Time-dependent changes in the risk of death in pure bronchiolitis obliterans syndrome (BOS)

Background: The timing of disease onset may affect the prognosis in chronic lung allograft dysfunction (CLAD). The relationship between the timing of disease onset and the prognosis of CLAD and its sub-types, bronchiolitis obliterans syndrome (BOS) and restrictive allograft syndrome (RAS), was examined. Methods: Clinical records and pulmonary function data of 597 patients who underwent bilateral lung transplantation from 1996 to 2010 and survived for >3 months were examined. Results: Among 155 patients with a final diagnosis of BOS, patient survival after disease onset was significantly different according to disease-onset timing (BOS onset/post-BOS median survival: overall/1,438 days; <1 year/511 days; 1-2 years/1,199 days; 2-3 years/1,403 days; >3 years/did not reach median survival; p < 0.0001). The prognosis of RAS was generally poorer than that of BOS (overall post-RAS median survival, 377 days). Treating non-CLAD, CLAD, BOS, and RAS as time-dependent covariates, recipient sex-adjusted and age-adjusted Cox regression analysis demonstrated an overall mortality risk of BOS (reference: no CLAD) of 6.7 (95% confidence interval, 4.6-9.9). However, when patients survived 3 years without CLAD, the mortality risk of subsequent BOS was only 1.9 (95% confidence interval, 0.8-4.4) compared with no CLAD. The number of RAS patients was too small to obtain sufficient power to estimate time-dependent mortality risk. Conclusion: Late-onset BOS showed a better prognosis than early-onset BOS. Studies that do not distinguish BOS from RAS may overestimate the mortality risk of BOS. Multicenter studies will be required to further elucidate risk factors toward the development of better management strategies for CLAD.