The serum human neutrophil lipocalin may be a promising biomarker for the diagnosis of cancer patients with bacterial infection

Background Cancer patients undergoing chemoradiotherapy are highly susceptible to bacterial infections. This study evaluated the diagnostic performance of inflammatory biomarkers in bacterial infections among cancer patients. Methods In this study, 334 tumor patients were enrolled, including 169 cases in the bacterial infection group and 165 cases in the control group. In the process of evaluating the diagnostic value of inflammatory markers, these 334 cases were divided into a training set (n = 233) and a validation set (n = 101). Serum levels of human neutrophil lipocalin (HNL), procalcitonin (PCT), prealbumin (PA) and routine hematological indices were measured. Diagnostic efficacy was assessed using receiver operating characteristic (ROC) analysis, while optimal biomarker combinations were identified via logistic regression. Pathogen-type discrimination and antimicrobial resistance patterns were assessed. Results Infected patients exhibited significantly elevated HNL (256.5 vs 164.2 ng/mL, P  < 0.001), PCT (0.33 vs 0.11 µg/L, P  < 0.001), white blood cell (WBC), neutrophils (NEU), D-dimer, and systemic inflammation indices neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII)), with reduced PA (0.146 vs 0.188 g/L) and lymphocytes (LYM) (0.82 vs. 0.97×10 ⁹ /L) (all p  < 0.05). HNL (area under the curve (AUC) = 0.853) and PCT (AUC = 0.828) demonstrated superior individual diagnostic performance. The combination of PCT, HNL, and PA yielded optimal diagnostic value (AUC = 0.949, 87% sensitivity, 95% specificity), with satisfactory calibration and learning curves. PCT levels were significantly higher in gram-negative vs gram-positive bacterial infections (0.41 vs 0.145 ng/L, p  < 0.001), showing discriminative capacity (AUC = 0.807), at a cutoff value of 0.195 µg/L. Conclusions The results demonstrated that HNL exhibited the most outstanding diagnostic performance in this study, and the combination of HNL, PCT and PA constituted the optimal diagnostic model for these infections. Furthermore, PCT exhibited significant diagnostic value in distinguishing between gram-negative and gram-positive bacterial infections. Collectively, these findings offered clinical guidance for selecting inflammatory biomarkers in diagnosing bacterial infections among cancer patients.