Background: Diffusion capacity of the lungs for carbon monoxide (DLCO) represents a biomarker of oxygen uptake across the alveolar capillary membrane, which is known to be deteriorated in acute heart failure (HF). It is not known whether DLCO has a predictive value for patients with chronic HF.

Methods: Data from the MyoVasc study (NCT04064450), a prospective cohort study on HF, were analyzed. During a highly standardized 5-hour examination in a dedicated study center participants underwent pulmonary function testing via body plethysmography (MasterScreen Body CareFusion, Germany). DLCO was determined via single breath method using carbon monoxide as test and Helium as tracer gas. According to current recommendations DLCO >80 % was defined as normal, 60-80 % as mildly reduced 40-<60 % as moderately reduced and <40 % as severely reduced. Information on clinical outcome was derived from structured follow-up with subsequent validation via source data and independent adjudication of clinical endpoints. The primary study endpoint was worsening of HF defined as a composite of HF hospitalization and cardiac death.

Results: The analysis sample comprised 1,311 subjects (mean age 67.1±10.1 years; female sex: 26.7%; median concentration of NT-proBNP 345.5pg/ml) with HF and available bodyplethysmographic data on DLCO. A total of N=242 individuals had a normal DLCO, whereas N=555 had a mildly reduced, N=408 a moderately reduced, and N=106 a severely reduced DLCO. During a median follow up period of 4.01 years, 238 participants experienced the primary study endpoint worsening of HF. The estimated cumulative incidence of worsening of HF ranged from 18.6% for participants with normal DLCO to 43.7% in subjects with severely reduced DLCO (P for trend <0.001). This relationship was consistent across the subcomponents HF-hospitalization (normal DLCO 12.4%, severely reduced DLCO 27.7%, P for trend <0.001) and cardiac death (normal DLCO 3.3%, severely reduced DLCO 23.5%, P for trend <0.001). As the cumulative incidence rates of worsening of HF, HF-hospitalization, and cardiac death did not differ between mildly reduced and normal diffusion capacity, DLCO <60 % was subsequently investigated as predictor of clinical outcome. In multivariable Cox competing risk analysis with non-cardiac death as competing risk adjusting for age, sex and height, worse DLCO (i.e. <60 %) was related to an approximately doubled risk for worsening of HF: hazard ratio (HR) 1.98, [95 % confidence interval 1.54 to 2.56]; P <0.001. Sensitivity analysis demonstrated that DLCO was a stronger predictor of cardiac death (HR 3.89 [2.49 to 6.06]; P<0.001) than for HF hospitalization (HR 1.96 [1.47 to 2.61], P<0.001). After additional adjustment for an established HF risk score (Meta-Analysis Group in Chronic Heart Failure (MAGGIC) Risk Score), DLCO was confirmed as strong predictor for worsening of HF (HR 1.60 [1.23 to 2.07]; P<0.001). Consistently, DLCO also remained an independent predictor for cardiac death (HR 2.86 [1.82 to 4.41]; P<0.001) and HF hospitalization (HR 1.60 [1.20 to 2.21], P <0.001) in Cox regression analyses with adjustment for age, sex, and the MAGGIC risk score.

Conclusion: DLCO represents a strong predictor of HF-specific outcome beyond established risk prediction models. This underlines the relevance of DLCO as markers of pulmonary structural remodeling for the clinical course of HF with potential implication for risk stratification and intervention.