Abstract
Azithromycin exposure during the early phase of allogeneic hematopoietic cell transplantation (HCT) has been associated with an increased incidence of hematological relapse. We assessed the impact of azithromycin exposure on the occurrence of relapse or new subsequent neoplasm (SN) in patients with bronchiolitis obliterans syndrome (BOS) after HCT, who are commonly treated with azithromycin alone or in combination with other agents. In a retrospective study of BOS patients from two large allograft centers, the effect of azithromycin exposure on the risk of relapse or SN was estimated from a Cox model with a time-dependent variable for treatment initiation. The Cox model was adjusted on time-fixed covariates measured at cohort entry, selected for their potential prognostic value. Similar models were used to assess the exposure effect on the cause-specific hazard of relapse, SN, and death free of those events. Sensitivity analyses were performed using propensity score matching. Among 316 patients, 227 (71.8%) were exposed to azithromycin after BOS diagnosis. The corresponding adjusted hazard ratio in patients exposed to azithromycin versus unexposed was HR = 1.51 (95% CI, 0.90 to 2.55) for relapse or SN, 0.82 (95% CI, 0.37 to 1.83) for relapse, and 2.00 (95% CI, 1.01 to 3.99) for SN. Patients exposed to azithromycin had a significantly lower cause-specific hazard of death free of neoplasm and relapse (adjusted HR= 0.54, 95%CI, 0.34 to 0.89). In conclusion, exposure to azithromycin after BOS after HCT was associated with an increased risk of SN but not relapse.
Keywords: pulmonary graft-versus-host disease, malignancy, relapse, hematopoietic stem cell transplant, bronchiolitis obliterans
Introduction
Recently, the European Medicines Agency (EMA) and the United States Food and Drug Administration (FDA) issued warnings against the long-term use of azithromycin in the setting of early allogeneic hematopoietic cell transplant (HCT) for hematological malignancies [1,2]. These warnings followed the results of ALLOZITHRO, a French randomized trial testing azithromycin as prophylaxis against lung chronic graft-versus host disease (GVHD) i.e. bronchiolitis obliterans syndrome (BOS) in recipients of allogeneic HCT [3]. The rationale for testing azithromycin in this population derives from the effect of azithromycin prophylaxis in reducing the incidence of BOS in lung transplant recipients [4]. ALLOZITHRO was terminated early due to an unanticipated reduction in survival attributed to increased rates of hematological relapse in patients who received azithromycin [3]. The mechanisms for relapse observed in the ALLOZITHRO trial are under investigation. It is hypothesized that azithromycin interferes with anti-tumor immune surveillance.
In light of these unexpected findings, the long-term use of azithromycin for the treatment of BOS after HCT has been called into question. Azithromycin is frequently used alone or as part of FAM (fluticasone, azithromycin, montelukast) treatment for established BOS [5,6], In the setting of HCT, BOS is usually diagnosed within the first 2 years after transplant [7] at a time when the risk for hematological relapse decreases and the risk of subsequent neoplasms (SN) increases [8,9].
The aim of this study was to determine if azithromycin treatment for BOS after allogeneic HCT is associated with an increased risk of cancer, including relapse of the original malignancy and SN.
Methods
Study cohort:
Patients with BOS, aged 18 and older, who survived at least 6 months posttransplant, from Fred Hutchinson Cancer Research Center (FHCRC)/Seattle Cancer Care Alliance in Seattle, Washington, and at the Hôpital St. Louis (SLS) in Paris, France, were included. Both sites have expertise in managing post-allogeneic HCT lung complications. Any patient who received an allogeneic HCT between 2000–2016 (FHCRC) or who was referred for clinical care between 2000–2017 (SLS) and met criteria for BOS diagnosis were included. BOS was defined by the following 2014 National Institutes of Health (NIH) spirometric criteria (forced expiratory volume in 1 second (FEV1) <75%, FEV1/vital capacity (VC) <0.7, ≥10% absolute FEV1 decline compared to pretransplant baseline)[10]. Absence of a bronchodilator response was not required for BOS diagnosis as this parameter was not uniformly available. Similarly, chest imaging was not used to determine BOS diagnosis as high-resolution studies were not available for many subjects. Chart review confirmed the absence of infectious diagnosis at the time of meeting spirometric criteria for BOS. Additionally, in the SLS cohort, as previously reported, BOS was also diagnosed in patients with a new-onset obstructive impairment characterized by a decrease in both FEV1 and VC, a normal FEV1/VC ratio, a normal total lung capacity (TLC) and elevated residual volume (RV) in the absence of alternative explanations for ventilatory impairment [11,12]. Prior to undergoing HCT, all subjects from both sites signed informed consent allowing the use of their clinical data for research. This study was approved by the appropriate institutional review boards for FHCRC and SLS.
