Published ahead of print on March 20, 2003, doi:10.1164/rccm.200212-1468OC
© 2003 American Thoracic Society
Airflow Obstruction after Myeloablative Allogeneic Hematopoietic Stem Cell TransplantationDivision of Clinical Research, Fred Hutchinson Cancer Research Center; and Department of Epidemiology, University of Washington, Seattle, Washington Correspondence and requests for reprints should be addressed to Dr. Jason W. Chien, M.D., Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., D3–190, P.O. Box 19024, Seattle, WA 98109–1024. E-mail: jchien{at}fhcrc.org
Despite advances in the management of myeloablative allogeneic hematopoietic stem cell transplants, airflow obstruction (AFO) remains a significant complication. We conducted a 12-year study to examine the recent epidemiology of AFO and its associated mortality. Using the rate of percent predicted FEV1 decline after transplant, we defined AFO as a more than 5% per year decline in percent predicted FEV1 with the lowest post-transplant FEV1/FVC ratio less than 0.8. New obstruction was more frequent than previous estimates (26% overall, 32% among patients with chronic graft-versus-host disease [GVHD]) and was significantly associated with older age at transplant, lower pretransplant FEV1/FVC ratio, history of both acute and chronic GVHD, and respiratory viral infection within the first 100 days after transplant. AFO was associated with significant attributable mortality rates of 9% at 3 years, 12% at 5 years, and 18% at 10 years after transplant, which were much higher for the subpopulation of patients with chronic GVHD (22% at 3 years, 27% at 5 years, and 40% at 10 years). These results suggest that the incidence of AFO may have been underestimated previously, and its presence significantly increases the mortality of long-term survivors of myeloablative allogeneic hematopoietic stem cell transplant patients.
Key Words: bronchiolitis obliterans • hematopoietic stem cell transplant • airflow obstruction Severe airflow obstruction (AFO) was first recognized in the early 1980s as a serious and often fatal complication of hematopoietic stem cell transplantation (HSCT) (1–4). Respiratory symptoms may include dry cough (60–100%), dyspnea (50–70%), and wheezing (40%) (1, 5, 6). New-onset AFO is the hallmark of this HSCT-related syndrome, otherwise known as bronchiolitis obliterans. The diagnosis is based primarily on spirometric measurements, provided that infection has been excluded (7). Other potentially useful studies include chest radiography, which may be normal or show hyperinflation, and high-resolution computed tomography, which may reveal areas of hypoattenuation, bronchial dilatation, bronchiectasis, and expiratory air trapping (5, 8–10). Bronchoalveolar lavage is usually nonspecific (11), and transbronchial lung biopsies are infrequently diagnostic because this process involves small distal airways. Open or video-assisted thoracoscopic lung biopsy, although rarely pursued, may demonstrate fibrinous obliteration of the lumen of the respiratory and membranous bronchioles (2, 3, 9, 12, 13). Most studies conducted to investigate new-onset AFO after HSCT have involved small, highly selected cohorts. The estimated incidence has ranged widely between 6 and 20% among long-term survivors of HSCT, partially because the definition of AFO in these studies varied (7). Among patients with chronic graft-versus-host disease (GVHD), the incidence has been estimated to be as high as 11% (5). Factors that have been associated with development of HSCT-related AFO include acute GVHD (14, 15), methotrexate use (16), hypogammaglobulinemia (17), and respiratory infections (13), but the strongest relationship is with chronic GVHD (6, 13–16, 18–20). Severe AFO has been associated with widely varying case-fatality rates ranging from 14 to 100% (5, 6, 9, 13, 21–26). It is not known whether AFO has a significant effect on mortality once other causes of death after HSCT have been accounted for. During the last decade, there have been significant advances in the field of HSCT. Hematopoietic stem cells from unrelated donors and matched or mismatched related donors are now more frequently used for treatment of a variety of hematologic disorders (27). Prophylaxis and treatment of GVHD have become more standardized with additional newer immunosuppressive agents (28). Standard prophylaxis with broad-spectrum antibiotics, antivirals, and antifungals has reduced the incidence of some respiratory infections (29). Despite these changes, HSCT-related AFO remains a significant complication for long-term survivors. A recent study assessing the existing diagnostic criteria (16) for HSCT-related AFO revealed that it was neither sensitive nor specific for identifying affected patients (30). Furthermore, the proposed alternatives performed no better. We performed a retrospective cohort study over a recent 12-year period to explore an alternative criteria for defining HSCT-associated AFO, update the estimated incidence and epidemiology of new-onset AFO after HSCT, and investigate the impact of AFO on mortality after HSCT. Some of the findings presented in this article have been reported previously in the form of an abstract (31).
