Phenotyping Established Chronic Lung Allograft Dysfunction Predicts Extracorporeal Photopheresis Response in Lung Transplant Patients


Mark Greer


Chronic lung allograft dysfunction (CLAD) remains the leading cause of mortality in lung transplant recipients after the first year. Treatment remains limited and unpredictable. Existing data suggests extracorporeal photopheresis (ECP) may be beneficial. This study aimed to identify factors predicting treatment response and the prognostic implications. A single center retrospective analysis of all patients commencing ECP for CLAD between November 1, 2007 and September 1, 2011 was performed. In total 65 patients were included, 64 of whom had deteriorated under azithromycin. Median follow-up after commencing ECP was 503 days. Upon commencing ECP, all patients were classified using proposed criteria for emerging clinical phenotypes, including “restrictive allograft syndrome (RAS)”, “neutrophilic CLAD (nCLAD)” and “rapid decliners”. At follow-up, 8 patients demonstrated ≥10% improvement in FEV1, 27 patients had stabilized and 30 patients exhibited ≥10% decline in FEV1. Patients fulfilling criteria for “rapid decliners” (n = 21, p = 0.005), RAS (n = 22, p = 0.002) and those not exhibiting neutrophilia in bronchoalveolar lavage (n = 44, p = 0.01) exhibited poorer outcomes. ECP appears an effective second line treatment in CLAD patients progressing under azithromycin. ECP responders demonstrated improved progression-free survival (median 401 vs. 133 days). Proposed CLAD phenotypes require refinement, but appear to predict the likelihood of ECP response.


bronchoalveolar lavage


bronchiolitis obliterans


bronchiolitis obliterans syndrome


chronic lung allograft dysfunction


double lung transplant


extracorporeal photopheresis


forced expiratory volume in one second


heart lung transplantation


lung transplantation


restrictive allograft syndrome


single lung transplant


total lung capacity


Lung transplantation (LTx) represents an accepted treatment option for selected patients with end-stage lung disease. In the past 20 years, there has been a steady increase in the number of transplantations being performed [1]. Long-term survival however remains disappointing, with overall median survival in adults of 5.5 years [1]. Beyond the first year, chronic graft dysfunction represents the leading cause of death [1]. Bronchiolitis obliterans syndrome (BOS) represents the agreed term for grading such losses in graft function [2]. Although derived from bronchiolitis obliterans (BO) classically seen in graft dysfunction, BOS is a spirometric diagnosis, reflecting deteriorating FEV1 from presumed small airway obliteration [2, 4-7]. Chronic lung allograft dysfunction (CLAD) represents evolving terminology, acknowledging that factors other than BO may account for graft loss [5]. Numerous studies have illustrated the highly variable clinical course in CLAD, ranging from gradual decline over years to sudden-onset, rapidly deteriorating graft function within weeks [8-10]. Given this heterogeneity, recent studies have attempted to define specific CLAD phenotypes based on various parameters [11-14], leading to the emergence of the restrictive allograft syndrome (RAS) and neutrophilic CLAD [2, 3, 15].

Various treatments have been suggested, with all demonstrating unpredictable outcomes and generally targeting minimizing further graft dysfunction. Macrolides remain the mainstay of medical therapy, with azithromycin having previously demonstrated a 30% response rate [12]. A small randomized placebo-controlled study has even demonstrated preventive effects [16]. Studies examining adjuvant benefit of montelukast show promise, but again impede disease progression rather than arresting or reversing CLAD [17].

Although exact etiology of CLAD remains unclear, numerous histopathological changes resulting from immune and inflammatory insults to airway epithelium and subcellular matrix are suspected. Given the rarity of similar immunological processes in nontransplant populations, reactivation of alloimmune Tcells is considered contributory [18]. Animal studies support an initial TH-1 alloimmune response, augmented by Interferon-γ, leading to lymphocytic infiltration of respiratory epithelial cells which then produce a pro-fibrotic milieu [19-21]. Treatments targeting recipient Tcell clonal populations have generated significant interest in managing CLAD [22]. Extracorporeal photopheresis (ECP) has emerged as one option, having been successfully used in cutaneous Tcell lymphoma and graft-versus-host disease [23]. The first successful application of ECP in lung transplant recipients was reported in 1995 [24]. Studies validating its efficacy in treating CLAD have shown promise, but been limited due to size, noncomparable adjuvant immunosuppression and short follow-up [22, 25-28]. Consequently, ECP is recommended in most centers as rescue treatment in CLAD [29, 30].

