The long term effects of COVID-19 on Pulmonary diffusing capacity for nitric oxide (DLNO) and carbon monoxide (DLCO)

Hypothesis: This study hypothesized that combining lung diffusing capacities for nitric oxide (DLNO) and carbon monoxide (DLCO) enhances detection of COVID-19-related lung pathologies over using DLCO-alone, or measures of spirometry and total lung capacity (TLC) or single measures alone. A combined DLNO+DLCO z-score was expected to better classify post-COVID lung function abnormalities, capturing a broader spectrum of pulmonary impairment.

Data Overview: The dataset includes 256 COVID-19 survivors & 76 controls (332 total) who completed all pulmonary function tests (PFT) across six centers in Italy, France, Spain, and Australia. About 97% of data from these centers is published in 6 peer-reviewed journals & 2 conference abstracts. Controls and COVID-19 cases were tested with the same equipment and protocols. The complete battery of PFTs included DLNO and DLCO (average breath-hold 5.4 s), spirometry, and total lung capacity (TLC), with z-scores calculated from reference equations. COVID-19 survivors were tested between 17 and 575 days post-infection (median 143 days). An additional 190 subjects completed all PFT measurements except TLC. Mixed model binary logistic regression was used with “Study” as a random intercept, COVID-19 (1) vs. control (0) as the dependent variable, and z-scores for spirometry, TLC, and diffusion capacity as independent variables. Matthews correlation coefficient (MCC) assessed model classification with a threshold of 0.5.

Notable Findings: Among 256 subjects with complete PFTs post-COVID-19, aged 18 to 89 (median 60 years), 56 (22%) of survivors had restrictive lung issues (TLC < lower limit of normal (LLN), defined by z-score < -1.645). Additionally, 57% showed impairment (airway obstruction, restriction, mixed disorder, or DLNO or DLCO < LLN). Of 34 models evaluated, the model with the lowest Bayesian Information Criterion (BIC) was a combined DLNO+DLCO z-score model, demonstrating superior COVID-19 detection. Five of the top nine models were combined DLNO+DLCO z-score models, while two were DLNO-only and two were DLCO-only z-score models. MCC values indicated 6 of the top 9 models were either combined DLNO+DLCO z-scores or DLNO-only models for highest classification accuracy, with MCCs of approximately 0.50 to 0.53. Furthermore. Dyspnea severity correlated with combined z-scores, DLNO5s-only and DLCO5s-only z-scores (p < 0.001).

Data Interpretation and Implications: This study highlights that DLNO, combined with DLCO, provides insights into COVID-19-altered lung diffusion, especially within the alveolar-capillary membrane. Combined z-scores enhance post-COVID abnormality classification, supporting tailored follow-up for symptomatic survivors. Adoption of the DLNO-DLCO approach would require further regulatory approval and standardization. This combined technique is suggested as a promising tool for detecting subclinical COVID-19-related lung impairments, enhancing sensitivity for long-term lung monitoring.

假设：本研究假设结合肺一氧化氮（DLNO）和一氧化碳（DLCO）的扩散容量可以增强对COVID-19相关肺部病理的诊断，优于单独使用DLCO，或仅测量肺活量（TLC）和肺总量（TLC），或单独的测量。预期的DLNO+DLCO联合z分数能够更好地分类COVID-19后的肺功能异常，捕捉更广泛的肺损伤谱。

数据概述：该数据集包括256名COVID-19幸存者和76名对照者（总计332人），他们在意大利、法国、西班牙和澳大利亚的六个中心完成了所有肺功能测试（PFT）。其中大约97%的数据已发表在6本同行评审期刊和2篇会议摘要中。对照者和COVID-19病例使用相同的设备和协议进行测试。完整的PFT测试包括DLNO和DLCO（平均屏气时间为5.4秒）、肺活量计和肺总量（TLC），并从参考方程中计算出z分数。COVID-19幸存者在感染后17至575天内接受测试（中位数为143天）。另外190名受试者完成了所有PFT测量，但未进行TLC测量。使用具有“研究”作为随机截距的混合模型二元逻辑回归，COVID-19（1）与对照（0）作为因变量，肺活量、TLC和扩散能力的z分数作为自变量。马修斯相关系数（MCC）用于评估模型的分类，阈值为0.5。

显著发现：在256名完成PFT的COVID-19后受试者中，年龄在18至89岁之间（中位数为60岁），其中56名（22%）幸存者存在限制性肺问题（TLC低于正常值下限（LLN），定义为z分数低于-1.645）。此外，57%的受试者表现出肺功能损害（气道阻塞、限制、混合障碍或DLNO或DLCO低于LLN）。在34个评估的模型中，具有最低贝叶斯信息准则（BIC）的模型是结合DLNO+DLCO z分数模型，显示出优越的COVID-19检测能力。前九个模型中有五个是结合DLNO+DLCO z分数模型，两个是DLNO-only模型，两个是DLCO-only z分数模型。MCC值表明，前九个模型中有六个是结合DLNO+DLCO z分数或DLNO-only模型，具有最高的分类准确性，MCC值约为0.50至0.53。此外，呼吸困难程度与结合z分数、DLNO5s-only和DLCO5s-only z分数相关（p<0.001）。

数据解释和影响：本研究强调了DLNO与DLCO的结合，可以揭示COVID-19改变后的肺扩散，特别是在肺泡-毛细血管膜中。结合z分数增强了COVID-19后异常分类，支持对症状性幸存者的定制随访。采用DLNO-DLCO方法需要进一步的监管批准和标准化。这种联合技术被建议为检测亚临床COVID-19相关肺损伤的有前途的工具，提高了长期肺监测的敏感性。