Clinical variables:
Modified disease risk index (DRI) was used to assess disease risk at the time of transplant [13]. Acute GVHD was graded from 1 to 4, according to consensus criteria [14,15]. Chronic GVHD was graded according to the 2014 NIH consensus guidelines [10] if the clinical data were available by chart review. Primary cause of death was determined by consensus when multiple morbidities were reported (GSC and AB). Causes of death were classified as respiratory (respiratory failure and respiratory infection), death related to relapse or SN, transplant-related mortality, others and unknown.
Azithromycin exposure:
Exposure to azithromycin was defined as the use of azithromycin at any time posttransplant, including for intentional treatment of BOS or other reasons, for any duration and at any dose, with the date of BOS diagnosis representing the entry into the cohort. Periods of exposure were defined by courses of azithromycin intake with a defined start and end date with at least one day free of intake between courses. Time-dependent dynamics of azithromycin exposure are displayed with a random sample of 50 patients in Figure 1 and demonstrate the following issues: (1) Azithromycin may have been given before BOS diagnosis; (2) Azithromycin may have been given at BOS diagnosis; (3) In some instances there is a time interval between BOS diagnosis and azithromycin intake. Each observation ends with patient death or last follow-up. Clinical practice for BOS treatment evolved during the study period for both cohorts; the use of azithromycin was more common in the latter part of the study period.
Figure 1.
Graphical time-dependent display of 50 randomly selected patients. Each line shows one observed patient. To display the time-dependency of azithromycin treatment, time free of any azithromycin exposure is marked in grey and a time of azithromycin exposure is marked in color according to the indication. The corresponding survival event is marked with a filled circle (death) or a transparent circle (alive). BOS: bronchiolitis obliterans syndrome. For instance, patient 2 (second line at the bottom) died early free of any exposure to azithromycin. At the top (first line), patient 50 only received a short course of azithromycin for prophylaxis.
Outcomes:
The primary outcome was cumulative incidence of all cancers subsequent to BOS, including relapse of the original hematological disease and new SN. Secondary outcomes included cumulative incidence of relapse, of SN, and of death free of relapse and SN, and event-free survival (event being either relapse or SN). Data for clinical outcomes were locked on November 30, 2017.
Statistical methods:
To account for the time-dependent dynamics of the data, we used a multistate model with following states: BOS diagnosis, azithromycin treatment, and death (see online supplement figure S1). The effect of cumulative treatment exposure on the risk of relapse and/or SN after BOS diagnosis was estimated from a Cox model with time-dependent variable for treatment initiation. This provides unbiased estimates of the hazard ratio allowing for control of survival bias while avoiding selection bias [16,17]. The Cox model was adjusted on time-fixed covariates measured at the time of transplant that were selected as potentially of prognostic value: age, gender, tobacco use, total body irradiation (TBI), DRI, prior autologous HCT and past exposure to azithromycin after HCT but before BOS. We also introduced chronic GVHD after BOS as a time-dependent covariate. Models were finally stratified on the site to handle potentially different baseline hazards.
Similar modeling strategies were used to assess the exposure effect on the cause-specific hazard of relapse, of SN, and of death free of relapse and SN. On each end point, we further tested the interaction of exposure effect according to the site using the Gail and Simon test [18].
To account for the use of azithromycin as an intermediate event and death as a potential competing event on the effect of azithromycin on the outcomes, we displayed cumulative hazards [19].
