INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The incidence of both acute rejection and chronic allograft failure appears to be higher in lung transplantation than that observed with other solid organ allografts (1). The principal cause of chronic allograft loss in lung transplantation is obliterative bronchiolitis (OB), a process defined by the presence of bronchiolar inflammation and fibrosis in conjunction with progressive airflow obstruction (4). OB responds poorly to augmented immunosuppression, and is thought to be the end result of multiple immunologic, virologic, genetic, and environmental effects on the transplanted lung. A number of studies examining risk factors for OB have identified multiple and high-grade antecedent episodes of acute perivascular rejection (pathologic Grade A2 or higher) and cytomegalovirus (CMV) pneumonitis as the major risk factors for OB (5). Although the pathogenetic mechanisms linking acute perivascular rejection and CMV pneumonitis to chronic airway obliteration remain undefined, presumably both conditions give rise to elaboration of profibrotic cytokine profiles.

It has been well established in other solid organ transplants that the more disparate the differences between donor and recipient at the HLA loci the greater the risk for allograft rejection (11, 12). Similar rejection risk in lung transplant recipients has been demonstrated, notably due to mismatching at the HLA-DR and HLA-B loci (3). This is predominantly due to direct recognition of donor HLA antigens on either graft endothelium or donor-derived passenger leukocytes by recipient T cells, leading to acute allograft rejection. Shared HLA structures between donor and recipient are not seen as foreign because a recipient's T cells have been educated to recognize exogenous or endogenous peptides presented by self-HLA molecules on autologous antigen-presenting cells. Consequently, it is possible that shared HLA structures between donor and recipient could be used by recipient T cells to more efficiently eliminate foreign antigens such as CMV from either the transplanted organ or passenger leukocytes.

The present study was undertaken with two objectives. First, we sought to confirm the previously identified variables for OB in a well-defined single center population. Second, we sought to investigate the specific effects of mismatching between donor and recipient at various HLA loci both directly on progression to OB and indirectly via influencing development of the identified risk factors for OB. On the basis of these relationships, we planned to generate a predictive formula to estimate the risk of OB after lung transplantation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Transplant Population

The present study represents a continuation of our previous work analyzing the relationships between HLA mismatching and lung allograft rejection (3). The study population has been described previously in detail (3). All patients have been monitored for an additional 2 yr and monitored for the development of OB. In brief, the study population consisted of 152 consecutive patients who underwent lung transplantation (100 single lung, 52 bilateral lung) at our institution between 1990 and 1996. The study population included 82 (54%) women and 70 (46%) men with a mean age of 45 ± 15 (SD) yr. Underlying diseases were emphysema (45%), cystic fibrosis (20%), pulmonary fibrosis (16%), pulmonary hypertension (13%), and miscellaneous (6%). After transplantation, patients were monitored by regular visits in the outpatient department. Complete evaluations were performed weekly for the first 2 mo, biweekly for the next 3 mo, and then monthly thereafter. Mean follow-up time after transplantation was 33 ± 27 (SD) mo (range, 0.1-103 mo), and follow-up was complete in all patients. Standard maintenance immune suppression consisted of cyclosporine or tacrolimus, azathioprine, and prednisone.

Diagnosis and Treatment of Rejection

Surveillance fiberoptic bronchoscopy with transbronchial biopsies was performed on a schedule of 3 wk, 3 mo, 6 mo, 9 mo, and 1 yr after transplantation. In addition, biopsies were performed at any time in the presence of clinical indications, such as cough, fever, and dyspnea; an abnormal chest radiograph; or a decline in spirometry (3). Follow-up biopsies were performed 4 wk after treated rejection to ensure resolution of rejection. During the second postoperative year, surveillance biopsies were performed every 4 mo, and thereafter, biopsies were obtained every 6 mo. At least six and usually eight specimens were obtained from different subsegments and immediately placed in formalin for routine histopathology. Biopsies were graded according to the standard criteria of the Lung Rejection Study Group (13). In this system, acute cellular rejection is defined by the presence of perivascular mononuclear cell infiltration. Acute rejection varies from minimal (Grade A1) and mild (Grade A2) to moderate (Grade A3) and severe (Grade A4). Normal pulmonary parenchyma without perivascular infiltrates is considered to have no rejection (Grade A0). Airway inflammation is designated Grade B; obliterative bronchiolitis is designated Grade C.

