The value of a nomogram based on 18F-FDG PET/CT metabolic parameters and metabolic heterogeneity in predicting distant metastasis in gastric cancer

Abstract Objective To investigate the value of metabolic parameters and metabolic heterogeneity from pretreatment deoxy-2-[fluorine-18]-fluoro-D-glucose positron emission tomography/computed tomography (18F-FDG PET/CT) in predicting distant metastasis in gastric cancer. Methods Eighty-six patients with pathologically confirmed gastric adenocarcinoma were included in this study. All patients underwent a whole-body 18F-FDG PET/CT scan before treatment. Clinicopathologic and imaging data were collected, including metabolic parameters such as maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) of the primary gastric cancer lesions. Heterogeneity index (HI)-1 was expressed as the absolute value of the linear regression slopes between the MTVs at different SUVmax thresholds (40% × SUVmax, 80% × SUVmax), while HI-2 was expressed as the difference between SUVmax and SUVmean. Patients were randomly divided into training and validation cohorts at a 7:3 ratio. The correlation between the above parameters and distant metastasis in gastric cancer was analyzed using the training cohort. A nomogram prediction model was then established and later verified with the validation cohort. Finally, decision curve analysis was used to evaluate the clinical utility of the model. Results This study included 86 patients with gastric cancer, with 60 (69.8%) in the training cohort and 26 (30.2%) in the validation cohort. There was no significant difference in the balanced comparison between both cohorts (all P > .05). Among all patients, 31 (36.0%) developed distant metastasis, while 55 (64.0%) did not. In patients who developed distant tumor metastasis, carcinoembryonic antigen, carbohydrate antigen (CA)12–5, CA19–9, CA72–4, MTV, TLG, and HI-1 were significantly higher than in patients without distant metastasis (all P < .05). Multivariate logistic regression analysis identified CA72–4 (OR: 1.151, 95% CI: 1.020–1.300, P = .023) and HI-1 (OR: 1.647, 95% CI: 1.063–2.553, P = .026) as independent risk factors for predicting distant metastasis in gastric cancer. The nomogram constructed from this analysis exhibited high predictive efficacy in the training (AUC: 0.874, 95% CI: 0.766–0.983) and validation (AUC: 0.915, 95% CI: 0.790–1.000) cohorts, providing a net clinical benefit for patients. Conclusion HI-1 is an independent risk factor for predicting distant metastasis in gastric cancer. A comprehensive prediction model combining HI-1 with the tumor marker CA72–4 can increase the net clinical benefit for patients.