Circulating Gastric Cancer Stem Cells as Blood Screening and Prognosis Factor in Gastric Cancer

Abstract

Gastric cancer (GC) is the fourth leading cause of cancer-related death, associated with late diagnosis and treatment resistance. Currently, screening tests for GC are not cost-effective or have low accuracy. Previously, we described an extended phenotype of gastric cancer stem cells (GCSCs; CD24⁺CD44⁺CD54⁺EpCAM⁺) that is associated with metastasis and tumor stage in GC patients. The goal of the current research is to evaluate the presence of these GCSCs in the peripheral blood of GC patients and healthy volunteers. A total of 73 blood samples were collected from 32 GC patients and 41 healthy volunteers. After peripheral blood mononuclear cell (PBMC) extraction, multiparametric flow cytometry was performed looking for GCSCs. Using clustering data through artificial intelligence (AI), we defined high/low levels of circulating GCSCs (cGCSCs) and proceeded to evaluate its association with clinical and prognostic variables. Finally, a diagnostic test analysis was performed evaluating patients and healthy volunteers. We found that cGCSCs are present in most GC patients with a mean concentration of 0.48%. The AI clustering showed two groups with different cGCSC levels and clinical characteristics. Through statistical analysis, we confirmed the association between cGCSC levels and lymph node metastasis, distant metastasis, and overall survival. The diagnostic test analysis showed sensibility, specificity, and area under the curve (AUC) of 83%, 95%, and 0.911, respectively. Our results suggest that the assessment of cGCSCs CD24⁺CD44⁺CD54⁺EpCAM⁺ could be a potential noninvasive test, with prognostic value, as well as highly sensitive and specific for screening or diagnosis of GC; however, a larger scale study will be necessary to confirm this.

Figure 1

cGCSCs are related with clinical characteristics and prognosis in GC patients: (a) analysis strategy in flow cytometry to evaluate cGCSCs (CD24⁺CD44⁺CD54⁺EpCAM⁺); (b) reducing dimensionality of clinical variables with artificial intelligence, it is possible to observe the two patient groups predicted by the MiniBatchKMeans algorithm, where the low cGCSC group is shown in blue and the high cGCSC group is in yellow; (c and d) stage of invasion to lymph nodes and distant metastasis according to cGCSCs %; and (e) Kaplan–Meier survival plot by cGCSC level.

Figure 2

cGCSCs in healthy volunteers and their potential as diagnostic test: (a) percentage of cGCSCs in GC patients and healthy volunteers, (b) flow cytometry analysis in healthy volunteer (#17), where CD24⁺CD44⁺ circulating cells have a EpCAM+ population but not a double-positive population, and (c) ROC curve graph with 0% of cGCSCs as cutoff for positive/negative test.  ^∗∗∗∗p < 0.0001.