术前全身炎症反应指数与慢性鼻窦炎伴鼻息肉患者内镜鼻窦手术后12个月内镜复发：同一机构两院区队列中的关联及增量预测效能

文献信息

中文摘要

背景： 全身炎症反应指数（SIRI）是一种由常规血细胞计数分析得出的易获取标志物，已被认为与慢性鼻窦炎伴鼻息肉的复发相关。其是否能在常规临床预测因素之外提供额外的预后信息仍不确定。
方法： 这项回顾性、单机构、两院区研究纳入805名年龄至少18岁（无年龄上限）并接受内镜鼻窦手术（ESS）的参与者。主院区开发队列包括445名参与者，黄山路院区留出队列包括360名参与者；两个队列中分别有400名和320名参与者具有可评估的12个月结局。阿司匹林加重性呼吸系统疾病（AERD）与对非甾体抗炎药（NSAIDs）的一般超敏反应分别进行临床判定。使用区分度、校准、Brier评分以及配对bootstrap估计的增量效能比较临床模型与扩展logistic模型。敏感性分析使用稳定化逆概率加权、200次结局插补和两种极端结局赋值，并排除无AERD的NSAID超敏参与者。
结果： 开发队列中124/400名参与者发生复发，留出队列中99/320名参与者发生复发。校正后，SIRI每翻倍与疾病复发相关[比值比1.47；95%置信区间（CI）1.06-2.04；P=0.022]。在开发队列评估中，加入SIRI使受试者工作特征曲线下面积变化0.007（bootstrap 95%CI -0.002至0.026），在留出队列评估中变化0.010（-0.016至0.030）；Brier评分的相应改善分别为0.0036（0.00006-0.0108）和0.0044（-0.0025至0.0111）。85名参与者缺乏可评估结局。SIRI比值比在加权时为1.46，在插补时为1.47，在AERD敏感性排除后为1.46。假设所有缺失结局均为复发，则估计值衰减至1.31（0.99-1.74）。
结论： 术前SIRI与ESS后12个月内镜复发相关。该关联在逆概率加权、多重插补和AERD限制分析中大体保持，但当所有缺失结局被赋值为复发时减弱。仅含SIRI的模型作为简单易得的全身炎症标志物显示出中等区分度，但其在常规临床预测因素之外的增量预测增益有限且不确定。因此，这些发现支持将SIRI作为候选预后标志物进行前瞻性外部验证，而非用于立即的治疗或监测决策。

英文摘要

BACKGROUND: The systemic inflammation response index (SIRI) is a readily obtainable marker derived from routine blood count analyses and has been associated with the recurrence of chronic rhinosinusitis with nasal polyps. Whether it provides added prognostic information beyond routine clinical predictors remains uncertain.
METHODS: This retrospective, single-institution, two-campus study included 805 participants who were at least 18 years of age (with no upper age limit) and who underwent a endoscopic sinus surgery (ESS). The Main Campus development cohort included 445 participants, and the Huangshan Road Campus held-out cohort included 360 participants; 400 and 320 participants from both cohorts, respectively, had evaluable 12-month outcomes. Aspirin-exacerbated respiratory disease (AERD) was clinically ascertained separately from general hypersensitivity to non-steroidal anti-inflammatory drugs (NSAIDs). Clinical and extended logistic models were compared using discrimination, calibration, Brier's score, and paired bootstrap estimates of incremental performance. Sensitivity analyses used stabilized inverse-probability weighting, 200 outcome imputations, and two extreme-outcome assignments and excluded NSAID-hypersensitive participants without AERD.
RESULTS: Recurrence occurred in 124/400 participants from the development cohort and in 99/320 participants from the held-out cohort. Each doubling of SIRI had been associated with the disease recurrence after adjustment [odds ratio, 1.47; 95% confidence interval (CI), 1.06-2.04; P = 0.022]. Adding SIRI changed the area under the receiver operating characteristic curve by 0.007 in the evaluation of the development cohort (bootstrap 95% CI, -0.002 to 0.026) and by 0.010 in the evaluation of the held-out cohort (-0.016 to 0.030); corresponding improvements in Brier's score were 0.0036 (0.00006-0.0108) and 0.0044 (-0.0025 to 0.0111). Eighty-five participants lacked an evaluable outcome. The SIRI odds ratios were 1.46 with weighting, 1.47 with imputation, and 1.46 after the AERD sensitivity exclusion. Assuming that every missing outcome was recurrence attenuated the estimate to 1.31 (0.99-1.74).
CONCLUSIONS: The preoperative SIRI was associated with 12-month endoscopic recurrence after ESS. This association was broadly preserved in inverse probability-weighted, multiple-imputation, and AERD-restricted analyses, although it was attenuated when all missing outcomes were assigned to recurrence. The SIRI-only model showed moderate discrimination as a simple and readily available marker of systemic inflammation, but its incremental predictive gain beyond routine clinical predictors was modest and uncertain. Therefore, these findings support SIRI as a candidate prognostic marker for prospective external validation rather than as a tool for immediate treatment or surveillance decisions.