Given limited data regarding efficacy of ECP in CLAD and gaining acceptance of separate phenotypes, we investigated if implementation of CLAD phenotypes assisted in predicting ECP response.

Materials and Methods

Study design

Retrospective analysis of all lung transplant recipients commencing ECP for progressive CLAD between November 2007 and September 2011 at our institution was performed. We analyzed baseline characteristics, colonization, CLAD phenotypes, azithromycin response and ECP initiation. Our local institutional review board approved analysis. Part of the study was registered in (identifier: NCT00502554)

Spirometry was performed according to ATS/ERS guidelines [31] at each outpatient attendance. FEV1 (Forced Expired Volume in 1 Second), FVC (Forced vital capacity) and TLC (Total lung capacity) values as well as broncho-alveolar lavage (BAL) cell counts at CLAD diagnosis, azithromycin initiation and ECP initiation were tabulated. Best FEV1 was calculated by averaging the two best values obtained greater than 3 weeks apart following transplantation. Corresponding FVC and TLC values at the time of best FEV1 were adopted as optimum baseline values. Significant BAL neutrophilia was accepted to be ≥15%, once bacterial and viral infections (IFT and PCR) had been excluded. Sudan staining >5% was considered evidence of aspiration. Transbronchial biopsies, performed as part of our routine surveillance program were reviewed for evidence of lymphocytic bronchitis, classified in accordance with ISHLT guidelines.

Phenotype definition

All patients were diagnosed and graded for graft dysfunction in accordance to International Society for Heart and Lung Transplantation (ISHLT) BOS criteria [5]. Given acceptance of the term “BOS Stage” in describing the extent of graft dysfunction, this has been retained in the interests of clarity. In acknowledging the emerging discordance in underlying etiology however, CLAD has been preferred for the purpose of descriptive classification. Phenotypes were defined at the time of ECP initiation, based on changes in TLC, BAL cellularity and average rate of decline in lung function between CLAD diagnosis and ECP initiation. Patients exhibiting a TLC ≤90% of baseline were labeled as “restrictive allograft syndrome” (RAS). Patients demonstrating BAL neutrophilia ≥15% were considered “neutrophilic CLAD”. Patients suffering a >100 mL/month decline in FEV1 before ECP initiation were classified as “rapid-decliner” (RD). These individual traits were not considered to be mutually exclusive, with a degree of overlap anticipated. Initial analysis was performed within the phenotypes as defined, with subsequent separate analysis of “overlap” patients.

Treatment algorithm in CLAD

Following exclusion of reversible causes of deteriorating graft function, 1st line treatment with azithromycin had been initiated according to accepted regimes [12]. Azithromycin response was defined as ≥10% increase in FEV1 at any stage post-initiation. ECP was initiated following subsequent ≥10% decline from baseline FEV1 under azithromycin. The ECP protocol was developed in collaboration with our Hematology department and although treatment was occasionally performed at neighboring hospitals, our standardized protocol was used in all patients. ECP was performed via a closed-loop intravenous system, with approximately 700 mL of whole blood being removed [22]. Leucocytes were then isolated by centrifugation, incubated with 8-methoxypsoralen (8-MOP) and subsequently exposed to ultraviolet-A (UVA) light (350 nm). This induced bonding between 8-MOP and pyrimidine bases in cellular DNA, resulting in lymphocyte apoptosis [29, 32]. ECP was initially performed fortnightly for 3 months. Thereafter, treatment intervals were lengthened subject to stabilized graft function. The maximum interval between ECP treatments was 8 weeks. Patients who did not complete the initial 3-month induction treatment were excluded. Treatment response was considered a ≥10% improvement in FEV1 compared to the value at the time of commencing treatment. A decline of ≥10% following ECP was termed progressive disease. Progressive patients continued to receive ECP at their steady-state interval unless clinical condition prevented continuation.


Data are reported as medians with interquartile ranges (IQR). Reported p-values are two-sided, unless otherwise indicated. For all analyses, p-values <0.05 were considered statistically significant. Categorical variables were analyzed using chi-squared test or Fisher's Exact test. Medians were compared with the Mann–Whitney test and the nonparametric Kruskal–Wallis-H test. Using Kaplan–Meier analysis, progression-free and nonresponder-free survival were calculated. Subsequent multivariate analysis with Cox regression was performed. All variables with a p value <0.10 were included, and variables with a p-value >0.10 were excluded in multivariate analysis.


