Background: Endoscopic ultrasound (EUS) is a guideline-recommended diagnostic test to estimate pretreatment clinical stage in gastric cancer. The impact of EUS to discriminate long-term outcomes has not been established.
Objectives: The objectives of our study were to (1) evaluate the association between EUS and pathologic stage; (2) evaluate the ability of EUS to predict disease-specific survival (DSS); and (3) determine how neoadjuvant chemotherapy (NCT) affects these relationships.
Methods: A prospective gastric cancer database at a tertiary care cancer center identified 734 patients who underwent curative intent resection. Patients were separated into EUS low-risk (T1-2, N0) and EUS high-risk (T3-4 Nany, or Tany N+) groups. Agreement statistics and 5-year DSS were estimated stratified by NCT.
Results: Between 1987 and 2015, 68% (502/734) of patients were not treated with NCT. Among these patients, percentage agreement between EUS and pathology was moderate (individual T stage: 52%; N stage: 70%; risk group: 73%). EUS accurately estimated pathologic risk group in 73% (365/502) of patients, whereas it overestimated pathologic risk group in 19% (93/502) of patients and underestimated risk in 8% (41/502) of patients. EUS in non-NCT staging was able to discriminate DSS for T stage (hazard ratio [HR] 5.07, p < 0.05), N stage (HR 3.58, p < 0.05), and risk group (HR 6.35, p < 0.05). Among patients treated with NCT, EUS was unable to discriminate DSS for T stage (HR 0.94, p > 0.05), N stage (HR 1.46, p > 0.05) and risk group (HR 0.50, p > 0.05).
Conclusions: Pretreatment clinical staging based on EUS alone could lead to over- or under treatment in 27% of patients and can discriminate DSS in NCT-naive patients. EUS should be used in the context of other validated clinical risk tools.