Published ahead of print on February 8, 2008, doi:10.1164/rccm.200706-951OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200706-951OC
Severity of Lymphocytic Bronchiolitis Predicts Long-Term Outcome after Lung Transplantation1 The Lung Transplant Unit, and 2 Department of Anatomical Pathology, St. Vincent's Hospital, Darlinghurst, New South Wales, Australia Correspondence and requests for reprints should be addressed to Allan R. Glanville, M.D., F.R.A.C.P., The Lung Transplant Unit, Xavier 4, St. Vincent's Hospital, Victoria Street, Darlinghurst NSW, Australia 2010. E-mail: aglanville{at}stvincents.com.au
Rationale: Severe and recurrent acute vascular rejection of the pulmonary allograft is an accepted major risk factor for obliterative bronchiolitis. Objectives: We assessed the role of lymphocytic bronchiolitis as a risk factor for bronchiolitis obliterans syndrome (BOS) and death after lung transplantation. Methods: Retrospective analysis of 341 90-day survivors of lung transplant performed in 1995–2005 who underwent 1,770 transbronchial lung biopsy procedures. Measurements and Main Results: Transbronchial biopsies showed grade B0 (normal) (n = 501), B1 (minimal) (n = 762), B2 (mild) (n = 176), B3 (moderate) (n = 70), B4 (severe) (n = 4) lymphocytic bronchiolitis, and Bx (no bronchiolar tissue) (n = 75). A total of 182 transbronchial biopsies were ungraded (8 inadequate, 142 cytomegalovirus, 32 other diagnoses). Lung transplant recipients were grouped by highest B grade before diagnosis of BOS: B0 (n = 12), B1 (n = 166), B2 (n = 89), and B3–B4 (n = 51). Twenty-three were unclassifiable. Cumulative incidence of BOS and death were dependent on highest B grade (Kaplan-Meier, P < 0.001, log-rank). Multivariable Cox proportional hazards analysis showed significant risks for BOS were highest B grade (relative risk [RR], 1.62; 95% confidence interval [CI], 1.31–2.00) (P < 0.001), longer ischemic time (RR, 1.00; CI, 1.00–1.00) (P < 0.05), and recent year of transplant (RR, 0.93; CI, 0.87–1.00) (P < 0.05), whereas risks for death were BOS as a time-dependent covariable (RR, 19.10; CI, 11.07–32.96) (P < 0.001) and highest B grade (RR, 1.36; CI, 1.07–1.72) (P < 0.05). Acute vascular rejection was not a significant risk factor in either model. Conclusions: Severity of lymphocytic bronchiolitis is associated with increased risk of BOS and death after lung transplantation independent of acute vascular rejection.
Key Words: lung transplantation • graft rejection • bronchiolitis obliterans
Obliterative bronchiolitis (OB) remains the most common late cause of morbidity and mortality after lung transplantation (LTx) (1). Acute pulmonary allograft rejection characterized by perivascular lymphocytic infiltrates is the major risk factor, but the mechanism linking acute rejection and small airway occlusion is not well established (2, 3). Lymphocytic bronchiolitis (LB) is considered to be a probable risk factor for OB but has not been embraced widely by the lung transplant community despite strong experimental evidence (2, 3). In 1988, Tazelaar and colleagues reported airway histology in nonimmunosuppressed rat lung allografts (4). In acute rejection, bronchioles were surrounded by a dense cuff of activated lymphocytes associated with focal necrosis, epithelial ulceration, and granulation tissue. In 1997, Boehler and associates demonstrated in the rat tracheal transplant model that lymphocytic infiltration was an invariable precursor to fibrous airway obliteration (5). In the clinical arena, Yousem in 1993 found that LB was observed more frequently in patients who developed OB than in those who did not (6). LB was preceded by acute vascular rejection in 20 of 26 cases and often persisted after the initial biopsy. Girgis and coworkers analyzed risks for the development of bronchiolitis obliterans syndrome (BOS) in 74 90-day survivors (7). By univariable analysis, significant risks were acute rejection score (sum of pathologic grades of each acute rejection episode) (P < 0.01) and LB (P = 0.03). LB in the absence of infection was associated with development of OB in 12 of 13 patients within 6 months of diagnosis. El-Gamel and colleagues found LB, analyzed as a categorical variable, in 21 of 26 patients with OB versus 4 of 50 patients without OB (8). The 1995 revisions to the 1990 International Society for Heart and Lung Transplantation (ISHLT) Lung Rejection Study Group (LRSG) Working Party recommended that each grade of acute rejection should mention the presence and intensity of coexistent airway inflammation (9). Husain and coworkers subsequently devised a quantitative method to retrospectively study transbronchial lung biopsy (TBBx) in 90-day survivors (10). TBBx were regraded using a scale of 0–4 for acute perivascular rejection (A score) and LB (B score) and the grades were summed over time. The mean A score for BOS (n = 66) was 6.2 compared with 3.2 for those without BOS (n = 68). B scores were 5.3 and 1.7, respectively. Late acute rejection and LB were significantly associated with BOS as was decreased immunosuppression. These remain the most comprehensive clinical data to support the notion that LB is a manifestation of acute rejection with similar implications for graft function as acute vascular rejection. We reasoned that increasing severity of LB should be associated with a higher risk of OB and now present results from a 10-year study using the revised criteria to analyze the predictive value of LB for the development of BOS and survival after LTx. Some of the results have been reported previously in the form of an abstract (11).
