Bronchiolitis obliterans syndrome ‘endotypes’ in haematopoietic stem cell transplantation

The recognition of disease presentations with similar phenotypic characteristics and their grouping into diagnostic categories has underpinned medical practice for over 100 years. This cataloguing of diseases has, in turn, led to much improved understanding of disease pathogenesis, as well as effective and more precise or even ‘personalized’ treatments, and continues to the present day. Perhaps the most pertinent contemporary example of this iterative process in respiratory medicine has been the clustering of asthma phenotypes, which point to distinct underlying pathophysiological mechanisms—referred to as ‘endotypes’.

In allogeneic transplantation, one of the most serious respiratory complications is the development of irreversible small airway narrowing due to the deposition of scar tissue—referred to as bronchiolitis obliterans syndrome (BOS). BOS affects recipients of lung allografts but is less common in individuals who receive multi-organ transplants (heart–lung, heart–lung–liver or lung–liver), with the volume of tissue transplanted correlating roughly inversely with BOS risk—an observation that remains unexplained.1 BOS affects approximately 10% of lung transplant recipients each year, and hence up to 50% by 5 years post-transplant, and is the main cause of mortality after the first post-transplant year.1 As haematopoietic stem cell transplantation is akin to having multiple solid organs transplanted, it is not surprising that BOS occurs, and it is perhaps also not surprising that the incidence is much lower, with an overall prevalence of <10% in most cohorts, higher in those with a history of chronic graft versus host disease.2

Over the past decade, distinct BOS phenotypes have been identified amongst lung transplant recipients, leading to a change in nomenclature and some changes in treatment approach. A predominantly restrictive syndrome, called ‘restrictive allograft syndrome (RAS), is characterized by largely upper zone fibrosis with no or little airway obstruction and a poor prognosis.3 At the same time, a potentially reversible BOS subgroup characterized by neutrophilic airway inflammation and a treatment response to azithromycin was identified.4 Together, these various manifestations of chronic graft injury have been termed ‘chronic lung allograft dysfunction’ (CLAD), with RAS accounting for up to 15% of CLAD but with BOS remaining the predominant CLAD subtype. The hope is that more accurate categorization will lead to better appreciation of differences in underlying pathogenesis, more effective treatment and hence classification of true endotypes. Analogous endotypes amongst haematopoietic stem cell transplant recipients are less well described, and perhaps as a consequence, management remains largely empirical. For example, there is currently no convincing evidence for the efficacy of azithromycin in the haematopoietic stem cell transplant BOS group,5 but the studies that have been carried out have not attempted to enrich for subgroups of patients potentially more likely to respond.

In this issue of the journal, Kwok et al.6 carefully interrogate a large and well-nurtured cohort of haematopoietic stem cell transplant recipients searching for distinct BOS phenotypes. Amongst the 1461 subjects in a cohort spanning almost 20 years, 95 (6.5%) developed BOS. Interestingly, the BOS group experienced fewer disease relapses than the non-BOS group. Amongst the group of subjects with BOS, the authors recognized that those subjects with a more rapid initial decline (>25% in the first 3 months) in forced expiratory volume in 1 s (FEV₁) appeared to represent a distinct subgroup who would suffer persistently inferior lung function at subsequent time points and inferior overall survival compared to the ‘slow decliners’.6 Females comprised 63% of the rapid decliner group (P = 0.014), but this was one of very few baseline differences between the groups, with trends only for pre-transplant FEV₁ and forced vital capacity (FVC) to be lower and for more patients with acute myeloid leukaemia and matched-unrelated grafts to suffer rapidly declining BOS.6 Unlike lung transplantation where the restrictive (RAS) subtype is associated with worse survival, the rapid declining group in Kwok et al.’s paper did not appear to have more restrictive physiology, and chest imaging findings were not reported.6 In contrast, in another recent observational cohort study from the United States, a restrictive phenotype was the strongest risk factor for subsequent mortality, in line with observations in lung transplantation.7

Although the limitations inherent to the study design adopted by Kwok et al. are well known and appreciated, the size and completeness of follow-up of the cohort are particular strengths. Recurrent problems in studies of BOS after haematopoietic stem cell transplantation have been the inclusion of only small numbers of study subjects, the lateness of the BOS diagnosis and the lack of uniform lung function follow-up.7 The granularity of the data ascertained by Kwok et al. is one of the significant strengths of the study and may explain why they were able to discern subgroups that were not apparent in other studies.7 Whilst it would be expected to an extent that subjects with any diagnosis who do worse initially will continue to do worse, and while, of course, their findings will need to be verified in other cohorts, they do point to the possibility of identifying disease subgroups with distinct clinical features amongst subjects suffering from BOS after haematopoietic stem cell transplantation. Compared to lung transplantation, the relative rarity of BOS; the lack of uniform recommendations for lung function follow-up; and the complexity of indications, induction regimens and transplant types inherent to the practice of haematopoietic stem cell transplantation add heterogeneity, which may obscure endotypes. Previous authors have identified bacterial colonization of the lower respiratory tract,8 pre-transplant airflow limitation2, 9, 10 and shorter duration of time from transplantation to diagnosis of BOS10 as poor prognostic factors; however, true endotypes remain elusive. Nevertheless, the experience in lung transplantation suggests that attempting to cluster subtypes of BOS together is a worthwhile endeavour.

As is now the case in asthma, more accurate classification of disease subtypes can provide clarity where previously there was confusion and lead to genuinely transformational improvements in management. At present, as it always has done, much of this pattern recognition and clustering relies on the astute eye of an experienced, inquiring clinician. However, in the age of big data, machine learning and increasing digitization of health care, the process will become more automated, sophisticated and timely. These developments, along with a commitment to consistently improving outcomes through multicentre collaboration, will be particularly helpful in making precision medicine a reality for patients with rare diseases such as BOS.