To account for changes in clinical practices in prescribing azithromycin, sensitivity analyses were performed using propensity score matching. The propensity score of azithromycin administration was estimated using a multivariable logistic model, including 9 potential confounders for relapse or SN (namely, age, gender, DRI, prior graft, acute leukemia, myeloablative conditioning, antithymocyte globulin, chronic GVHD, and time from HCT to BOS). Estimates of propensity scores were pooled from 30 imputed datasets, obtained by multiple imputations with chained equations. Quality of the score was measured on standardized mean difference of confounders, and c-index of the model [20]. Then, 1:1 matching on the pooled propensity score was individually performed using the nearest neighbor method within a caliper of 0.20 standard deviation of the logit of the propensity score, with and without replacement [21]. Estimates of azithromycin exposure used generalized linear models to handle the matching, with inverse-probability weighting and design-based standard errors.
Summary statistics, namely median [interquartile range, IQR] and percentage, are reported. All tests are 2-sided, with P values less than .05 considered significant. Analyses were performed on R 3.5.1 software (https://www.R-project.org/).
Results
Description of the cohort.
A total of 316 patients with BOS were included in the study: 185 patients from FHCRC and 131 from SLS. Baseline characteristics of the cohort are reported in Table 1.
Table 1:
Characteristics of the study cohort at baseline and at BOS diagnosis according to azithromycin exposure after BOS diagnosis
| Total (n=316) | Azithromycin exposure | ||
|---|---|---|---|
| No (n=89) | Yes (n=227) | ||
| Baseline characteristics at HCT | |||
| Women, n (%) | 129 (40.8%) | 31 (34.8%) | 98 (43.1%) |
| Age, median (IQR), y | 48.6 [33.6–58.6] | 47.9 [36–59] | 48.7 [33–57.6] |
| Diagnosis, n (%) | |||
| Acute leukemia | 140 (44.3%) | 35 (39.3%) | 105 (33.5%) |
| Chronic myeloid leukemia | 32 (10.1%) | 12 (13.5%) | 20 (8.8%) |
| Other myeloproliferative disorders | 8 (2.5%) | 3 (3.4%) | 5 (2.2%) |
| Myelodysplastic disorders | |||
| Lymphoid malignancies | 52 (16.4%) | 20 (22.5%) | 32 (36.6%) |
| Others | 73 (23.1%) | 16 (18%) | 57 (25.11%) |
| 11 (3.5%) | 3 (3.4%) | 8 (3.5%) | |
| Disease risk index at HCT, n (%) | |||
| Low | 47 (14.9%) | 18 (21.7%) | 29 (13.3%) |
| Intermediate | 176 (55.7%) | 43 (51.8%) | 133 (61%) |
| High | 68 (21.5%) | 19 (22.9%) | 49 (22.5%) |
| NA | 25 (7.9%) | 9 (10.1%) | 16 (7%) |
| History of smoking, n (%) | 122 (38.6%) | 45 (51.1%) | 77 (33.9%) |
| History of solid cancer prior to HCT, n (%) | 31 (9.8%) | 4 (4.5%) | 27 (11.9%) |
| Prior autologous HCT, n (%) | 96 (30.4%) | 19 (21.3%) | 77 (33.9%) |
| Donor type, n (%) | |||
| Related | 138 (43.7%) | 42 (47.2%) | 96 (42.3%) |
| Haploidentical | 5 (1.6%) | 2 (2.2%) | 3 (1.3%) |
| Unrelated HLA-match# | 130 (41.1%) | 32 (36%) | 98 (43%) |
| Unrelated HLA-mismatch## | 36 (11.4%) | 9 (10.1%) | 27 (11.9%) |
| Donor/recipient sex, n, (%)* | |||
| Male/male | 91 (29.5%) | 28 (32.9%) | 63 (28.3%) |
| Male/female | 57 (18.5%) | 16 (18.8%) | 41 (18.4%) |
| Female/male | 92 (29.9%) | 27 (31.8%) | 65 (29.1%) |