Only histologically proven episodes of acute perivascular rejection were included in the analysis. Episodes of rejection Grade A1 were monitored without treatment. Episodes of rejection Grade A2 or higher were treated with daily infusion of methylprednisolone, 1 g intravenously for 3 d. Rejections that persisted after methylprednisolone therapy were treated with OKT3 or anti-thymocyte globulin. There were no episodes of Grade A4 rejection.

Diagnosis and Treatment of Obliterative Bronchiolitis

Obliterative bronchiolitis was diagnosed by the earlier occurrence of either (1) pathologic evidence of OB, or (2) an unexplained decline in absolute FEV₁ by 20% from highest baseline levels (bronchiolitis obliterans syndrome, BOS) (4). In the BOS system, severity of BOS varies from mild (Stage 1, FEV₁ 66 to 80% of baseline value) to moderate (Stage 2, FEV₁ 51 to 65% of baseline value) and severe (Stage 3, FEV₁  50% of baseline value). Each BOS stage may be subcategorized as "a" (without evidence of obliterative bronchiolitis) or "b" (with evidence of obliterative bronchiolitis) (4). Initial treatment of OB consisted of methylprednisolone infusion, similar to acute rejection. Decline in pulmonary function that persisted after methylprednisolone therapy was treated with anti-thymocyte globulin, mycophenolate, or cyclophosphamide.

HLA Typing

Recipients and donors were typed for HLA-A, -B, and -DR antigens using monospecific anti-HLA sera from our laboratory together with HLA-typing trays for ethnic subgroups obtained from One Lambda (Los Angeles, CA). HLA typing was performed with purified T cells for detection of HLA Class I antigens and purified B cells for detection of HLA Class II antigens. Complete HLA typing for both donor and recipient was available in 93% of cases. Zero, one, and two mismatches occurred at the HLA-A locus in 1, 44, and 55%, respectively; at the HLA-B locus in 0, 22, and 78%, respectively; and at the HLA-DR locus in 1, 35, and 64%, respectively.

Analyzed Variables

Variables applied in univariate and multivariable analysis were grouped in five broad categories:

1. 1. Preoperative donor variables: Donor age, sex, race, smoking, Gram stain, HLA type, CMV status
2. 2. Preoperative recipient variables: Recipient age, sex, race, diagnosis, preoperative corticosteroids, HLA type, CMV status
3. 3. Preoperative donor-recipient matching: Age, sex, race, HLA type, CMV status
4. 4. Intraoperative variables: Ischemic time, single lung versus bilateral lung transplant, cardiopulmonary bypass, surgeon
5. 5. Postoperative complications: Ischemia-reperfusion injury, postoperative pneumonia, bronchial ischemia, bronchial dehiscence, bronchial stenosis, diaphragmatic paralysis, acute perivascular rejection, lymphocytic bronchitis/bronchiolitis, CMV infection, CMV pneumonitis

Ischemia-reperfusion was assessed by hypoxemia (Pa_O2/FI_O2) and lung injury scores (14). The diagnosis of postoperative pneumonia was based on the presence of a new intrathoracic opacity (other than atelectasis) and either (1) culture, serologic, cytologic, or histopathologic evidence of a specific organism or organisms or (2) a clinical course in which pulmonary infection appeared highly likely (15). Bronchial ischemia was defined by bronchoscopic visualization of dark, necrotic mucosa overlying the bronchial anastomosis (16). Anastomotic dehiscence was defined as disruption of more than 25% of the circumferential suture line observed at bronchoscopy (16). Stenosis was defined as narrowing of the bronchial lumen to less than 4.9 mm in diameter (the outer diameter of the bronchoscope) (16). CMV pneumonitis was diagnosed by demonstration of characteristic large cells with haloed "owl-eyed" basophilic intranuclear inclusion bodies in bronchoalveolar lavage (BAL) fluid or lung tissue, together with tissue evidence of interstitial pneumonitis (17). Isolation of CMV in culture of lavage fluid without evidence of pneumonitis or inclusion bodies was considered CMV infection (17).

