To investigate the risk factors of moderate to severe myelosuppression in patients with pancreatic ductal adenocarcinoma (PDAC) after the first chemotherapy, and to establish a predictive model.

A cluster sampling method was used to select the clinical data of 123 patients with PDAC confirmed by pathological diagnosis and chemotherapy in the First Affiliated Hospital of Xinjiang Medical University from January 2018 to December 2022. 68 patients with mild or no myelosuppression were set as the control group, and 55 patients with moderate or severe myelosuppression were set as the observation group. Logistic univariate and multivariate regression analysis screened out the risk factors of moderate and severe myelosuppression after the first chemotherapy and constructed the prediction model. Hosmer-Lemeshow test and receiver operating characteristic (ROC) curve were used to determine the predictive efficiency of the regression model.

By multivariate analysis, absolute neutrophil count <3.40×10⁹/L (OR=2.890, 95% CI: 1.332-6.269, P=0.007), hemoglobin <118 g/L (OR=2.374, 95% CI: 1.092-5.163, P=0.029) and CONUT≥2 (OR=2.635, 95% CI: 1.206-5.756, P=0.015) were independent risk factors for moderate to severe myelosuppression after chemotherapy in PDAC patients. The constructed regression prediction model had sensitivity of 74.5%, specificity of 64.7%, Yoden index of 0.393, and area under ROC curve of 0.716 (95% CI: 0.625-0.807) had good predictive value, and the Hosmer-Lemeshow test had a good fitting degree (χ²=3.335, P=0.766).

The scores of absolute neutrophil count, hemoglobin and nutrient control before chemotherapy in PDAC patients are important factors affecting moderate to severe myelosuppression after the first chemotherapy. Early diagnosis and intervention before chemotherapy are expected to improve the prognosis of patients.