Sixty-five patients (27 female, 42%) commencing ECP for established CLAD between November 01, 2007 and September 01, 2011 were included. Fifty-one (78%) patients had undergone double-lung transplantation with a median age at transplantation of 43 years (IQR 34–53). Remaining baseline characteristics are summarized in Table 1. Median time to CLAD diagnosis was 1114 days (IQR 601–1,994). Median begin of ECP followed 292 days (IQR 95–741) after CLAD onset; 63/65 patients started >1 year after CLAD onset with ECP. Median follow-up after ECP initiation was 503 (IQR 404–741) days (Table 2).

Table 1. Baseline characteristics (n = 65)
  1. aOnly patients participating in maintenance follow-up were included (n = 44).
  2. BOS = bronchiolitis obliterans syndrome; CLAD = chronic lung allograft syndrome; DLTx = double lung transplantation; ECP = extracorporeal photophoresis; f/u = follow up; HLTx = heart lung transplantation; SLTx = single lung transplantation; Tx = transplantation.
AgeYears53 (42–59)
Gendern° (%) 
Female 27 (42)
Male 38 (58)
Underlying diseasen° (%) 
Emphysema 25 (38)
Cystic fibrosis 16 (25)
Idiopathic fibrosis 11 (17)
Pulmonary hypertension 3 (5)
Other 10 (15)
Transplantation proceduren° (%) 
HLTx 5 (8)
DLTx 51 (78)
SLTx 9 (14)
Immunosuppression at inclusionn° (%) 
Cyclosporin 25 (39)
Tacrolimus 40 (62)
Mycophenolat 53 (82)
Prednisolone 59 (91)
Evelorimus 4 (6)
Azathioprin 7 (11)
Azithromycinn° (%)64 (98)
Time on azithromycin before ECP, median, IQRDays221 (109–506)
Azithromycin respondern° (%)9 (14)
FEV 1 baseline, median, IQR% best68 (57–77)
FEV 1 baseline, median, IQRml1790 (1360–2375)
BOS Stagen° (%) 
Stage 0p 3 (5)
Stage 1 6 (9)
Stage 2 21 (32)
Stage 3 35 (54)
Time to CLAD, median, IQRDays1114 (601–1994)
Time to ECP, median, IQRDays1411 (1014–2869)
ECP Follow up, median, IQRDays503 (404–741)
ECP cycles, median, IQR15a (12–18)
ECP interval, median, IQRDays28a (21–37)
Table 2. Univariate and multivariate analysis of all patients receiving n > 3 cycles ECP for CLAD
 Univariate analysis 
 ECP nonprogressiveECP progressivepMultivariate analysis Hazard ratio (95% CI)
Patients, n30   
Age > 55 years, n (%)12 (34)15 (50)0.2 
CLAD >1000 days post-Tx, n (%)14 (40)21 (70)0.28 
Time to ECP > 1500 days post-Tx, n (%)18 (51)11 (37)0.23 
Early CLAD <1 year, n (%)02 (6)0.12 
CLAD onset < 3 years, n (%)9 (26)5 (17)0.38 
BOS Stage 1/217 (49)13 (43)0.27 
BOS Stage 318 (51)17 (57)0.67 
Colonization, n (%)7 (20)4 (13)0.48 
ECP cycles > 10, n (%)20 (57)12 (40)0.11 
ECP interval < 2 weeks, n (%)10 (29)7 (23)0.23 
Azithromycin responder, n (%)4 (11)5 (17)0.54 
Onset Azithromycin – Onset ECP < 1 year, n (%)25 (71)14 (47)0.03 
“Rapid decliner” (>100 mL FEV1/month), n (%)6 (17)15 (50)0.0056.41 (2.77–14.81)
BAL No-Neutrophilia (≤15%), n (%)19 (54)25 (83)0.012.13 (0.86–5.31)
Restrictive pattern (≤90% Best TLC), n (%)6 (20)16 (43)0.0024.06 (1.84–9.02)
≥2 “Overlap” Phenotypes5 (19)21(81)<0.001 
Redo Tx (before), n4 (11)1 (3)0.22 
Death, n114<0.001 