Statistical Analysis

Risk factors for obliterative bronchiolitis were evaluated by Kaplan- Meier product-limit estimate (univariate analysis) and standard contingency tables, censoring individuals at the time of death. The Cox proportional hazards model was used for the multivariable analysis of the time to diagnosis of OB. This provided relative risks of failure and 95% confidence intervals, adjusting for the other independent risk factors. The relationship between CMV pneumonitis and HLA mismatches was evaluated by the Kaplan-Meier method. Predictions of freedom from OB at 1, 2, and 3 yr were generated by the Cox proportional hazards model, using the methods described by Allison (18). It is important to note that 1-, 2-, and 3-yr Kaplan-Meier probabilities differ from those generated by the Cox model because of the proportional hazards requirement of the Cox model. In addition, the Cox model estimate utilizes all patients whereas the stratified Kaplan- Meier computes only the estimate for the patients with a particular outcome, such as acute rejection. The hazard function (instantaneous risk of OB) was estimated parametrically, by means of the hazard function regression model of Blackstone and coworkers, using maximum likelihood methods (19). Data were analyzed with SAS system software (SAS Institute, Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Time to Onset of OB

The probability of OB after lung transplantation, diagnosed by the earlier occurrence of either pathologic evidence of OB, or unexplained decline in absolute FEV₁ by 20% from highest baseline levels, is shown in Figure 1. The data indicate that 13% of patients developed OB by 1 yr after transplantation, 30% by 2 yr, and 56% by 3 yr. The median time to onset of OB for the entire study population was 2.7 yr. By diagnostic criteria using only BOS spirometric criteria (unexplained decline in absolute FEV₁ by 20% from highest baseline levels), the median time to onset of OB was longer (4.2 yr). (No statistical value can be assigned because these are the same population.)

View larger version (23K): [in this window] [in a new window]   Figure 1.   Probability of freedom from obliterative bronchiolitis (OB) after lung transplantation. Kaplan-Meier product-limit estimate demonstrating that 50% of patients developed OB by 2.7 yr after transplantation.

Hazard function analysis (instantaneous rate of developing OB) revealed a rising rate of OB during the first year after lung transplantation, followed by a relative plateau during the second year (Figure 2). Subsequently, there was decline in the instantaneous rate of developing OB up to 6 yr after transplant (Figure 2).

View larger version (16K): [in this window] [in a new window]   Figure 2.   Hazard function (instantaneous risk) of developing OB after lung transplantation, demonstrating a rising rate of OB during the first year, a relative plateau during the second year, and a decline in risk thereafter.

Non-HLA Risk Factors for OB

The major risk factors for OB are listed in Table 1. By univariate analysis, any episode of Grade A2 or A3 acute rejection (p = 0.0021) was the most significant risk factor for OB. Episodes of Grade A1 acute rejection showed no significant association with OB. Addition of the variable Grade A1 acute rejection to Grades A2 and A3 (any Grade A rejection) reduced the association between acute rejection and OB (p = 0.0354) as compared with Grade A2 or A3 alone. The second significant risk factor for OB was CMV pneumonitis (p = 0.0101). Bilateral lung transplants and lymphocytic bronchitis (Grade B histology) showed trends as risk factors for OB (p = 0.0733 and p = 0.1280, respectively). None of the other postoperative complications such as reperfusion injury, bronchial ischemia, postoperative pneumonia, or CMV infection showed any significant association with OB.

The influence of Grade A2 or A3 acute rejection on the probability of OB is shown in Figure 3. The upper curve represents recipients with no episodes of Grade A2 or A3 acute rejection, whereas the lower curve represents recipients with at least one episode of Grade A2 or A3 acute rejection. The curves are significantly different (p [log-rank] = 0.0021). The data show that by 1 yr after transplant, 27% of recipients with at least one episode of Grade A2 or A3 acute rejection developed OB, compared with 11% of recipients with no episodes of Grade A2 or A3 acute rejection (Figure 3). By 2 yr after transplant, 52% of recipients with at least one episode of Grade A2 or A3 acute rejection developed OB, compared with 35% of recipients with no episodes of Grade A2 or A3 acute rejection. A similar curve demonstrating the influence of CMV pneumonitis on the probability of OB is shown in Figure 4A.