Patient Population and Clinical Variables All patients who received their first myeloablative allogeneic HSCT at the Fred Hutchinson Cancer Research Center between January 1, 1990 and December 31, 2000 were considered for study. The last date of follow-up was December 31, 2001. Figure 1 lists the details regarding excluded patients. Clinical variables are listed in Table 1 . Disease risk for all hematologic malignancies at transplant was classified as low, intermediate, or high as published previously (32). Solid malignancies and nonhematologic diseases were classified as high risk. Donor type categories were related matched, related mismatched, and unrelated. Donor match status was determined according to donor–recipient ABO compatibility and HLA-A, HLA-B, and HLA-DR status. Respiratory virus infection (RVI) with parainfluenza, respiratory syncytial virus, or influenza during the first 100 days after HSCT was documented by either culture or direct fluorescent antibody staining of nasopharyngeal aspirates and/or bronchoalveolar lavage fluid (33, 34). Acute GVHD was graded based on stages of organ involvement using standard criteria (35, 36) and categorized as yes (Grades 3–4) or no (Grades 0–2) acute GVHD. The diagnosis and staging of chronic GVHD were established by using clinical, histologic, and laboratory criteria published previously (20, 37) and were categorized according to the presence or absence of clinical extensive chronic GVHD. Acute and chronic GVHD was then integrated and categorized as no acute or chronic GVHD, acute GVHD alone, de novo chronic GVHD (not preceded by acute GVHD), quiescent-onset chronic GVHD (preceded by acute GVHD that was followed by a period of quiescence), or progressive-onset chronic GVHD (preceded by acute GVHD with no period of quiescence) (20). Prophylaxis for GVHD included methotrexate, cyclosporine, and/or FK-506.
Pulmonary Function Tests All pulmonary function tests (PFTs) were performed at the Fred Hutchinson Cancer Research Center according to the American Thoracic Society guidelines (38) using the following equipment: the Gould 1001 (Gould, Inc., Dayton, OH) from January 1990 to June 1991, the Sensormedics 2100 (Sensormedics Co., Yorba Linda, CA) between July 1991 and August 1999, and the Sensormedics V-Max 22 with Autobox #6200 (Sensormedics Co.), between September 1999 and December 2001. Published equations for children and adults were used to determine predicted values of FEV1 and FVC (39, 40). If a bronchodilator was administered, the postbronchodilator values were used for analysis to minimize misclassification of reversible AFO. The PFT obtained immediately before HSCT was considered the baseline lung function. PFTs obtained within the first 6 months after HSCT were excluded because changes in pulmonary function during the first 6 months after transplant may be reversible and reflect temporary functional changes related to peritransplant events (41). All PFTs obtained after the initial 6 months were used to calculate an annualized rate of decline for the pFEV1 (see STATISTICAL METHODS). These were obtained routinely during long-term follow-up visits at 1 and 5 years after transplant or as otherwise indicated. Patients were defined as having significant AFO if the annualized rate of pFEV1 decline was more than 5% per year and the lowest documented post-transplant FEV1/FVC ratio was less than 0.8. The latter criterion was applied to prevent misclassification of patients with a restrictive pulmonary process. Patients whose pFEV1 decreased by more than 5% per year but had a FEV1/FVC ratio more than 0.8 were categorized as not having AFO. Patients were also classified according to the definition of AFO published previously, defined by a decrease of the FEV1/FVC ratio to less than 0.7 after transplant (16). The mortality cohort consisted of patients who had a 1-year follow-up PFT (365 ± 100 days after transplant) available. Survival time was determined from the 1-year PFT to the last documented date in the database (date of last correspondence with Fred Hutchinson Cancer Research Center, date of last PFT, date of death, or December 31, 2001).
