<p>The combination of immune checkpoint inhibitors (ICIs) and chemotherapy has become the standard first-line treatment for advanced gastric cancer and gastroesophageal junction cancer (GC/GEJC), yet reliable biomarkers predicting efficacy remain limited. We performed longitudinal plasma proteomic profiling in 31 patients receiving immunochemotherapy in this prospective observational cohort study. Serial plasma samples collected at baseline and every two treatment cycles were analyzed to identify dynamic protein changes correlated with therapeutic response and survival. The results showed that responders exhibited significantly lower baseline interleukin-15 (IL-15) levels (<i>P</i> = 0.01). After two cycles of treatment, a 161.9% reduction in plasma Mucin-16 (MUC-16), also known as Cancer Antigen-125 (CA-125), was identified as an independent factor for prolonged overall survival (HR 7.40, 95%CI 2.22–24.60, <i>P</i> = 0.001), which was validated by conventional serum CA-125 testing. Integrating baseline IL-15 with dynamic changes in MUC-16 and Matrix Metalloproteinase 12 (MMP12), the constructed composite risk model achieved an Area Under the Curve (AUC) of 0.799 for predicting response, and this model was also an independent factor for progression-free survival (HR 3.35, 95% 1.56–7.20, <i>P</i> = 0.002). In summary, baseline IL-15 helps identify patients likely to benefit from immunochemotherapy, while MUC-16 and the integrated model facilitate dynamic monitoring of therapeutic efficacy. These findings provide new insights into non-invasive blood-based patient stratification and treatment optimization.</p><p><b>Trial registration</b> ChiCTR2300078000. Registered 27 November 2023—prospectively registered, <a href="https://www.chictr.org.cn/hvshowproject.html?id=241839&amp;v=1.0">https://www.chictr.org.cn/hvshowproject.html?id=241839&amp;v=1.0</a></p>