Optimal postoperative surveillance strategies for cancer survivors with gastric neuroendocrine carcinoma based on individual risk: a multicenter real-world cohort study

Abstract

Background: The best follow-up strategy for cancer survivors after treatment should balance the effectiveness and cost of disease detection while detecting recurrence as early as possible. Due to the low incidence of gastric neuroendocrine carcinoma and mixed adenoneuroendocrine carcinoma [G-(MA)NEC], high-level evidence-based follow-up strategies is limited. Currently, there is a lack of consensus among clinical practice guidelines regarding the appropriate follow-up strategies for patients with resectable G-(MA)NEC.

Materials and methods: The study included patients diagnosed with G-(MA)NEC from 21 centers in China. The random forest survival model simulated the monthly probability of recurrence to establish an optimal surveillance schedule maximizing the power of detecting recurrence at each follow-up. The power and cost-effectiveness were compared with the National Comprehensive Cancer Network, European Neuroendocrine Tumor Society, and European Society for Medical Oncology Guidelines.

Results: A total of 801 patients with G-(MA)NEC were included. The patients were stratified into four distinct risk groups utilizing the modified TNM staging system. The study cohort comprised 106 (13.2%), 120 (15.0%), 379 (47.3%), and 196 cases (24.5%) for modified groups IIA, IIB, IIIA, and IIIB, respectively. Based on the monthly probability of disease recurrence, the authors established four distinct follow-up strategies for each risk group. The total number of follow-ups 5 years after surgery in the four groups was 12, 12, 13, and 13 times, respectively. The risk-based follow-up strategies demonstrated improved detection efficiency compared to existing clinical guidelines. Further Markov decision-analytic models verified that the risk-based follow-up strategies were better and more cost-effective than the control strategy recommended by the guidelines.

Conclusions: This study developed four different monitoring strategies based on individualized risks for patients with G-(MA)NEC, which may improve the detection power at each visit and were more economical, effective. Even though our results are limited by the biases related to the retrospective study design, we believe that, in the absence of a randomized clinical trial, our findings should be considered when recommending follow-up strategies for G-(MA)NEC.

Trial registration: ClinicalTrials.gov NCT05671393.

Graphical abstract

Figure 1

The study flow chart.

Figure 2

Risk-adjusted cumulative recurrence rate of patients in each mTNM group.

Figure 3

The risk-adjusted recurrence rate of each month after surgery in each mTNM group.

Figure 4

Comparison of effectiveness among the different follow-up strategies in each mTNM group. The x-axis represents the given number of follow-up visits within 5 years after surgery. The y-axis represents the month of delayed detection of disease recurrence. The dark blue curve represents the risk-based surveillance strategies. The red, blue, orange, and green dots represent the follow-up strategies of ENETS-v1, NCCN, ENETS-v2, and ESMO, respectively. The error bars indicate the 95% CIs. For some points, the 95% CIs would be shorter than the height of the symbol.

Figure 5

Postoperative follow-up details for each mTNM group. The highlighted ‘o’ represents the months recommended for visits.