A gene expression signature based on tumor cell lineage/state compositions can predict survival in patients with peritoneal carcinomatosis.
A gene expression signature based on tumor cell lineage/state compositions can predict survival in patients with peritoneal carcinomatosis (PC), according to a study out of The University of Texas MD Anderson Cancer Center that was published in Nature Medicine.1,2
The discovery was made by profiling more than 45,000 individual cells from patients with PC, revealing extensive cellular heterogeneity.
“In order to better treat patients with PC, we first have to understand the populations of metastatic cells in the peritoneal cavity,” said co-author Linghua Wang, MD, PhD, assistant professor of genomic medicine at The University of Texas MD Anderson Cancer Center. “This is the most detailed analysis of these cells performed to date. That is the power of single-cell analysis—we are able to look at every single cell and get a picture of the landscape.”
PC, a subtype of metastatic gastric cancer, represents a major unmet clinical need, as overall survival is less than 6 months.
In the analysis, researchers dissected, at single-cell resolution, the cellular and transcriptomic intratumoral heterogeneity (ITH) of cryopreserved PC tumor cells from 20 patients with gastric adenocarcinoma (GAC) using single-cell transcriptome sequencing (scRNA-seq) technology in combination with integrative computational analyses.
“A key finding of this study is that the diversity in tumor cell lineage/state compositions appears to mirror and may even dictate the inherent ITH of PC tumor cells at multiple levels,” the investigators wrote.
Based on tumor cell lineage compositions, PC samples were classified into 2 main subtypes: gastric-dominant (mainly gastric cell lineages) and gastrointestinal (GI)-mixed (with mixed gastric and colorectal-like cells).
All cases were clinically diagnosed as PC from GAC, but transcriptome-based analysis revealed a high degree of cellular heterogeneity in inferred tumor cell lineages. Only approximately 70% of mapped PC tumor cells were defined as cells of stomach origin. However, the expression profiles of a subset of PC tumor cells (26%) transcriptomically resembled cells of other GI organs, particularly the intestine (21%).
A significant survival difference was found between the 2 subtypes. All 6 cases with a GI-mixed phenotype were long-term survivors, whereas 6 of 8 cases with a gastric-dominant phenotype were short-term survivors (P = .05).
A 12-gene expression signature was generated from single-cell differentially expressed gene (DEG) analysis on PC tumor cells between the gastric-dominant and GI-mixed subtypes, followed by filtering the DEGs to list to identify the most significant DEGs.
“The signature demonstrated an excellent power to prognosticate patient survival,” the authors wrote. Consistently, patients whose PCs were in the gastric-dominant group had significantly shorter survival than those whose PCs were in the GI-mixed group (7.8 vs 24.5 months). “Multivariate Cox regression analysis showed that this signature was a strong prognosticator of short survival, with a hazard ratio of 12.7 (95% CI, 3.2–51.0; P = 3.3 x 10−4), and it was independent of clinical/histopathological variables.”
The gene signature retained its prognostic significance when evaluated in 4 other large-scale localized GAC cohorts totaling 1336 patients, even though the signature was derived from a cohort with advanced GAC, the authors noted.
If the gene expression signature is validated in prospective studies, it may be useful in stratifying patients with gastric cancer and directing them for more effective treatment strategies.
“Based on our findings, we need to move toward profiling these [PC tumor] cells in each patient in order to offer more tailored treatment options,” commented co-author Jaffer Ajani, MD, professor of gastrointestinal medical oncology at The University of Texas MD Anderson Cancer Center.
“This is an important first step toward a better understanding of the single-cell biology of these cancer cells, but we have more work to do,” he concluded. “We foresee that understanding this heterogeneity could one day be used to guide clinical decision making that is most beneficial to each patient.”