| Female/female | 68 (22.1%) | 14 (16.5%) | 54 (24.2%) |
| Source of stem cells graft, n (%) | |||
| Peripheral blood | 273 (86.4%) | 74 (83.1%) | 199 (87.7%) |
| Bone marrow | 36 (11.4%) | 11 (12.4%) | 25 (11%) |
| Cord blood | 7 (2.2%) | 4 (4.5%) | 3 (1.3%) |
| Conditioning regimen, n (%) | |||
| Myeloablative | 180 (57%) | 47 (52.8%) | 133 (58.6%) |
| Nonmyeloablative | 136 (43%) | 42 (47.2%) | 94 (41.4%) |
| TBI | 159 (50.3%) | 33 (37.1%) | 126 (55.5%) |
| Dose, median (IQR), Grays | 2(2–12) | 2 (2–12) | 2(2–12) |
| Antithymocyte globulin### | 39 (12.3%) | 14 (15.7%) | 25 (11.0%) |
| Lung function | |||
| FEV1 (% predicted), median (IQR), | 92.8 [82.8–100.2] | 91.1 [83.1–97.9] | 92.9 [82.7–100.7] |
| FVC (% predicted), median (IQR) | 95.1 [85.9–105.6] | 92.0 [83.9–103.7] | 96.2 [86.5–106.0] |
| Characteristics at BOS diagnosis | |||
| Months from HCT, median (IQR), | 16.8 [10.8–30.6] | 13.9 [10.7–24.5] | 18.1 [11–33.5] |
| Chronic GVHD | |||
| Before BOS | 273 (86.4%) | 76 (85.4%) | 197 (86.8%) |
| At or after BOS, n (%) | 33 (10%) | 10 (11%) | 23 (10%) |
| Grade max | |||
| Mild | 25 (7.9%) | 13 (14.8%) | 12 (5.4%) |
| Moderate | 117 (37%) | 33 (37.5%) | 84 (37.8%) |
| Severe | 161 (50.9%) | 39 (44.3%) | 122 (54.9%) |
| Death free of chronic GVHD | 5 (1.6%) | 2 (2.2%) | 3 (1.3%) |
| Absence of chronic GVHD | 5 (1.6%) | 1 (1.1%) | 4 (1.8%) |
| Lung function | |||
| FEV1 (% predicted), median (IQR), | 55.9 [42.9–65.5] | 60.4 [51.1–68.8] | 53.2 [40.4–63.7] |
| FVC (% predicted), median (IQR) | 71.9 [64.0–83.5] | 72.9 [65.5–83.6] | 71.3 [63–83.5] |
| FEV1/FVC, median (IQR) | |||
| RV (% predicted), median (IQR) | 0.63 [0.51–0.69] | 0.67 [0.60–0.73] | 0.61 [0.48–0.67] |
| 131.7 [103.0–161.3] | 134.8 [113.9–161.1] | 130.8 [100.8–161.3] | |
| Treatments for BOS†, n (%) | |||
| Azithromycin | 197 (62%) | 0 (0%) | 197 (87%)‡ |
| Systemic steroids | 105 (33%) | 16 (18%) | 89 (39%) |
| ICS/LABA | 153 (48%) | 44 (49%) | 109 (48%) |
| ICS without LABA | 154 (49%) | 18 (20%) | 98 (43%) |
| Montelukast | 146 (46%) | 7 (8%) | 139 (62%) |
Excluding cord blood recipients, and 1 missing value for a recipient of a peripheral blood stem cell graft
Treatments administered for BOS at the time of diagnosis or thereafter; of the 227 patients who received azithromycin after the diagnosis of BOS, 197 received it specifically for the treatment of BOS and 30 for another indication (antimicrobial prophylaxis or treatment of an infection)
A total of 95 patients (30%) received the azithromycin in the FAM regimen (fluticasone, azithromycin, montelukast).
74 patients minimal 8/8 HLA-match and 56 patients 10/10 HLA-match
34 patients 7/8 HLA-match and 2 patients 9/10 HLA-match
no patients had HCT with ex vivo depletion of donor T-cells
IQR: interquartile range; HCT: hematopoietic cell transplant; FEV1: forced expiratory volume in 1 second; FVC: forced vital capacity; GVHD: graft-versus-host disease; ICS/LABA: inhaled corticosteroids/long-acting beta-agonist; ICS: inhaled corticosteroids; HLA: human leukocyte antigen; NA: not available
The median time to BOS diagnosis after HCT was 16.8 months [IQR 10.8 to 30.6]. The description of the patients according to the transplant center (FHCRC vs. SLS) is reported in the online supplement, (Table S1, Figure S2, Figure S3).
