4CPS-106 Multi-state model to estimate the optimal duration of first-line chemotherapy in advanced gastric cancer: data from the national registry agamenon

Background To date, there has been no phase III trial to establish the optimal duration of first-line chemotherapy for patients with advanced gastric cancer (AGC): limited treatment, maintenance of some drugs or treatment until progression. Purpose Assess prognostic factors and progression-free survival (PFS) in each stratum stablished according to the duration of the treatment. Material and methods The sample comprises patients from the AGAMENON multicentre registry in which 31 Spanish and one Chilean centre participated. The eligibility criteria include adults (≥18 years) with a histologically confirmed unresectable or AGC and first-line polychemotherapy without progression in the second evaluation of response at approximately 6 months. Multi-state models were used to model processes in which participants undergo transitions from one state to another (e.g., from initiation to cessation of drugs, and from that point to progression or death). In order to examine time-varying states, a Markov multi-state model was used. On the cumulative scale, the transition probability matrix was established by the Aalen–Johansen estimator. Results We analysed 415 patients treated between January 2008 and September 2017. The patients were divided into three strata: discontinuation of platinum and maintenance with fluoropyrimidine until progression (30%, n=123); complete treatment withdrawal prior to progression (52%, n=216); and full treatment until progression (18%, n=76). Compared to those receiving treatment until progression, no decrease in PFS was observed in participants who discontinued all treatment (HR 1.16, 95% CI, 0.70 to 1.92) or in whom platinum was suspended (HR 0.92, 95% CI, 0.54 to 1.58). With respect to PFS prognostic factors, a significant effect was observed for ECOG ≥2 in stratum 3 (HR 4.06, 95% CI, 1.40 to 11.7). The presence of ≥3 metastatic sites revealed a prognostic effect after discontinuing platinum (HR 1.65, 95% CI, 1.06 to 2.56) or all chemotherapy (HR 1.65, 95% CI, 1.06 to 2.56). Bone metastases were an adverse prognostic factor in all the strata (HR 1.64, 95% CI, 1.13 to 2.37). Complete response was a protective factor after withdrawing the entire regimen (HR 0.31, 95% CI, 0.16 to 0.57) or platinum (HR 0.12, 95% CI, 0.03 to 0.41). No significant interactions between covariates were detected. Conclusion In this registry of AGC, treating until progression did not impact PFS compared to maintenance or discontinuation after a predefined number of cycles. References and/or acknowledgements We thank all the researchers of the AGAMENON registry. No conflict of interest.