Statistical Methods
Patient Characteristics and Definition of AFO There were 3,002 first myeloablative allogeneic HSCTs between January 1, 1990 and December 31, 2000. The excluded patients are described in Figure 1. The clinical characteristics and outcomes of the remaining 1,131 patients are summarized in Table 1. By applying the definition of AFO published previously to the cohort, 139 cases (12%) of AFO were identified (Figure 2A) . Although this definition was quite specific for identifying patients with AFO, it did not accurately identify patients who experienced a significant decline in airflow after HSCT. For instance, approximately 50% of the patients classified as having AFO by this definition had a modest decrease of the FEV1/FVC ratio by 0.1 or less, and approximately 5% of patients classified as not having AFO had a significant decrease of the ratio by 0.1 or more. This suggested that the FEV1/FVC ratio criterion did not accurately identify patients who experienced the most dramatic decline in airflow after transplant. For comparison, the annualized rate of decline in pFEV1 was calculated for the cohort (median rate of decline 2.8% ± 11.6 per year [SD]). Two hundred ninety-nine patients (26%) were categorized as having AFO defined by an annualized rate of decrease in pFEV1 by more than 5% per year with the lowest post-HSCT FEV1/FVC ratio less than 0.8 (Figure 2B). The median annualized decrease in pFEV1 among patients defined as not having and having AFO was 1.3% ± 7.8 and 10.4% ± 13.2, respectively (p < 0.001). In comparison with the criteria published previously, our proposed criteria were not only able to quantify the magnitude of the rate of airflow decline after HSCT but also able to identify a subset of patients whose airflow decline was the most dramatic and perhaps the most clinically meaningful. Therefore, we proceeded with further analyses using the annualized rate of decrease in pFEV1 by more than 5% per year with the lowest post-HSCT FEV1/FVC ratio less than 0.8 as the definition of AFO.
Clinical Characteristics Associated with Development of AFO Univariate analysis demonstrated that race, disease risk, donor match, stem cell source, cytomegalovirus serologic status, total body irradiation exposure, number of PFTs, GVHD prophylaxis, and year of transplant were not significantly associated with AFO. AFO was significantly associated with increasing age at transplant (  2 26.4, p < 0.001), pretransplant FEV1/FVC ratio (2 80.1, p < 0.001), GVHD category (2 25.3, p < 0.001), recipient–donor sex match (2 11.6, p = 0.009), and RVI during the first 100 days (2 10.5, p = 0.001). There was no statistically significant difference in the residual volume and lung diffusion capacity between the two groups. Table 2 summarizes the distribution of cases of AFO according to the category of GVHD. Two hundred and twenty-five of the cases of AFO (75%) occurred among patients with chronic GVHD (de novo, quiescent-onset, and progressive-onset chronic GVHD; Table 2). The majority of these cases of AFO (191 cases) occurred among patients with both acute and chronic GVHD (quiescent- and progressive-onset chronic GVHD). Among patients classified as having neither acute nor chronic GVHD, there were 63 cases of AFO. Only 1 of these 63 cases was truly GVHD-free. The remaining cases had some combination of lower grade acute (Grades 1 or 2) and chronic GVHD (subclinical or limited clinical).
Multivariate analysis of the clinical factors found to be associated with the development of AFO resulted in the adjusted RR depicted in Table 3 . Age at the time of transplant, pretransplant FEV1/FVC ratio, GVHD category, and a history of RVI each conferred a significant risk for the development of AFO. Age more than 60 at transplant resulted in the highest risk (RR 2.5, 95% CI 1.3–3.4). A pretransplant FEV1/FVC ratio less than 80% was significantly associated with AFO, with the risk highest among those with a ratio less than 70% (60–70%, RR 2.4, 95% CI 1.8–2.8; < 60%, RR 2.3, 95% CI 1.3–3.1). Whereas acute GVHD alone was not associated with an increased risk for developing AFO, chronic GVHD alone (RR 1.5, 95% CI 1.01–2.0), quiescent-onset (RR 1.6, 95% CI 1.3–2.4), and progressive-onset (RR 1.9, 95% CI 1.4–2.4) chronic GVHD were significantly associated with AFO. Recipient–donor sex match was no longer statistically significant. Inclusion of covariates not significantly associated with AFO in the univariate analysis did not significantly affect the risk estimates in the multivariable logistic analysis.