Azithromycin exposure.
Overall, 237 patients (75%) received azithromycin during their follow-up for one to 3 courses. Characteristics of both azithromycin-exposed and unexposed cohorts are summarized in Table 1. The description of azithromycin exposure any time after HCT is summarized in Table 2. The median length of exposure to azithromycin after BOS diagnosis was 16 months [IQR 7 to 36].
Table 2:
Description of azithromycin exposure any time after HCT for 237 patients*
| Number of azithromycin exposures per patient† | N (%) |
|---|---|
| 1 | 165 (52.2%) |
| 2 | 60 (19%) |
| 3 | 12 (3.8%) |
| Characteristics of 1st exposure | |
| Duration (days, median, IQR) | 230 [7;847] |
| Before BOS | 66 (20.9%) |
| At or after BOS | 167 (52.9%) |
| Both before and after BOS | 4 (1.3%) |
| Characteristics of 2nd exposure | |
| Duration (days, median, IQR) | 244.5 [90;884.8] |
| Before BOS | 7 (2.2%) |
| At or after BOS | 64 (20.2%) |
| Both before and after BOS | 1 (0.32%) |
| Characteristics of 3rd exposure | |
| Duration (days, median, IQR) | 191.5 [69.2;783.2] |
| Before BOS | 0 (0%) |
| At or after BOS | 12 (3.8%) |
| Both before and after BOS | 0 (0%) |
10 patients received azithromycin only before BOS for a median time of 7 days among the 237 patients who received azithromycin during the study period;
a single exposure of azithromycin was defined by azithromycin intake with a start and end date; the minimum duration of the 1st exposure after BOS was 7 days. A subsequent azithromycin exposure was considered if there was at least one day free of intake between two courses of azithromycin.
BOS: bronchiolitis obliterans syndrome; IQR: interquartile range
Cancer outcomes.
The median time of follow-up after BOS was 41 months [IQR 17 to 95]. There were 53 (16.8%) patients who relapsed; 30 relapses were documented after BOS (Table 3).
Table 3:
Type of hematological relapses and subsequent neoplasms after BOS diagnosis according to prior exposure to azithromycin
| Azithromycin exposure | ||
|---|---|---|
| No n=89 | Yes n=227 | |
| Hematological relapses type after BOS, n (%) | 10 (11%) | 20 (9%) |
| Acute leukemia | ||
| Chronic myeloid leukemia | 3 | 5 |
| Lymphoid malignancies | 1 | 2 |
| Myelodysplastic disorders | 4 | 8 |
| Others | 2 | 5 |
| 0 | 0 | |
| Subsequent neoplasms*, n (%) | 8 (9%) | 35 (15%) 33 after azithromycin† |
| Types | ||
| Squamous cell carcinoma | 3 (37.5%) | 21 (60%) |
| Adenocarcinoma | 2 (25%) | 8 (22.9%) |
| Verrucous carcinoma | 1 (12.5%) | 0 |
| Carcinoma NOS | 0 (0%) | 2 (5.7%) |
| Bowen’s disease | 0 (0%) | 2 (5.7%) |
| Malignant melanoma | 0 (0%) | 2 (5.7%) |
| Lymphoma | 1 (12.5%) | 0 |
| Mast cell leukemia | 1 (12.5%) | 0 |
| Anatomic Site | ||
| Breast | 0 (0%) | 3 (10.7%) |
| Gut | 1 (16.7%) | 3 (10.7%) |
| Skin | 2 (33.3%) | 14 (50%)† |
| Oral cavity | 1 (16.7%) | 7 (20%) |
| Pancreas | 1 (16.7%) | 0 |
| Penis | 0 (0%) | 1 (3.6%)† |
| Prostate | 1 (16.7%) | 3 (10.7%) |
| Uterus | 0 (0%) | 2 (5.7%) |
| Blood | 1 (12.5%) | 0 (0%) |
| Lung | 1 (16.7%) | 1 (3.6%) |
| Brain | 0 (0%) | 1 (3.6%) |
Basal cell carcinomas excluded;
2 malignancies occurred after BOS but before azithromycin onset, including 1 malignant melanoma of the upper extremity and 1 squamous cell carcinoma of the penis. BOS: bronchiolitis obliterans syndrome; NOS: not other specified
Median time from first azithromycin exposure after BOS to relapse was 15.6 months [IQR 8.5–36.3]; from transplant to relapse was 41.8 months [IQR 23.1–60.0]. Excluding basal cell carcinomas, 43 (13.6%) patients developed a SN after BOS, including 10 patients who did not receive azithromycin prior to the diagnosis of SN, and 33 in patients previously exposed to azithromycin. Among these 43 patients with SN after BOS, 18 developed more than one subsequent malignancy, including only 2/10 free of any azithromycin exposure and 16/33 after azithromycin exposure (p=0.15). The median time of developing a SN after azithromycin exposure was 43.5 months [IQR 8.5–36.3]. Median time from transplant to SN was 81 months [IQR 44.7–117.5]. Twenty-four (56%) of SN were of squamous cell histology (Table 3). The type of hematological relapse and SN according to sites are reported in the online supplement Table S2, as well as details on the duration of follow-up in each group of patients (Table S3). There was no evidence of any difference in relapses/SN according to the association with fluticasone and montelukast (p=0.42 by the exact Fisher test).
