Memorial Sloan Kettering Cancer Center (MSK) researchers have identified genomic changes in recurrent gastric tumors that could improve personalized care and post-operative monitoring of this deadly type of cancer. In the study, researchers used large-scale genetic sequencing data to identify patients at risk of recurrence of primary gastric cancer, and to predict the pattern of metastasis after surgery. Findings showed that the oncogenes KRAS and PIK3CA were linked to shorter disease-free survival (DFS), and revealed specific genomic changes associated with patterns of metastasis. The findings were published online on August 5, 2026 in JAMA Surgery.
Genomic changes in other types of cancers including lung and colorectal cancer are currently being used to inform clinical decision-making and may provide information about disease pathogenesis. The greater diversity of gastric cancer, however, has made efforts to understand its genomics more challenging. The MSK study is one of the first in an American cohort to characterize certain genomic changes in gastric cancer that could potentially improve identification and management of tumors at higher risk for recurrence.
“This evaluation of genomic alterations in over 350 gastric tumors resected and sequenced at a single institution identifies distinct genomic associations with postoperative outcome, including DFS and patterns of recurrence. These findings complement current molecular and histologic classifications of gastric cancer. Clinically, the genomic insights from our cohort may improve risk stratification, informing the initiation of adjuvant treatment, surveillance strategies, and entry into clinical trials,” said senior author Vivian Strong, MD, FACS, who holds the Iris Cantor Chair as a distinguished senior translational researcher, with specialization in stomach cancer, adrenal tumors, minimally invasive surgery, laparoscopic surgery, and robotic surgery.
Gastric Cancer: Deadly and Diverse
Globally, gastric cancer ranks as the fifth leading cause of death from cancer. While surgery can cure gastric cancer, recurrence rates are high. Relapse typically develops within two years of surgery and is associated with shorter survival. Recent improvements in systemic therapy around the time of surgery have decreased recurrence rates, although this risk remains high.
When gastric cancer recurs, it predominantly spreads via the blood or lymphatic system, or to the peritoneum. Other studies, including an analysis of past patients with recurrent gastric cancer enrolled in the large JCOG 1001 Japanese randomized controlled clinical trial found primary tumor characteristics that may increase risk for spread, with the most consistently identified risk factor being Lauren histological subtype. This classification system groups gastric cancer into the following variants based on examination under the microscopic:
- Intestinal subtype: More common, characterized by gland formation
- Diffuse subtype: Less common, characterized by intercellular adhesion molecules, infiltrative growth, and without gland formation)
- Mixed subtype: Characteristics of both intestinal and diffuse subtypes
- Indeterminate subtype: Unable to assign to other categories
Past studies have linked higher rates of recurrence to the diffuse Lauren histological subtype.
Study Design
The retrospective cohort study took place at MSK, and included 438 patients (median age 65 years, median DFS 5.17 years and median overall survival 6.56 years), who had surgery for gastric cancer with the intent of cure between 2010 and 2024. Patients received perioperative systemic therapy according to National Cancer Network Guidelines (NCCN) and were staged according to the 8th edition of the American Joint Committee on Cancer manual. Recurrence after surgery, organ involvement, and spread of disease were determined by physician medical record review.
The primary outcomes were DFS and patterns of disease spread and were correlated with primary tumor genomic profiles. Patterns of disease spread were defined as:
- Hematogenous: Metastasis through the blood to distant organs including adrenal gland, bone, brain, liver, lung, pleura, soft tissue, or spinal cord
- Peritoneal: Lesions in the peritoneal surface or ovary
- Lymphatic: Recurrence in the lymph nodes or solid organs.
Genetic testing was conducted using MSK-IMPACT a next generation laboratory assay that analyzes hundreds of genes in a single test and can detect 341 of 505 cancer-associated genes. Researchers used MSK’s precision knowledge data base OncoKB to classify clinically significant genomic changes as oncogenic or likely oncogenic, which may guide treatment.
In a subanalysis, researchers evaluated surgical pathology reports of patients with lymph node negative or lymph node positive disease. This analysis excluded patients with tumors with serosal invasion and a large amount of lymphatic spread. It also excluded those who received neoadjuvant therapy before surgery, which could downstage the tumor and confound results.