BAL data at CLAD diagnosis was available in all patients, with neutrophilia evident in 29/65 (45%) patients prior to commencing azithromycin. Mean BAL neutrophilia was 43% (18–63%) in this group. In the remaining patients, mean BAL neutrophilia was 5% (2–7%). Follow-up lavage was performed after median 14 weeks (IQR 9–27 weeks) azithromycin treatment, identifying neutrophilia in 40/64 (63%) patients. At ECP initiation, neutrophilia prevalence had improved, affecting 21/65 (32%) patients, 11 of whom had displayed refractory neutrophilia under azithromycin. Due to deteriorating clinical condition, further BAL analysis was possible in only 30/65 patients. Median time to repeat BAL after commencing ECP was 27 weeks(IQR 18–43 weeks). Thirteen patients were neutrophilic, 8 of whom were persistent despite ECP. Three patients displayed refractory neutrophilia throughout the entire period of investigation.

Aspiration was assessed using Sudan staining of BAL samples. Pre-CLAD data was available for all patients, with 13/65 (20%) patients testing positive on at least one occasion. At the outset, 58/65 (89%) patients were receiving anti-reflux treatment, with 53/58 patients taking proton-pump inhibitors (PPI) and the remaining five patients H2-antagonists. No fundoplications were performed. Only two Sudan-positive patients were not regularly taking treatment due to side effects. Similar GERD prevalence was observed following initiation of azithromycin and ECP, with five patients demonstrating recurrent aspiration despite medical treatment.

Airway colonization was confirmed in 23/65 (35%) patients prior to CLAD treatment. The majority (n = 16) had Pseudomonas aeruginosa (PSA), with a further five culturing MRSA. Asymptomatic colonization was initially managed conservatively. Following a suspected exacerbation, PSA-colonized patients received sensitive intravenous antibiotics, before 6/16 pts. commenced inhaled maintenance treatment with either colistin (n = 4) or tobramycin. Exacerbations attributed to MRSA were treated with intravenous antibiotics. Under azithromycin the numbers of colonized patients fell with 16/65 (25%) demonstrating positive BAL cultures of who 9/16 had PSA and only 2/16 MRSA. Inhaled antibiotics were administered in 5/9 of these patients. Of the 30 patients undergoing BAL following ECP initiation, 11 were colonized, 4 of who had PSA. All of these patients were receiving inhaled antibiotics.

Although transbronchial biopsies (TBB) were available in 43 (66%) patients, these had mostly been performed before CLAD onset. In their final TBBs, 14 patients exhibited B1R with a further three patients demonstrating B2R. The remainder were B0. Subsequent mucosal biopsies in 23 patients with established CLAD have shown all to be at least B1R.

Sixty-four patients had previously received azithromycin, over a median duration of 221 days (IQR 109–506) before ECP. One patient proceeded directly to ECP due to known contradictions to azithromycin (long QT syndrome). Nine patients demonstrated temporary improvement of ≥10% in FEV1 (IQR 11.0–29.9%) and were considered transient azithromycin responders. The median time to ECP following transplantation was 1411 days (IQR 1014–2869). At ECP initiation, 35 patients were graded BOS stage 3, 21 patients BOS 2 and the remaining 9 patients BOS 1 or 0p. Considered collectively, the median number of ECP cycles completed by the cohort was 11 (IQR 8–15). However, 8 patients died within the first year of ECP treatment and a further 13 patients in the progressive group terminated ECP due to clinical frailty. Within the remaining 44 patients participating in consolidation ECP beyond the initial 3 months, a median of 15 cycles (IQR 12–18) were performed at a median interval of 28 days (IQR 21–37). Median follow-up after commencing ECP was 503 days (IQR 404–741). Eight patients (12%) exhibited ≥10% FEV1 response under ECP, which persisted throughout follow-up. A further 27 (42%) patients achieved and maintained FEV1 stabilization until the end of the study. Thirty patients (46%) progressed despite ECP.

During the study, 32 (49%) patients were commenced on montelukast. Of these, 29 (91%) started treatment subsequent to ECP at a median interval of 196 days (IQR 81–446). All of these patients had exhibited sufficient loss in graft function to be considered progressive under ECP. All three patients starting montelukast before ECP, did so due to delays or difficulties in accessing (n = 1) or performing ECP: intravenous access (n = 2). Upon resolution, ECP was immediately commenced.