The cumulative hazards for relapse or SN are displayed in Figure 2.
Figure 2.
Cumulative hazard of relapse and/or subsequent neoplasm, either overall (Fig. 2a), or for relapse (Fig. 2b) and subsequent neoplasm (Fig. 2c), separately, and death free of relapse and subsequent neoplasm (Fig. 2d). Cumulative hazard can be interpreted as the probability of failure at time t given survival until time t. Note that when the cumulative hazard function is a straight line (such as in Fig.2a, for the non-exposed group), the underlying hazard function is constant over time.
The corresponding unadjusted hazard ratio (azithromycin-exposed vs. unexposed) are 1.49 (95%-CI, 0.91 to 2.44) for relapse or SN, 0.90 (95%-CI, 0.42 to 1.93) for relapse, and 1.88 (95%-CI, 1.00 to 3.51) for SN. The effect of azithromycin exposure on hazard of malignancy was confirmed in adjusted multivariate models (HR 2.00 [95%CI, 1.01 to 3.99, p=0.048], Table 4), adjusting for factors associated with the occurrence of relapse or SN or both after BOS on univariable analyses (see online supplement Table S4). Further adjusting on year of allograft, indication for the graft (distinguishing AML from other diagnoses), TBI dose, the use of a myeloablative regimen, and antithymocyte globulin did not modify these findings (see online supplement Table S5).
Table 4:
Estimation of azithromycin effect using multivariable time-dependent Cox models
| Relapse or subsequent neoplasm | Relapse | Subsequent neoplasm | ||||
|---|---|---|---|---|---|---|
| HR (95%CI) | P value | HR (95%CI) | P value | HR (95%CI) | P value | |
| AZM exposure | 1.51 (0.90 to 2.55) |
0.12 | 0.82 (0.37 to 1.83) |
0.63 | 2.00 (1.01 to 3.99) |
0.048 |
| TBI | 1.17 (0.62 to 2.23) |
0.62 | 0.55 (0.22 to 1.40) |
0.21 | 1.96 (0.90 to 4.27) |
0.091 |
| DRI | 1.64 (1.20 to 2.24) |
0.002 | 2.07 (1.35 to 3.16) |
0.0008 | 1.53 (1.02 to 2.29) |
0.038 |
| cGVHD-t | 0.86 (0.34 to 2.15) |
0.74 | 1.28 (0.37 to 4.41) |
0.69 | 0.65 (0.18 to 2.37) |
0.52 |
| Age | 1.00 (0.98 to 1.02) |
0.91 | 0.99 (0.96 to 1.02) |
0.40 | 1.02 (0.99 to 1.04) |
0.14 |
| Sex/Male | 1.07 (0.64 to 1.79) |
0.79 | 1.06 (0.46 to 2.46) |
0.89 | 0.94 (0.50 to 1.77) |
0.86 |
| History of smoking | 1.28 (0.74 to 2.23) |
0.38 | 1.45 (0.64 to 3.33) |
0.38 | 1.08 (0.55 to 2.13) |
0.82 |
| Prior autologous HCT | 1.99 (1.03 to 3.87) |
0.041 | 4.24 (1.61 to 11.1) |
0.003 | 1.37 (0.68 to 2.77) |
0.38 |
| AZM exposure before BOS | 0.59 (0.30 to 1.17) |
0.13 | 0.85 (0.31 to 2.36) |
0.76 | 0.53 (0.23 to 1.22) |
0.14 |
HR: Hazard ratio; CI: confidence interval; TBI: Total body irradiation; ATG: antithymocyte globulin; DRI: disease risk index; cGVHD-t: time dependent occurrence of chronic graft-versus-host-disease; HCT: Hematopoietic cell transplantation; AZM: azithromycin; BOS: bronchiolitis obliterans syndrome; HR: hazard ratio; CI: confidence interval
