Background: For locally advanced or metastatic gastroenteropancreatic neuroendocrine carcinomas (GEP-NECs), standard chemotherapy typically incorporates etoposide or irinotecan plus platinum agents. Herein, we initiate a phase I/II trial to apply Nal-IRI in combination with carboplatin as the 1^st line treatment in patients with GEP-NECs to define the dose-limiting toxicity (DLT) and recommended phase II dose (RP2D). Methods: In the phase I part, traditional 3+3 design was applied. Enrolled patients would receive escalating dose of 60 (level -1), 80 (starting dose, level 0) or 100 (level 1) mg/m² salt-form nal-IRI plus carboplatin AUC=4 with maximum total dose of 600 mg on D1, every 21 days as a cycle. DLTs were evaluated during the first one cycle of treatment with tumor assessment every 6 weeks. Prophylaxis G-CSF was not allowed in the first cycle Paired testing of tissue (TBx) and liquid biopsy (LBx) was done by Illumina TSO500 platform before treatment. The second LBx would be repeated for confirmed partial responders when progression. Results: Totally 6 GEP-NEC patients with primary sites of colon in three, and each one in esophagus, ampulla of Vater and pancreas were enrolled to level 0. One patient had DLT of grade 4 neutropenia lasting for 3 days even under salvage G-CSF. Treatment-related adverse events showed grade 3/4 neutropenia (33.4%), grade 2 anemia (16.7%), grade 3 fatigue (16.7%), and grade 2 skin rash (16.7%). Four patients achieved confirmed partial response and 2 patients got stable disease, with the objective response rate of 66.7% (95% confidence interval: 22%-96%). After the data and safety monitoring board meeting, level 0 was decided as RP2D but not escalation to level 1 because of acceptable toxicity profiles and promising efficacies. Comprehensive genomic profiling showed that driver mutations were identified in three cases including two harboring BRAF^V600E mutation and one with KRAS^G12D mutation. The other three cases lacking driver mutation showed frequent copy number alteration including 1 with MDM2 amplification (copy number 27) and a wild-type TP53. Homologous recombination deficiency (HRD), as determined by GIS score, was identified in a colon GEP-NEC who remained to be HRD-positive at progression but a higher TMB (from 4.7 to 10.2 Muts/Mb) and a higher BRAF^V600E mutation allele frequency (from 24.1% to 43.6%) in tumor tissue, compared to the baseline profiling. Conclusions: Based on current data, the RP2D is 80 mg/m² of nal-IRI (salt-form) and carboplatin of AUC=4, every 3 weeks. The extension phase II study in GEP-NEC is ongoing. Clinical trial information: NCT05385861.