View larger version (13K): [in this window] [in a new window]   Figure 3.   Influence of Grade A2 or A3 acute rejection on the development of OB after lung transplantation. Recipients with no episodes of Grade A2 or A3 acute rejection (upper curve) had a higher probability (p [log-rank] = 0.0021) of freedom from OB as compared with recipients with at least one episode of Grade A2 or A3 acute rejection (lower curve).

View larger version (14K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Figure 4.   (A) Influence of CMV pneumonitis on freedom from OB after lung transplantation. Recipients who did not develop CMV pneumonitis (upper curve) had a higher probability (p [log-rank] = 0.0100) of freedom from OB as compared with recipients who developed CMV pneumonitis (lower curve). (B) Influence of mismatching between donor and recipient at the HLA-A locus on freedom from OB after lung transplantation. Recipients with at least one match at the HLA-A locus (upper curve) had a higher probability (p [log-rank] = 0.0341) of freedom from OB as compared with recipients with no matches at the HLA-A locus (lower curve).

HLA Risk Factors for OB

The only significant HLA risk factor for OB was mismatching at the HLA-A locus (p = 0.0366) (Table 1). The influence of mismatching at the HLA-A locus on the probability of OB is shown in Figure 4B. Mismatching at the HLA-B and -DR loci showed no significant association with OB (Table 1).

Multivariable Analysis of Risk Factors for OB

By Cox proportional hazards modeling, three risk factors for OB were significant, namely Grade A2 or A3 acute rejection (p = 0.0126), mismatching between donor and recipient at the HLA-A locus (p = 0.0144), and CMV pneumonitis (p = 0.0358) (Table 2). The risk of OB associated with Grade A2 or A3 acute rejection was 2.2 relative to that associated with no episodes of Grade A2 or A3 acute rejection (95% confidence intervals, 1.2 to 4.2). The risk of OB associated with HLA-A mismatching was 2.1 (95% confidence intervals, 1.2 to 3.7), and the risk of OB for CMV pneumonitis was 1.9 (95% confidence intervals, 1.0 to 3.4).

Predicting the Risk of OB

On the basis of Cox proportional hazards modeling, a predictive formula was derived to estimate the risk of OB after lung transplantation. In the least favorable situation, namely, a recipient who was mismatched with the donor at the HLA-A locus, and who sustained Grade A2 or A3 acute rejection and CMV pneumonitis, the probability of OB at 1, 2, and 3 yr was 42, 73, and 92%, respectively. In the most favorable situation, for a recipient who had at least one match with the donor at the HLA-A locus, and who did not sustain either Grade A2 or A3 acute rejection or CMV pneumonitis, the probability of OB at 1, 2, and 3 yr was 7, 15, and 25%, respectively. All other combinations of predictive risk factors for OB are tabulated in Table 3.

HLA Risk Factors for CMV Pneumonitis

In the study population, CMV pneumonitis occurred with similar frequency in all donor-recipient matchings except the pairing of a CMV-negative donor and CMV-negative recipient (D/R). The frequencies of CMV pneumonitis in the three pairings with either a CMV-positive donor or CMV-positive recipient were 44% in the D⁺/R⁺ group, 30% in the D⁺/R group, and 23% in the D/R⁺ group, respectively (p = NS). The frequency of CMV pneumonitis in the D/R group was 3% (p = 0.0001).

Mismatching at the HLA-DR locus was a risk factor for CMV pneumonitis only when the recipient had been exposed to CMV before transplantation (R⁺) (Table 4). When the recipient had been exposed to CMV before transplantation (D⁺/R⁺ and D/R⁺), the occurrence of CMV pneumonitis was 28 of 54 (52%) if there were no HLA-DR matches, whereas the occurrence of CMV pneumonitis was 6 of 28 (21%) if there was at least one HLA-DR match (p = 0.0097). In contrast, when the recipient had not been exposed to CMV before transplantation (D⁺/R), the occurrence of CMV pneumonitis was 4 of 15 (25%) if there were no HLA-DR matches, and 3 of 8 (35%) if there was at least one HLA-DR match (p = NS). A different way of showing this relationship is by Kaplan-Meier analysis (Figure 5). The median time to CMV pneumonitis was 0.5 yr for D⁺/R⁺ and recipients with no HLA-DR matches compared with 3.2 yr for those with at least one HLA-DR match (p = 0.0199). In contrast, the median times to CMV pneumonitis did not differ between D⁺/R recipients who had either complete HLA-DR mismatches or had partial matches with their donors. Finally, mismatching at the HLA-A and -B loci showed no statistical association with CMV pneumonitis, nor did episodes of acute rejection or any other postoperative complication.