Patients were selected for LTx on the basis of standard criteria (12–14). A total of 341 of 366 (93.2%) consecutive recipients survived more than 90 d and were evaluable for BOS (Figure 1) (2, 3, 9). Underlying diseases included cystic fibrosis (n = 100), chronic obstructive pulmonary disease (n = 123), diffuse parenchymal lung disease (n = 52), pulmonary hypertension (n = 28), and other (n = 38). Type of transplant included heart–lung (n = 27), single lung (n = 69), and bilateral LTx (n = 245) (Table 1). Follow-up was per protocol (15). Patients monitored FEV1 daily on home spirometers and were instructed to report reductions of 10% or more. Formal lung function testing was performed at each clinic visit.
Surveillance TBBx were performed at 3, 6, 9, and 12 weeks and where indicated by symptoms or for follow-up of acute rejection or cytomegalovirus (CMV) pneumonia using standard techniques (16, 17). TBBx were also performed at 6, 9, 12 months in patients with acute rejection (ISHLT > A1) within the first 4 months (18). TBBx were graded as A0, A1, A2, A3, and A4, with concomitant B grades B0, B1, B2, B3, B4, and Bx using ISHLT criteria (9). "Bx" biopsies were evaluable for A grading but did not show bronchiolar epithelium to allow a B grade to be determined. "Inadequate" biopsies did not show either bronchiolar epithelium or sufficient alveolar tissue to meet ISHLT criteria for grading. TBBx with CMV pneumonia or fungal infection were excluded from B-score analysis as were TBBx with histopathologic evidence of an acute neutrophilic bronchiolitis with or without positive cultures. TBBx were analyzed by the same pathologist (S.R.) who was blinded to clinical status. BOS was diagnosed using ISHLT criteria (3). Pulmonary function tests met guidelines of the ATS (19). All LTx recipients received triple-drug immunosuppressive therapy with cyclosporine or tacrolimus, azathioprine or mycophenolate mofetil, and prednisolone (20). Cyclosporine monitoring used C2 targets in the latter half of the time period of study (21, 22). Acute rejection (ISHLT grade > A1) was treated with intravenous methyl prednisolone (12.5 mg/kg/d) for 3 days followed by an oral taper of prednisolone. Symptomatic A1 rejection was treated with an oral prednisolone alone (18). There was no standard protocol for treatment of isolated B-grade rejection. CMV-seronegative patients who received their transplant from a CMV-seropositive donor received antiviral prophylaxis with intravenous ganciclovir (5 mg/kg thrice weekly) for a total of 10 weeks. CMV pneumonia was treated with intravenous ganciclovir (5 mg/kg, twice a day for 14–21 d, and thrice weekly at 10 mg/kg/d for the next 14 d) (23, 24). Surveillance bronchoscopy included assessment for fungi, mycobacteria, and chlamydia (25, 26).
