN, protein loss was found in 35 cases (12.2%). Thus, PTEN mutations were significantly associated with PTEN protein loss (p = 5.232 × 10−10), high MSI (p = 3.936 × 10−8), and EBV-positivity (p = 0.0071). In conclusion, our results demonstrate that loss-of-function mutations in PTEN are a frequent genetic mechanism of PTEN inactivation in GC.
Background: Recent advances in circulating cell-free DNA (cfDNA) analysis from biofluids have opened new avenues for liquid biopsy (LB). However, current cfDNA LB assays are limited by the availability of existing information on established genotypes associated with tumor tissues. Certain cancers present with a limited list of established mutated cfDNA biomarkers, and thus, nonmutated cfDNA characteristics along with alternative biofluids are needed to broaden the available cfDNA targets for cancer detection. Saliva is an intriguing and accessible biofluid that has yet to be fully explored for its clinical utility for cancer detection. Methods: In this report, we employed a low-coverage single stranded (ss) library NGS pipeline “Broad-Range cell-free DNA-Seq” (BRcfDNA-Seq) using saliva to comprehensively investigate the characteristics of salivary cfDNA (ScfDNA). The identification of cfDNA features has been made possible by applying novel cfDNA processing techniques that permit the incorporation of ultrashort, ss, and jagged DNA fragments. As a proof of concept using 10 gastric cancer (GC) and 10 noncancer samples, we examined whether ScfDNA characteristics, including fragmentomics, end motif profiles, microbial contribution, and human chromosomal mapping, could differentiate between these two groups. Results: Individual and integrative analysis of these ScfDNA features demonstrated significant differences between the two cohorts, suggesting that disease state may affect the ScfDNA population by altering nuclear cleavage or the profile of contributory organism cfDNA to total ScfDNA. We report that principal component analysis integration of several aspects of salivary cell-free DNA fragmentomic profiles, genomic element profiles, end-motif sequence patterns, and distinct oral microbiome populations can differentiate the two populations with a p value of < 0.0001 (PC1). Conclusion: These novel features of ScfDNA characteristics could be clinically useful for improving saliva-based LB detection and the eventual monitoring of local or systemic diseases.
Gastric cancer (GC) has the fifth highest incidence among cancers and is the fourth leading cause of cancer-related death GC has predominantly a higher number of cases in certain ethnic groups such as the Korean population. GC found at an early stage is more treatable and has a higher survival rate as compared with GC found at a late stage. However, a diagnosis of GC is often delayed due to the lack of early symptoms and available screening programs in United States. Extracellular RNA (exRNA) is an emerging paradigm; exRNAs have the potential to serve as biomarkers in panels aimed at early detection of cancer. We previously reported the successful use of a panel of salivary exRNA for detecting GC in a high-prevalence Korean cohort, and that genetic changes reflected cancer-associated salivary exRNA changes. The current study is a case-control study of salivary exRNA biomarkers for detecting GC in an ethnically distinct U.S. cohort. A model constructed for the U.S. cohort combined demographic characteristics and salivary miRNA and mRNA biomarkers for GC and yielded an area under the receiver operating characteristic (ROC) curve (AUC) of 0.78. However, the constituents of this model differed from that constructed for the Korean cohort, thus, emphasizing the importance of population-specific biomarker development and validation.
Although immune checkpoint inhibitors can induce durable responses in patients with multiple types of advanced cancer, only a limited number of patients have a known reliable biomarker. This study aimed to validate the IMmunotherapy Against GastrIc Cancer (IMAGiC) model, which was developed based on a previous study of four-gene and PD-L1 level, to predict immunotherapy response. We developed a clinical assay for formalin-fixed paraffin-embedded samples using quantitative real-time polymerase chain reaction to measure the expression level of the previously published four-gene set. The predictive performance was validated in a cohort of 89 patients with several advanced tumor types. The IMAGiC score was derived from tumor samples of 89 patients consisting of eight cancer types, and 73 out of 89 patients available for clinical response were analyzed with clinicopathological factors. The IMAGiC group (responder vs. non-responder) was determined with a specific value of the IMAGiC score as a cutoff, which was set by log-rank statistics for progression-free survival (PFS) divided the patients into 56 (76.7%) non-responders and 17 (23.3%) responders. Clinical responders (complete response/partial response) were higher in the IMAGiC responder group than in the non-responder group (70.6 vs. 21.4%). The median PFS of the IMAGiC responder group and non-responder was 20.8 months (95% CI 9.1-not reached) and 6.7 months (95% CI 4.9–11.1, p = 0.007), respectively. Among the 17 IMAGiC responders, 11 patients had tumor mutation burden-low and microsatellite-stable tumors. This study validated a predictive model based on a four-gene expression signature. Along with conventional biomarkers, our model could be useful for predicting response to immunotherapy in patients with advanced cancer.