Univariate analysis comparing “progressive” patients with those who had stabilized or improved was performed (Table 2). Time to CLAD diagnosis and initiation of ECP treatment did not influence ECP response. Neither the degree of graft dysfunction, based on BOS Stage (p = 0.80), nor evident colonization (p = 0.48) could predict ECP efficacy. “Rapid-decliners” were more likely to exhibit an ECP refractory course (6 vs. 15 pts, p = 0.005; hazard ratio [HR] 6.41, 95% CI: 2.77–14.81). Similarly, patients fulfilling RAS criteria were less likely to stabilize under ECP (6 vs. 16 pts, p = 0.002; HR 4.06, 95% CI: 1.84–9.02). “Non-neutrophilic” patients (n = 44, p = 0.01) also exhibited worse prognosis. Patients exhibiting a combination of at least two phenotypes were 21/26 progressive under ECP. Ten patients failed to fulfill criteria for any of the defined CLAD sub-types.

An initial response to azithromycin did not predict ECP response, however, patients progressing early under azithromycin necessitating ECP within 1 year, were more likely to stabilize with ECP than those progressing later (25 vs. 14 pts, p = 0.03). Of the nine patients responding to azithromycin, only three exhibited a similar response to ECP.

Survival correlated with ECP response, with median progression-free survival in those stabilizing under ECP of 401 days (inter-quartiles 301–576 days), compared to 133 days (IQR 80–190 days) in the ECP refractory group. Colonization did not influence ECP response (p = 0.48). Five patients underwent retransplantation and 15 patients died during follow-up, of whom 14 exhibited ECP refractory CLAD. The single death in the responder group was due to a sudden cardiac event.


This is the largest study examining efficacy of ECP in treating chronic allograft dysfunction following lung transplantation. Our findings demonstrate that the majority of CLAD patients (54%) progressing under azithromycin achieve spirometric stabilization with ECP during median follow-up of 1.5 years, obtaining clear survival benefit with median survival improving from 133 to 401 days (Figure 1A). No distinguishable prognostic benefit was evident for the small number of patients actually exhibiting improved lung function (Figure 1B). Finlen Copeland et al. previously characterized the clinical course of CLAD, reporting median survival after diagnosis of 2.5 years [33]. Taken in this context, the improved survival observed in patients with ECP-sensitive CLAD (2-year mortality: 1/35 pts, 2.8%) appears promising. Given that this response appears independent of CLAD duration as well as BOS stage at initiation, ECP should be considered a viable second line treatment in patients with established CLAD. These findings corroborate and substantiate previous studies alluding to ECP benefit in CLAD [22, 25, 26, 42].

Figure 1.

Kaplan-Meier curves demonstrating survival between patients demonstrating stabilized (A) or improved lung function (B) in response to ECP treatment.

As well as confirming survival benefit in ECP responders, we identified basic CLAD phenotypes that assisted in predicting ECP efficacy. Based on several recent studies, it is becoming increasingly clear that chronic allograft dysfunction can no longer be considered a single entity, going far beyond a functional description for classical bronchiolitis obliterans. For this reason we adopted the more generic term CLAD, to allow and encourage more structured and coherent terminology. Although clinical practice clearly demonstrates a variable and unpredictable clinical course in CLAD, incorporation of this in study designs remains rare. Descriptive data exists, with both Sato et al. and Verleden et al. providing convincing evidence for a sizeable RAS subtype with markedly poorer prognosis [2, 3]. Although undoubtedly complicating the current concept of BOS, it may represent a crucial step in addressing the frustratingly inconsistent outcomes characteristic of studies assessing CLAD treatments. Existing data relating to ECP typifies this with previous studies demonstrating unpredictable benefit, small patient numbers and short follow-up limiting subtype analysis and prognostic implications [25, 26]. Although a subsequent, much larger study confirmed reduced mean ΔFEV1 decline, prognostic benefit could not be confirmed [22]. Morrell et al. acknowledged variable ECP response within their undifferentiated cohort but were unable to identify factors predicting FEV1 response. Jaksch et al. recently reported on an undifferentiated CLAD cohort of 51 patients, and while demonstrating survival benefit, they too failed to identify positive predictors of ECP response [44]. Through basic phenotyping, we attempted to address this issue, both for the purposes of predicting treatment response as well as guiding prognosis. Our results clearly demonstrate advantages in this approach, confirming that rapid CLAD progression (6 vs. 15 pts. p < 0.005, HR = 6.41) or RAS (6 vs. 16 pts. p = 0.002, HR = 4.06) is less likely to respond to ECP and require early consideration regarding either palliation or re-transplantation. Although it could be argued that these findings merely represent the inevitable outcomes of particularly aggressive CLAD, our data shows that 20% of patients who stabilized and less than half of those progressing however fulfilled RAS criteria. Similarly, a small proportion of patients (n = 6 pts) with more aggressive forms of obstructive CLAD also stabilized under ECP.