Specific Oncogenes and Chromosomal Instability
The analysis included 377 patients with negative surgical margins, genetic testing of their tumors, and recurrence within 2 years of follow-up or sufficient follow-up (median [IQR] age, 64 [55-71] years; 113 female [30%] and 264 male [70%]). Among these, 179 developed recurrence and 198 had no evidence of disease. Median time to recurrence was 1.10 years, with peritoneal spread occurring later than spread through the blood (1.34 years vs 0.88 years; P = .005). Median survival after diagnosis of relapse was 1.47 years.
Patients with no evidence of disease had tumors that were more often high in microsatellite instability (MSI) than those who developed recurrence (14% vs 3%; P < .001). MSI refers to short, repetitive stretches of DNA that can indicate a deficiency in cellular DNA mismatch repair mechanisms. Gastric tumors that are high in MSI can be more responsive to immune checkpoint inhibitors.
Analyses restricted only to microsatellite-stable (MSS) tumors (n = 345) showed that larger tumor size and higher pathologic stage were independently associated with shorter DFS, as were the following:
- KRAS oncogene (hazard ratio [HR], 1.54; 95% CI, 1.04-2.28; P = .03). When mutated, KRAS can promote uncontrolled cell growth and spread in gastric cancer. In the study, changes in KRAS were more prevalent in recurrent tumors than tumors from patients with no evidence of disease (18% vs 10%; P = .03; Q = 0.46)
- PIK3CA oncogene (HR, 2.15; 95% CI,1.25-3.69; P =.006 ). PIK3CA can cause an enzyme called P13K to become overactive. This can promote cancer cell proliferation and survival, particularly in Epstein-Barr virus positive gastric tumors
Analyses looking at patterns of disease spread showed that hematogenous spread was most common. Greater chromosomal instability was associated with hematogenous spread (fraction genome altered [FGA] 0.11 vs 0.03; P = .001), whole-genome duplication [WGD] 47% vs 15%; P = .02), and more frequent changes in genes modulating cell cycle regulation (39% vs 6%; P <.001), compared to peritoneal spread. However, bone metastasis was associated with more genomically stable primary tumors, compared to hematogenous spread to other sites of the body (FGA 0.005 vs 0.114; P < .001).
Additional analyses highlighted the following:
- Lauren diffuse-type histology: Associated with bone metastasis compared with other sites of hematogenous spread (36% vs 3%; P < .001). This association has been shown in studies in Europe and Asia.
- Lauren mixed-type gastric cancers: Associated with lymph node metastasis compared to all other tumors (76% vs 45%; P=.01). The MSK study is the first American cohort to show this result, along with associated genomic findings. The authors note that these findings are exploratory and require further investigation.
Limitations
Limitations of the study include its retrospective design. A further limitation is possible selection bias related to including tumor tissue that had already been sequenced after diagnosis of relapse, which could have overestimated relapse rates. Additionally, not all tumors were sampled before patients received neoadjuvant therapy, which could have resulted in tumor downstaging and affected results. Advances in cancer care over the 15 years of the study could also have influenced the findings. Finally, peritoneal recurrences can be challenging to diagnose, which could have resulted in inaccurate estimates of this pattern of spread.
Personalized Medicine: Genomic Profiling and Accessibility
The results suggest a role for genomic profiling to identify patients at risk for recurrence of primary gastric cancer and to guide personalized management after surgery. KRAS and PIK3CA have been associated with recurrence in other cancer types and could potentially be used as tumor markers to identify patients with worse prognosis.
Place holder: “These alterations have therapeutic relevance, as both KRAS targeted and PIK3CA-targeted therapies have demonstrated efficacy and earned regulatory approval in other solid tumors. As the use of molecularly guided therapy gains momentum in oncology, our findings present a potential opportunity for next generation inhibitors to be evaluated in KRAS-altered and PIK3CA-altered gastric cancer,” said Dr. Strong.
Changes in KRAS and PIK3CA could also be used for selection of monitoring modalities that could be used to identify recurrence earlier after surgery, for trial enrollment, and to help guide treatment decisions to lower the risk of recurrence.
“Importantly, access to next generation sequencing is not universal, and gene panel sequencing typically costs more than $1000 per sample. Clinical application of molecular diagnostics ultimately depends on both discovery and accessibility,” Dr. Strong added.
Disclosures
Dr. Shar reports grants from Astrazeneca and Bristol Myers Squibb. Dr. Laszkowska reports research funding from AI Medical Services. Dr. Vardhana reports grants from Bristol Myers Squibb. Dr. Strong reports personal fees from AstraZeneca. No other disclosures were reported.