Definitions
Statistical Analysis
We followed 341 patients for a total 1,371 patient-years, with a mean (±SD) follow-up duration of 4.02 ± 2.76 years (median, 3.39 yr; range, 0.33–11.00 yr). Demographic details of the four groups are shown in Table 1. There was no significant difference between groups with respect to demographics, transplant type, underlying condition, or proportion of patients evaluable for BOS. Distribution of B grades and other diagnoses for each biopsy taken is shown in Table 2 by group. Twenty-three of 341 (6.7%) patients could not be ascribed a B group due to inability to obtain diagnostic tissue, presence of exclusion criteria (as above), or an unfavorable risk–benefit ratio for performing TBBx. More biopsies were performed in the B2 and B3–4 groups due to the fact that follow-up biopsies were generally performed after each abnormal biopsy. Table 3 demonstrates the median times to the first diagnosis of LB in each group and the median times to the development of BOS and death after the diagnosis of LB.
Univariable risks for BOS are described in Table E1 in the online data supplement and included highest B grade, average B grade, sum of B grades, highest A grade, average A grade, and sum of A grades. Longer ischemic time, type of transplant bilateral lung transplant, and recent year of transplant were associated with a reduced risk. Univariable risks for death are described in Table E2 and included BOS, highest B grade, average B grade, sum of B grades, highest A grade, average A grade, and sum of A grades. Longer ischemic time and recent year of transplant were associated with a reduced risk. Multivariable analysis of risks for BOS is described in Table 4, which shows that highest B grade (relative risk [RR], 1.62; 95% confidence interval [CI], 1.31–2.00) (P < 0.001) increased risk, whereas longer ischemic time (RR, 1.00; CI, 1.00–1.00) (P < 0.05) and recent year of transplant (RR, 0.93; CI, 0.87–1.00) (P < 0.05) reduced risk. Multivariable risks for death described in Table 4 included BOS as a segmented time-dependent covariable (RR, 19.10; CI, 11.07–32.96) (P < 0.001) and highest B grade (RR, 1.36; CI, 1.07–1.72) (P < 0.05).
Figure 1 demonstrates the distribution of all patients who underwent transplantation in 1995–2005 and their allocation to groups by highest B grade determined before the diagnosis of BOS for the 93.2% who survived greater than 3 months. Figure E1 shows outcomes on all biopsies performed in 1995–2005 with a focus on 1,513 biopsies that were evaluable for B grading. The number of lung transplant recipients with a particular sum of A-grade and B-grade biopsies is shown in Figures 2 and 3, respectively, which demonstrates that a substantial number of recipients did not record any grade-A rejection. However, the absence of grade-B rejection was rare. Figure E2 shows the frequency of the association of the highest A-grade TBBx ever detected with the highest B-grade TBBx ever detected in an individual patient sorted by number of biopsies performed. Figures 4 and 5, respectively, show the cumulative incidence of BOS by B group after the diagnosis of LB and mortality by B group after the diagnosis of LB, and in each case show a significant difference between groups (Kaplan-Meier, P < 0.001, Mantel-Cox). Importantly, both for BOS and mortality, the B2 group was significantly different from either the B1 or B3–4 groups.
The B0 Group Only 12 patients met criteria for the B0 group. Freedom from BOS stage of 1 or greater at 5 years post-transplant was 100% in this group compared with 60% in the B1 group (Figure 4), but because two patients developed late BOS, the difference was not significant thereafter (B0 vs. B1,  2 = 1.82, P = 0.18, log-rank). Post-transplant survival was equivalent to the B1 group, at 80% at 5 years (B0 vs. B1, 2 = 0.16, P = 0.90, log-rank).
The B3–4 Group A grades recorded concurrently with the 74 B3–4 TBBx were A0 (n = 27), A1 (n = 16), A2 (n = 9), A3 (n = 20), and A4 (n = 2). Individual A grades recorded within the 4 weeks before the first B3–4 TBBx were A0 (n = 20), A1 (n = 7), A2 (n = 8), and A3 (n = 4) with CMV pneumonia (n = 4). Follow-up A grades recorded within the 4 weeks after a B3–4 TBBx were A0 (n = 32), A1 (n = 16), A2 (n = 8), and A3 (n = 3) with CMV pneumonia (n = 3). The average A grade (n = 74) concurrent with the B3–4 TBBx was 1.38 ± 1.30. The highest A grades recorded per patient at any time post-transplant by the 51 patients in the B3–4 TBBx group were A0 (n = 2), A1 (n = 12), A2 (n = 14), A3 (n = 21), and A4 (n = 2).