View larger version (15K): [in this window] [in a new window]   Figure 5.   Influence of mismatching between donor and recipient at the HLA-DR locus on freedom from CMV pneumonitis in recipients who had been exposed to CMV before transplantation (R+). Recipients who were CMV positive with at least one match at the HLA-DR locus (upper curve) had a higher probability (p [log-rank] = 0.0199) of freedom from CMV pneumonitis as compared with recipients who were CMV positive with no matches at the HLA-DR locus (lower curve).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study reports data from a single institution examining the influence of donor and recipient HLA locus mismatching on the development of obliterative bronchiolitis (OB) after lung transplantation. The overall incidence of OB in the study population was quite high. Thirty percent of patients developed OB by 2 yr after transplant and 56% developed OB by 3 yr, with a median time to onset of OB of 2.7 yr. The median time to onset of OB in the present study was similar to the range of 2.5-3.4 yr reported from other lung transplant centers (5, 20). Preliminary data from a 12-center multinational database reported a 52% incidence of OB at 3 yr after transplantation (21). The data in the present study, combined with previous reports, highlight the abbreviated time course to OB after lung transplantation, which is considerably shorter than the median time to chronic allograft rejection after heart and kidney transplantation (8 and 14 yr, respectively) (11, 12). In addition, the data highlight the importance of precise definition of OB on outcome. The median time to onset of OB, using only spirometric criteria, was longer because it did not include patients who had pathologic evidence of OB but who did not yet sustain a decline in FEV₁ of more than 20%.

The data in the present study reconfirmed previous reports that Grade A2 or A3 acute rejection, CMV pneumonitis, and HLA-A mismatching were significant risk factors for OB (5- 10). In contrast, mismatches between donors and recipients at the HLA-DR and HLA-B loci were not important variables determining OB. These observations appear to be at odds with our previous demonstration of the importance of mismatches between donors and recipients at the HLA-DR and HLA-B loci in determining early, high-grade rejections after lung transplantation (3). We interpret these findings to relate to a multifactorial etiology of OB. In this regard, bronchial epithelial injury (reperfusion, bronchial ischemia, aspiration) has been postulated to be a potent determinant of OB (5, 22). Although our data did not provide any significant correlation between clinical indices of reperfusion lung injury or bronchial ischemia and OB, these parameters may be too nonspecific to demonstrate a linkage between bronchial epithelial injury and OB. It has been demonstrated that bronchial epithelial injury leading to OB is driven by both immunologic and nonimmunologic mechanisms (23).

The data in the present study also demonstrated a significant relationship between HLA-DR locus matching and reduced risk of CMV pneumonitis in recipients who had been exposed to CMV before transplant. We interpreted these data to mean that the memory T cell response in recipients previously exposed to CMV could utilize the shared donor HLA-DR molecule to prevent donor-derived CMV replication in the transplanted lung, and consequently the development of CMV pneumonitis. In contrast, under immunosuppression naive T cells from previously unexposed recipients could not be educated to mount an immune response to CMV in the donor tissue even when shared HLA-DR determinants were present.

One limitation of the present analysis was that the study population was relatively small, and many trends in the data did not achieve significance. Nevertheless, the present study enjoyed many of the advantages of a single center study including uniformity of diagnostic criteria, bronchoscopic biopsies, immune suppression, pulmonary function laboratory, pathology interpretation, thorough record keeping, and complete follow-up in all patients. These advantages guarantee accuracy of the recorded data within the recognized limitation of relatively small numbers of patients.

The results of the predictive model for OB in the present study (Table 3) provide opportunities to reduce the risk of OB after lung transplantation. More effective therapies to reduce the rate of high-grade rejections and CMV pneumonitis will, it is hoped, delay the time to onset of OB. The approximate 2-fold higher frequency of OB for HLA-A mismatches in this study does not yet justify prospective matching of lung transplant recipients, as is currently done for kidney transplant recipients. Nevertheless, retrospective knowledge of HLA-A mismatching and prediction of OB risk may define high-risk patients who require augmented immune suppression or more vigorous surveillance for diagnosis and treatment of OB.