不同宿主背景下肺NTM感染CT主导影像模式及影像归因分析

CT-Dominant Imaging Patterns and Imaging Attribution Analysis of Pulmonary Nontuberculous Mycobacterial Infection across Different Host Backgrounds

摘要: 目的：探讨不同宿主背景下肺非结核分枝杆菌感染的CT主导影像模式特点及影像归因分析价值。方法：回顾性纳入2022年1月至2026年4月安徽医科大学第二附属医院住院NTM相关病例。按宿主背景分为普通宿主、结构性肺病和免疫受损组。由2名胸部影像医师判定胸部CT主导影像模式，并对复杂背景或归因困难病例进行影像归因分析。结果：共纳入64例患者，其中普通宿主31例、结构性肺病17例、免疫受损16例。CT主导影像模式以气道中心型(39例)最常见，其次为肿块实变型(12例)和纤维空洞型(9例)；不确定型4例。排除不确定型后，不同宿主背景间CT主导影像模式分布差异无统计学意义(P = 0.500)。影像归因分析识别出4例不能可靠纳入预设CT主导影像模式分析的病例，包括淋巴结受累主导型、磨玻璃影主导型及复杂背景相关病例。结论：肺NTM感染CT主导影像模式以气道中心型最常见，本研究未能在不同宿主背景间发现具有统计学意义的CT主导影像模式分布差异，相关结果仍需更大规模研究进一步验证。部分复杂背景病例虽具有NTM相关病原学证据，但主要肺部异常不能直接归入既定CT主导影像模式。影像归因分析有助于识别适合进入CT主导影像模式分析的病例，并提高对真实世界复杂病例的理解。

Abstract: Objective: To investigate CT-dominant patterns of pulmonary nontuberculous mycobacteria (NTM) infection across different host backgrounds and to evaluate the role of imaging attribution analysis in complex real-world cases. Methods: This retrospective study included hospitalized patients with NTM-related disease at the Second Affiliated Hospital of Anhui Medical University between January 2022 and April 2026. Patients were categorized according to host background as general host, structural lung disease, or immunocompromised host. Two thoracic radiologists assessed the CT-dominant pattern on chest CT, including airway-centered pattern, fibrocavitary pattern, mass/consolidation pattern, and indeterminate category. Imaging attribution analysis was performed for cases with complex clinical backgrounds or uncertain attribution of imaging findings. Results: A total of 64 patients were included: 31 general hosts, 17 patients with structural lung disease, and 16 immunocompromised hosts. The most common CT-dominant pattern was the airway-centered pattern (n = 39), followed by the mass/consolidation pattern (n = 12) and fibrocavitary pattern (n = 9); 4 cases were classified as the indeterminate category. After exclusion of indeterminate cases, the distribution of CT-dominant patterns did not differ significantly among host-background groups (P = 0.500). Imaging attribution analysis identified 4 cases that could not be reliably incorporated into the predefined CT-dominant pattern analysis, including cases dominated by lymph node involvement, ground-glass opacity, or complex background-related abnormalities. Conclusion: In pulmonary NTM infection, the airway-centered pattern was the most frequent CT-dominant pattern. This study failed to identify statistically significant differences in the distribution of CT-dominant imaging patterns among different host backgrounds, and these findings require further validation in larger-scale studies. Although some complex cases had microbiologic evidence related to NTM, their predominant pulmonary abnormalities could not be directly assigned to the predefined CT-dominant patterns. Imaging attribution analysis may help identify cases suitable for CT-dominant pattern analysis and improve interpretation of complex real-world presentations.