Our data show that the severity of LB detected on TBBx is the most significant risk factor for the development of BOS, which is the major risk factor for death after lung transplant. LB was also an independent risk factor for death. Of interest, severity of vascular rejection was a risk factor in univariable analysis (Tables E1 and E2), confirming many other studies, but was not an independent risk factor for either BOS or death in multivariable analyses (Tables 4 and 5). This suggests that it is the concomitant grade of LB that determines outcome when acute vascular rejection is diagnosed. Dogma regarding the strong relationship between acute vascular rejection and subsequent development of BOS has arisen largely from early studies that did not consider LB at all and later studies that have not analyzed LB comprehensively as a multivariable risk (27).
By design, surveillance procedures detect asymptomatic rejection, which has been treated uniformly in our program with augmented immune suppression (18). In contradistinction, LB was not treated unless associated with vascular rejection, which was a common but not invariable accompaniment (Figure E2). Our study was not designed to allow a prospective analysis of therapy for LB with or without concomitant acute vascular rejection. Previous attempts at treating LB have highlighted the fact that associated changes in lung function are often refractory to corticosteroid therapy. Ross and colleagues found that BOS developed in 13 of 20 (65%) recipients after detection of an isolated grade-B lesion (in the absence of perivascular infiltrates), despite therapy with corticosteroids, and suggested that alternative therapies should be used (28). Given the data reported by Husain and coworkers (10) and Ross and colleagues (28), it is somewhat puzzling as to why other centers have not embraced analysis of LB lesions found on TBBx. Potential problems with determining an accurate LB grade include absence of evaluable bronchiolar tissue and confounding by the presence of infection, which may not be known by the pathologist at the time of reporting. Close clinicopathologic liaison is therefore necessary to exclude false-positive results, and we deliberately excluded cases in which acute infection was diagnosed and treated (Figure E1). Similarly, the experience and familiarity of the pathologist with the grading criteria are key factors in achieving reproducible results. Ultimately, the utility of a grading system depends on its sensitivity and specificity, and we acknowledge that the grading reliability of the current LRSG system has been questioned by a study that found only modest interreader (A-grading kappa, 0.65; 95% CI, 0.60–0.70; B-grading kappa 0.26; 95% CI, 0.14–0.39) and intrareader (A-grading kappa, 0.65; 95% CI, 0.53–0.76; B-grading kappa, 0.33; 95% CI, 0.15–0.51) agreement of grades of acute allograft rejection (29). Conversely, another study of interobserver agreement found kappa values of 0.79–0.82 for both LB and alveolar lesions but cautioned greater discordance between observers for moderate/severe alveolar lesions (30). Grading reliability also depends on number and adequacy of samples, with animal studies suggesting that five pieces of lung tissue were needed to yield a sensitivity of 92% (CI, 82–100%) to identify mild rejection with TBBx (31). We averaged over eight evaluable samples per procedure and all procedures were performed, or supervised for adequacy, by senior consultant thoracic physicians (17). To determine the specificity of perivascular infiltrates for the diagnosis of rejection, Tazelaar reviewed 42 cases of pneumocystis and CMV pneumonia diagnosed by surgical (33 cases) and transbronchial lung biopsy (9 cases) from non–LTx immunosuppressed patients (32). Perivascular lymphocytic infiltrates similar to those observed in lung rejection were identified in 21% of cases with pneumocystis (n = 33) and 42% with CMV (n = 7). These findings argue for caution in interpreting the presence of perivascular inflammation on TBBx until a diagnosis of infection is excluded. Therefore, we excluded cases of CMV and fungal infections from our analysis despite morphologic observations that suggest CMV infection of bronchiolar wall cells is rare (33). Although conclusions must be guarded due to small sample size, the B0 group recorded no BOS within 5 years of transplant. A direct link between LB and OB is plausible and conceptually appealing, perhaps facilitated by epithelial injury rather than associated acute vascular rejection (34, 35). Duncan and coworkers used T-cell antigen receptor β-chain variable region RNase protection assays to quantitate circulating CD4+ and CD8+ repertoires of transplant recipients with OB (36). Finding CD4+ expansions had 100% sensitivity and 80% specificity for the presence or imminent development of OB, suggesting that proliferations of CD4+ T cells were important in OB pathogenesis. CD4 proliferations are most likely part of a major histocompatibility complex class II–dependent process of indirect alloantigen presentation as predicted by Burke and colleagues (37), perhaps mediated by epithelial presentation of both class 1 and 2 antigens (38). Milne and associates, but not Hasegawa and coworkers, found that class II antigens in BOS were up-regulated and acute rejection manifested intense HLA-DR expression in epithelia and endothelial cells (39, 40). HLA-DR expression