Phenotyping CLAD after Single Lung Transplant: Limits and Prognostic Assessment of the 2019 ISHLT Classification System

Purpose Phenotyping chronic lung allograft dysfunction (CLAD) in single lung transplant (SLTX) recipients is challenging, due to the native lung contribution to pulmonary function tests (PFT). We aimed to assess the ISHLT CLAD classification and its prognostic performance in SLTX. Methods In this retrospective cohort of adult, first, SLTX from 2009 to 2017, patients with persistent drop in FEV1>20% were assessed by 2 independent reviewers to determine CLAD status and phenotype, specifically noting presence or absence of RAS-like opacities. Disagreements were resolved by a 3rd reviewer. Interobserver agreement (IOA) was calculated by Cohen's Kappa. Association of CLAD phenotypes with time to death/retransplant (ReTx), adjusted for age, sex, CMV mismatch and native lung condition, were assessed with Cox models. Results Out of 172 SLTX patients, 92 had a persistent drop in FEV1>20%, of whom 67 got a diagnosis of CLAD. We noted a moderate IOA for CLAD status (Kappa 0.69) and poor IOA for phenotype (Kappa 0.5). When applying the ISHLT criteria strictly (based on exact cut-offs for PFT, along with imaging), 34 patients had BOS (50.7%), 9 RAS/mixed (13.4%), 7 undefined (10.4%), and 17 unclassified (25.5%). We found no association of these strict phenotypes with death/ReTx (Fig A). When using adjudicated CLAD phenotypes, 31 patients had BOS (46.3%), 15 RAS/mixed (22.4%), 2 Undefined (3%), and 19 Unclassified (28.3%). Using these adjudicated phenotypes, RAS/mixed was significantly associated with higher risk of death/ReTx (HR 2.98, 95%CI [1.39-6.4]) (Fig B). Finally, the specific adjudication of RAS-like opacities had the best IOA (Kappa 0.73). Presence of RAS-like opacities was a strong predictor of death/ReTx (HR 2.3, 95%CI [1.2-4.5]) (Fig C). Conclusion PFT interpretation is challenging in SLTX. Using a classification relying more on imaging analysis that harboured good IOA, we obtained better prognostic performances than with a strict application of published physiologic cut-offs.