AFO and Mortality The presence of AFO 1 year after transplant was determined for 944 patients (mortality cohort, 83% of the original cohort), 270 (29%) of whom developed AFO by 1 year after HSCT. There was 91.1% agreement (  0.8, SE 0.03) between the original and mortality cohorts. Kaplan–Meier analysis demonstrated that AFO had a significant effect on mortality rates among all patients and among the subset with chronic GVHD (p < 0.001, Figure 3)
. Patients with AFO consistently had higher crude cumulative mortality rates than patients without AFO at 3, 5, and 10 years after transplant. These rates were also consistently higher for the subset of patients with chronic GVHD and AFO. The crude mortality rates attributable to AFO were 12% at 3 years, 14% at 5 years, and 21% at 10 years for all patients and 27% at 3 years, 18% at 5 years, and 29% at 10 years for the subset of patients with chronic GVHD. After the mortality rates were adjusted for pretransplant FEV1/FVC ratio, GVHD category, disease risk, disease relapse, and history of RVI, the attributable adjusted mortality rates for AFO among all patients were 9% at 3 years, 12% at 5 years, and 18% at 10 years. Among the subgroup of patients with chronic GVHD, the attributable adjusted mortality rates for AFO were much higher (22% at 3 years, 27% at 5 years, and 40% at 10 years).
Cox proportional hazard analysis for patients with AFO compared with patients without AFO, after adjusting for pretransplant FEV1/FVC ratio, age at transplant, GVHD categories, disease risk, disease relapse, and history of RVI, demonstrated a risk of 2.3 (95% CI 1.6–3.3) for mortality after HSCT (Table 4) . Subanalysis demonstrated that the risk for mortality was only significant among patients whose pFEV1 decreased by more than 10% per year (146 patients, hazard ratio [HR] 3.2, 95% CI 2.2–4.8, p < 0.001). As expected, age more than 60, progressive-onset chronic GVHD, underlying disease risk at transplant, underlying disease relapse, and history of RVI also increased the risk for mortality. The presence of AFO at baseline (FEV1/FVC ratio < 70%) did not significantly increase the risk for mortality. When the model mentioned previously was applied to the subpopulation of 561 patients with chronic GVHD (de novo, quiescent, and progressive type of onset), the HR for AFO was similar (HR 2.9, 95% CI 1.9–4.5, p < 0.001). Neither did sex match, donor match, stem cell source, cytomegalovirus serologic status, GVHD prophylaxis, total body irradiation exposure, and year of transplant significantly increase the mortality risk nor did further adjustment for these characteristics qualitatively affect the risk estimates presented in Table 4.
Despite the medical advances in the field of HSCT in the past two decades, new-onset AFO remains a significant transplant complication. In this large study, we have found that the true incidence of AFO after myeloablative allogeneic HSCT may have been underestimated by previous reports. Our study also found the development of AFO to be associated with increasing age at the time of transplant, evidence of AFO before transplant, the history of both acute and chronic GVHD, and a history of early RVI. Finally, we demonstrated that our proposed criteria for AFO have identified an increase in mortality attributable to AFO, particularly among patients with chronic GVHD. According to our data using the criteria published previously, it is unlikely that the incidence of AFO has increased significantly in the last two decades. It is more likely that the incidence is higher in our study because we used a different definition to define AFO. We believe that our rate of pFEV1 decline definition may more accurately identify patients who experienced a significant decline in their airflow after HSCT and that the true incidence of HSCT-related AFO has been underestimated in the past. The comparison between the two definitions demonstrated that the FEV1/FVC ratio definition misclassified a significant proportion of the population, often categorizing patients with a modest decline of the ratio (< 0.1) as having AFO. Our definition, although sensitive to development of AFO, was not so sensitive as to result in significant misclassification of normal individuals. Indeed, the mean rate of pFEV1 decline in a normal population is less than 1% per year, well over two SDs lower than our 5% criteria (44). The use of an annualized rate of decline in lung function is a well-established epidemiologic criterion for identifying subsets of a study population with rapid decline of lung function (45–47). This measure makes it possible to recognize patients with rapidly declining airflow, including some who do not meet criteria for AFO by other standards. The use of this definition to identify patients with rapidly declining airflow will be an important epidemiologic step to selecting appropriate cohorts at risk for AFO for studies of pathogenesis and early intervention trials. However, the utility of this definition for clinical decision making in individual patients will require further investigation. Previous studies that have attempted to examine the relationship of AFO with acute and chronic GVHD have had conflicting results. Although most of these studies have consistently found an association with chronic GVHD, the same studies have also found the relationship with acute GVHD to be either present (14, 15, 48, 49) or absent (5, 16, 19, 50). Only one small study has found that decline in the pFEV1 is associated with both acute and chronic GVHD (51). Our analysis provides further insight into the relationship between acute GVHD, chronic GVHD, and AFO. First, it is clear that chronic GVHD alone, but not acute GVHD alone, is a risk for developing significant AFO after