The influence of BAL neutrophilia on ECP response requires careful consideration. It should be reiterated, that the proposed phenotypes are specific to established CLAD patients failing on azithromycin. Extrapolation to general CLAD populations cannot be supported and in particular comparison with neutrophilic reversible airway dysfunction [12, 14] as proposed by Verleden et al. [5] should be avoided. Doubts are emerging about the usefulness of BAL neutrophilia in measuring disease activity or treatment response in established CLAD. Our experience with ECP correlates with the findings of Meloni et al. regarding azithromycin [43] where clinical response appeared discordant with evolution in BAL neutrophilia. Intriguingly, we observed a transient increase in the prevalence of BAL neutrophilia after initiating azithromycin. This subsequently abated, before gradually increasing again under ECP. This appeared independent of clinical response. It is our impression that many confounding factors such as colonization or recurrent infection undermine the utility of BAL neutrophilia as a specific marker of disease activity in advancing CLAD. Mucosal biopsies, assessing ISHLT B-Grading have been suggested as a more disease-specific alternative. Previous studies have demonstrated correlation between B-Grade and BAL neutrophilia in acute rejection [45] and that B-Grade severity represents an independent CLAD risk factor [46]. Further studies in relation to established CLAD are, however needed.

Comparisons between azithromycin and ECP responses in CLAD based on our findings also require careful interpretation. Although a small number of patients displayed transient benefit, direct comparison between our cohort and azithromycin responders is not possible. Interestingly, our data does however suggest that patients progressing earlier under azithromycin were more likely to profit from ECP, compared to those who transiently responded to azithromycin. Based on the presented data, we consider these findings further evidence for a multi-modal basis to CLAD and reinforce the need for refined phenotypes to allow early implementation of tailored combination therapy. Our data suggests significant overlap within the groups defined and that approximately 30% of patients were classified by exclusion, undermining somewhat the overall prognostic interpretability (Figure 2). Overlapping phenotypes clearly have worst prognosis, given that 21/26 were progressive under ECP. Further research is essential, with further refinement of these CLAD phenotypes inevitable.

Figure 2.

Illustration of how our ECP treatment groups conform to the emerging CLAD phenotypes. Significant overlap was observed and a large number of patients were unable to be positively classified.

The immunological effects of ECP are not completely understood, undermining its usefulness in understanding CLAD. Upregulation of CD4+CD25+ T-regulatory cells has been demonstrated in CLAD patients demonstrating FEV1 response to ECP [25]. Laboratory studies demonstrated that ECP-treated leucocytes contain apoptotic cells or cells destined to undergo apoptosis, with subsequent in vivo phagocytosis suppressing inflammatory responses, preventing dendritic cell maturation and reducing release of Tumour Necrosis factor, transforming growth factor β1 (TGF-β1) and interleukin-10 [32]. T-Cell downregulation and enhancement of dendritic cell immune-tolerance ensue, leading ultimately to a tolerogenic state [29]. These downstream effects help explain treatment response, given that only 5–10% of leucocytes actually undergo treatment. No evidence exists, that ECP further enhances immunosuppression, with some reports suggesting paradoxical Treg upregulation [34, 35]. With respect to CLAD itself, Meloni et al. reported on downregulation of peripheral CD4+CD25+ Tcells corresponding with CLAD development [36], Following ECP, spirometric benefit was observed in 60% of patients, all of whom exhibited stabilized or increased CD4+CD25+ Tcell expression, whereas refractory patients demonstrated a corresponding decline [25]. Within our cohort, a limited amount of immunological data was available (n = 27, 13 non-progressive) at ECP initiation and following 6 months of treatment. Over this period no significant changes in serum CD4+CD25+ Tcell counts were observed. No data exists examining the histological effects of ECP in CLAD. The characteristic gradual-onset stabilization, which appears to be independent of actual disease stage suggests inhibition rather than reversal of the propagating immunological processes in certain patients.

