Clinical features and predictive model for invasive pulmonary fungal infections in hematologic disorders

Abstract

Objective

This study was to investigate the clinical features of hematological disorders complicated by invasive pulmonary fungal infections and identify factors affecting treatment outcomes, with the aim of developing a predictive model.

Methods

Clinical data were collected from patients with hematological disorders and invasive pulmonary fungal infections between January 2020 and June 2023. Based on metagenomics next generation sequencing (mNGS) of bronchoalveolar lavage fluid (BALF), patients were categorized into three groups: Candida, Mucor, and Aspergillus. General conditions, clinical features, treatments, and outcomes were compared. Treatment outcomes were assessed two months after therapy and classified as either improved or not improved. Factors influencing outcomes were analyzed, and a risk prediction model for treatment failure was developed.

Results

A total of 89 patients with hematological diseases and invasive pulmonary fungal infections were included: 26 with Candida, 25 with Mucor, and 38 with Aspergillus. Significant differences were observed between groups in long-term corticosteroid use, hematological disease outcomes, neutropenia duration, treatment duration, central venous catheter placement, galactomannan (GM) test results, CD4⁺ T-cell count, and clinical manifestations. After two months of antifungal therapy, improvement rates were 96.15 % for Candida, 76.00 % for Mucor, and 63.16 % for Aspergillus. Logistic regression analysis identified elevated platelet count (OR = 0.9823, 95%CI: 0.9663–0.9945), D-dimer (OR = 1.2130, 95%CI: 1.0544–1.4934), C-reactive protein (OR = 1.0066, 95%CI: 1.0026–1.0111) and medication adjustments based on mNGS results (OR = 0.0495, 95%CI: 0.0108–0.1624) as significant prognostic factors. A nomogram prediction model based on these factors demonstrated good discrimination with a C-index of 0.86.

Conclusion

The clinical features and treatment outcomes differ among fungal types in patients with hematological disorders and invasive pulmonary fungal infections. The nomogram prediction model, incorporating platelet count, D-dimer, C-reactive protein and mNGS-guided therapy adjustments, offers robust predictive performance for two-month treatment outcomes.