transplant. Second, this analysis suggests that the risk for AFO differs slightly depending on the type of chronic GVHD onset. Finally, we have demonstrated that significant AFO can develop among patients with low-grade acute (grades 1–2) and low-grade chronic GVHD (subclinical and clinical limited) and that AFO almost never occurred in the complete absence of GVHD. These findings suggest that the relationship between AFO and GVHD is probably more complicated than was believed previously. Acute GVHD is believed to occur when donor T cells respond to alloantigens and attack host tissues (52), resulting in cytokine production by activated donor T cells, monocytes, and macrophages. Despite the well established association between acute and chronic GVHD, the pathogenesis of chronic GVHD, which has distinctive clinical and pathologic manifestations that mimic autoimmune disease, is probably different (52). The slight differences in risk for developing AFO suggests that acute GVHD, in the presence of chronic GVHD, may play a role in the pathogenesis of AFO, such that pre-existing acute GVHD may alter the risk for developing AFO. Case-fatality rates associated with the development of AFO after HSCT have ranged widely from 14 to 100% in previous studies (7, 10, 13). In an earlier study of allogeneic HSCT recipients conducted at our center, the crude 3-year mortality rate of patients with AFO was 65%, significantly higher than the 44% rate found in a comparison group of allogeneic marrow recipients without AFO. None of these studies investigated whether AFO had an adverse effect on mortality after adjusting for other common causes of death among long-term survivors of HSCT. The current study revealed that AFO increased the risk of mortality, an effect that was more pronounced among patients with chronic GVHD. This effect existed even after the most common causes of mortality among long-term survivors were considered in the analysis, particularly for patients with chronic GVHD, and persisted for at least 10 years after transplantation. Our study is not without limitations. First, the data, particularly the mortality data, are specific to long-term survivors of myeloablative allogeneic HSCT (survival to at least 1 year for the mortality data) at our institution and may not apply for other institutions. Second, our data might be biased because patients may be more likely to return for follow-up and more likely to have a PFT if there is a transplant-related complication such as GVHD. However, although the true incidence of AFO among all patients may be lower than 27%, our data likely represent an accurate assessment of the prevalence of AFO among patients with GVHD. Third, the increased risk associated with older age at transplant and a lower pretransplant FEV1/FVC ratio may be an overestimation because as lung function declines due to either increasing age or some underlying pulmonary disease, a smaller absolute decline in pFEV1 is required to meet the 5% per year criteria for AFO. Fourth, our study was not designed to identify the etiology of new-onset AFO. Patients with "severe AFO" as reported previously may have a different pathologic process than the milder form of AFO that is now being reported in our study. Despite accounting for viral infections before Day 100, other less common processes such as adult-onset asthma, chronic obstructive lung disease, bacterial or viral respiratory infections after Day 100, or nonspecific reactive airway disease could have resulted in a transient decline in the pFEV1. Although the cause of the AFO might have been determined by pursuing additional studies, these were not always performed, probably due to low clinical suspicion and differences in clinical practice. Incorporation of any such existent data would surely bias our analysis. Finally, although our mortality analysis does not take into account all potential risk factors for mortality after HSCT, we believe that our hazard model has accounted for the most common risk factors for death among long-term survivors of myeloablative allogeneic HSCT. Although severe infections are a common cause of death in this cohort, these tend to occur in the most immunosuppressed patients, which are accounted for by the GVHD variables in the model. In summary, our study has demonstrated that the incidence of new-onset AFO among myeloablative allogeneic HSCT patients, according to our definition, is probably much higher than suspected previously. Our findings also suggest that chronic GVHD alone and acute and chronic GVHD together, regardless of grade, are significant risk factors for developing post-transplant AFO. Finally, our study demonstrated that AFO is associated with increased mortality after HSCT even after controlling for other well recognized risk factors for mortality. Future studies of new-onset AFO after HSCT should consider the rate of decline in pFEV1 as a quantitative phenotype to identify specific groups of patients who might represent a population susceptible to this syndrome. These studies should include investigations that will determine the natural history of this syndrome, examine immunologic and genetic risk factors that may be involved in the development of both acute and chronic GVHD, the impact of AFO on quality of life after HSCT, and the response of AFO to immunosuppressive therapy.
The authors would like to thank Dr. Joachim Deeg for his insightful review of this manuscript.
Supported by National Institutes of Health grants T32 HL07287, PO1 CA 18029, HL36444, CA18221, 1R01 HL71914–01, 1K23HL69860–01, and American Lung Association of Washington Research Grant. Received in original form December 13, 2002; accepted in final form March 18, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