may be a critical step in the subsequent development of LB and BOS. Indeed, T-cell alloreactivity to donor HLA class II molecules likely plays a role in the pathogenesis of BOS after LTx (41). In a 4-year prospective study of 51 LTx recipients, Girnita and coworkers found risk factors for BOS were as follows: LB as a categorical variable (P < 0.0001), persistent acute rejection (P < 0.05), and number of HLA-DR mismatches (P < 0.05). Importantly the presence of HLA-specific antibodies exhibited a cumulative effect on BOS when associated with either LB or persistent acute rejection (42). Yousem and associates noted the presence of B cells in the grafts of patients with acute rejection (AR) who did not respond to bolus corticosteroids and postulated a role for antibody-mediated rejection in 1994 (43). Subsequently, Magro and colleagues, in 2003, investigated the role of humoral immunity in the pathogenesis of BOS and concluded that potential antigenic targets included the bronchial wall microvasculature, bronchial epithelium, and chondrocytes (44). Although unproven, humoral rejection could contribute to the development of OB in the absence of LB on TBBx, which may explain the development of BOS in some of our patients. Missed LB, either due to biopsy sampling error or timing of sampling, is another possible explanation as is the effect of immediate primary graft dysfunction (PGD), which has been associated with an increased risk of BOS, independent of acute rejection, LB, and community-acquired respiratory viral infections (45). We were not able to assess the impact of PGD in our population because early radiographs had been destroyed on conversion to a digital system, which prevented accurate allocation of the scoring system (46). Nor were we able to assess the putative role of gastroesophageal reflux, community-acquired viral infections, or HLA antibodies (3) because uniform data were not collected prospectively throughout the time period of observation. Moreover, the majority of patients with community-acquired viral infections were diagnosed by nasopharyngeal swabs (47, 48) and did not undergo TBBx, which reduces one potential confounding variable in the interpretation of the B grades. We acknowledge difficulties in modeling outcomes for time-dependent variables using the Cox methodology, so we used a time-segmented analysis of BOS as a risk factor for death. Also inherent in our Cox analysis is the assumption that grading is ordinal, which may introduce error, so we also described time to BOS and time to death as principal outcome parameters using Kaplan-Meier analysis. Other limitations of our study include the fact it is a single-center study in which management strategies have changed over time, the lack of a standard therapeutic approach to LB, operating characteristics of the grading system, and potential confounding by other untested variables, such as the influence of humoral rejection. Confounding by patients who were not able to be classified into a group, not able to be evaluated for BOS, or by sampling error of TBBx may have led to bias, but the likely impact should have been insignificant because these represented less than 7% of all subjects. Our study is one of the largest and longest studies of TBBx in the literature to date and the first to examine severity of LB as a risk factor for BOS and survival after lung transplant. Importantly, the competing risks of LB and acute vascular rejection were assessed in tandem. In multivariable analysis, severity of A-grade rejection was not identified as an independent risk factor for BOS or death, which challenges current dogma. In conclusion, severity of LB is an important factor in determining outcome after LTx because it is a major risk factor for BOS. A single B2 or higher lesion is associated with reduced survival. Conversely, lung transplant recipients who survive more than 90 days, and who only ever have B0–B1 TBBx, have a 70% 10-year survival (Figure 4) at which point 50% of patients remain free of BOS (Figure 5), which portends superior longevity than previously anticipated. The data are sufficiently compelling to suggest future therapeutic trials should focus on LB severity as a powerful surrogate marker of long-term outcome.
The authors thank the members of the Lung Transplant Unit who assisted in the care of these patients, and Matthew Law for statistical advice.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200706-951OC on February 8, 2008
Conflict of Interest Statement: A.R.G. has participated as an unpaid speaker in scientific meetings and educational courses organized and financed by various pharmaceutical companies (Novartis, Roche, Astellas), received honoraria of $1,000 in 2005–2007 for serving on an advisory board for Roche, $1,000 in 2005–2007 for serving on an advisory board for Novartis, and $500 in 2006–2007 for serving on an advisory board for Janssen-Cilag. A.R.G.'s institution is the sponsor for the CeMyLungs trial with a budget of Received in original form June 27, 2007; accepted in final form February 5, 2008
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
1
2,400,000 from 2004–2012, from Novartis to manage this study. C.L.A. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. A.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.P. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.A.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. 