Regarding optimal ECP implementation in CLAD, limited data exists. Our protocol reflects availability of resources and efforts to maintain patient adherence given treatment time demands. Currently, we initiate fortnightly treatment for the first 3 months (six cycles). Thereafter, patients demonstrating stabilization or improvement are cautiously considered for interval lengthening between cycles. At present the maximum interval of maintenance therapy is 8 weeks. The median interval for patients (n = 44) participating in the “maintenance phase” was however 28 (IQR 21–37) days. Treatment is intensified if progression recurs following interval lengthening, but only to the patient's previous treatment intensity. No further intensification beyond fortnightly treatment is attempted. In the absence of data to the contrary, ECP was previously continued in refractory patients for as long as practicable, being considered analogous to palliative therapy. In light of recent data from Jaksch et al. [44] we now favor an intensive trial of ECP over 3 months, with subsequent withdrawal of treatment after a minimum of 10 cycles if no signs of treatment response exist. Our protocol is admittedly less intensive than those used in previous ECP studies. Morrell et al. performed a total five complete cycles in the first month and then fortnightly until 3 months. Thereafter, patients automatically continued with monthly cycles regardless of FEV1 [22]. Meloni et al. used similarly intense initiation with three cycles in the first week and bi-weekly thereafter [25]. Benden et al. did not employ an induction phase and started 4–6 weekly maintenance cycles [26]. Given broadly comparable outcomes between our results and those of Morrell et al., there is little evidence to support an intensive initiation schema. Regarding long-term maintenance therapy protocols however, little useful data exists and the need for multicenter prospective trials is clear.

ECP represents a further attempt to harness Tcell function in chronic allograft dysfunction. Alternative therapies, such as total lymphoid irradiation (TLI) have also been shown to impair immunological reactions to allogeneic cells in renal [37], cardiac [38] and lung allograft rejection following transplantation [39]. Diamond et al. investigated 11 LTx patients, of whom only four managed to complete the 5-week treatment course. Although those completing treatment benefitted in terms of FEV1, high-associated mortality and morbidity raised serious questions. Fisher et al. subsequently demonstrated better tolerability with extended follow up to 12 years in a larger series [40]. Approximately 73% of patients completed comparable radiotherapy, demonstrating a statistically significant reduction in FEV1 decline. Significant bone marrow suppression (8/37 pts.) was noted, though clinically relevant complications were low (2/37 pts.). More recently Verleden et al. examined TLI in a small number of CLAD patients progressing under azithromycin. Using the same 8 Gy protocol they once again demonstrated reduced FEV1 decline, though no patients improved [41]. Although all patients completed TLI, they were unable to demonstrate significant mortality benefits compared to a matched historical group receiving best conservative therapy. Given broad comparisons between this cohort and our patients, the evidence suggests ECP is preferable to TLI as second line treatment in CLAD following progression under azithromycin.

The retrospective nature of our study confers several limitations. The lack of a control group or treatment randomization impairs assessment of treatment response versus the natural disease course. Although standardized treatment protocols were formulated, decisions regarding initiation and subsequent treatment intensity remained vulnerable to clinician variability. Initial reluctance given the invasive and time-intensive nature of ECP explains the high number of BOS 3 patients included. Mounting confidence in the benefits of ECP however led to more aggressive treatment initiation during the course of the study. As with all studies relying on spirometry in patients with advanced lung disease, device sensitivity and the patient's ability to accurately perform the test introduces problems in reliably interpreting dynamic change in advance disease.

In conclusion, ECP represents a viable and prognostically relevant treatment option for lung transplant recipients demonstrating established azithromycin-refractory CLAD. This effect appears to be independent of BOS stage at the time of initiation. Patient phenotyping appears to predict treatment response and ultimately prognosis. The proposed forms of RAS and neutrophilic CLAD appear to predict ECP response. Patients with RAS or rapidly progressive disease (non-neutrophilic) and especially overlap patients were significantly less likely to benefit from ECP therapy and have a worse prognosis. Limitations in the proposed CLAD phenotypes remain, with evidence of significant overlap and a large proportion of patients falling outside the current nomenclature. With the exception of FEV1, no reliable markers assist to assess disease activity or treatment response in established CLAD, with BAL neutrophilia appearing increasingly unreliable. Many questions remain regarding optimal initiation, implementation, duration and ending of ECP treatment. Investigator consensus and larger multicenter trials are needed to fully evaluate treatment benefit.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.