A Nomogram Integrating CD169%, Neutrophil CD64 Index, and C-Reactive Protein for Differential Diagnosis of Mixed...

Background: Early discrimination between viral, bacterial, and mixed respiratory tract infections remains challenging in clinical practice. Conventional microbiological methods may be limited by turnaround time and specimen-related factors, whereas routine inflammatory markers lack sufficient specificity for etiological classification. We investigated whether flow cytometry-derived immune markers combined with C-reactive protein (CRP) could improve early infection stratification. Methods: In this prospective single-center study, adult patients with acute respiratory tract infection and healthy controls were enrolled between 29 July 2025 and 31 January 2026. Peripheral blood CD169 positivity on monocytes (CD169%), neutrophil CD64 index (nCD64 index), monocyte HLA-DR positivity (HLA-DR%), and CRP were measured and compared across viral infection (influenza A, influenza B, or COVID-19), bacterial infection, mixed infection, and healthy control groups. Logistic regression models were developed for three diagnostic tasks: viral versus bacterial infection, mixed versus bacterial infection, and mixed versus viral infection. Model performance was evaluated using receiver operating characteristic (ROC) analysis, DeLong testing, calibration, decision curve analysis (DCA), and bootstrap internal validation. Nomograms were constructed from the final models. Results: A total of 216 infected patients were included: 112 with viral infection, 67 with bacterial infection, and 37 with mixed infection; 50 healthy individuals served as controls. CD169% was increased in viral and mixed infection, whereas the nCD64 index was increased in bacterial and mixed infection. HLA-DR% was lower in bacterial infection. For differentiating viral from bacterial infection, the optimized three-marker model combining CD169%, nCD64 index, and CRP achieved an area under the ROC curve (AUC) of 0.8947 (95% CI: 0.8463-0.9432), with a sensitivity of 81.25% and a specificity of 80.70%. In mixed infection, CD169% was the main discriminator versus bacterial infection, while the nCD64 index and CRP were more informative versus viral infection. The combined model yielded an AUC of 0.7618 (95% CI: 0.6485-0.8751) for mixed versus bacterial infection and 0.8406 (95% CI: 0.7594-0.9218) for mixed versus viral infection. Longitudinal analyses showed that CD169% declined during recovery from viral infection, whereas the nCD64 index and CRP decreased after treatment in bacterial infection; all three markers declined in mixed infection. Conclusions: CD169%, the nCD64 index, and CRP reflect complementary dimensions of the host response to acute infection. Their combined assessment showed good performance for differentiating viral from bacterial infection and provided clinically relevant information in mixed infection. This host-response-based nomogram may serve as a practical adjunct for early etiological stratification and mixed-infection risk assessment.