Sensitivity analysis considering only extensive chronic GVHD rather than overall chronic GVHD and adding time from HCT to BOS diagnosis in the adjusted multivariate models confirmed the azithromycin exposure effect on the occurrence of SN (HR 2.03, 95%CI 1.01 to 4.08, p=0.047) (see online supplement Table S6). There was no effect of cumulative months of exposure to azithromycin before the occurrence of relapse or SN; restriction of azithromycin exposure to greater than 7 days, or greater than 28 days (adjusted HR=2.05, 95%CI, 1.01 to 4.19, P=.047) did not modify the hazard of malignancy in exposed patients compared to unexposed, likely due to the duration of exposure of at least 7 months in 75% of the exposed patients.
There was no evidence of any azithromycin exposure by site interaction (FHCRC or SLS) on adjusted estimates, with p-values of Gail and Simon interaction tests (p=0.38 for relapse or malignancy; p=0.18 for relapse; p=0.78 for subsequent malignancy) (see online supplement Figure S4 and Figure S5). That is, the effect of azithromycin appeared to be the same for patients treated at FHCRC and those treated at SLS.
Survival outcomes.
Since death is a competing event for SN or relapse, the effect of azithromycin exposure after BOS on deaths free of relapse and/or SN were examined. One hundred and twenty (38%) subjects had died as of last follow-up including 32 patients who died of relapse and 7 of cancer (Table 5).
Table 5:
Causes of death
| Primary cause of death (n=120) | N (%) |
|---|---|
| Respiratory causes | 54 (45%) |
| Respiratory failure* | 37 (30.8%) |
| Respiratory infection | 17 (14.2%) |
| Relapse/SN causes | 32 (26.7%) |
| Relapse | 25 (20.8%) |
| SN | 7 (5.8%) |
| Transplant-related mortality | 8 (6.7%) |
| Others | 12 (10%) |
| Unknown cause | 14 (11.7%) |
6 patients died of multiple causes after lung transplantation; SN: subsequent neoplasm
Besides relapse, respiratory causes (n=54, 45%) were the most common primary causes of death. Regarding the effect of azithromycin exposure, the hazard of death free of malignancy and relapse was not modified, either unadjusted or adjusted for prognostic variables, compared to unexposed patients (HR=0.69, 95%CI, 0.42 to 1.12, P= .13; HR=0.71, 95%CI, 0.43 to 1.19, P=0.19, respectively). Patients exposed to azithromycin had a significantly decreased cause-specific hazard of death free of malignancy (adjusted HR= 0.54, 95%CI, 0.34 to 0.89, P= .014; Figure 2d) while their cause-specific hazard of death free of relapse was not significantly decreased (adjusted HR=0.62, 95%CI, 0.38 to 1.03, P= .06). Cause-specific hazard of death from all respiratory causes was similar in both groups of patients (HR=1.01, 95%CI, 0.68 to 1.51, P= .96), as well as that of death from respiratory failure (HR= 1.13, 95%CI, 0.68 to 1.86, P= .64) and from respiratory infection (HR=0.84, 95%CI, 0.43 to 1.64, P= .61).
Sensitivity Analyses.
A propensity score for receiving azithromycin was estimated; this differed between exposed and unexposed patients as measured by the c-index of the model (at 0.66) and the standardized mean differences (SMD) of the potential confounders across the treatment groups with an average value at 0.17. Of the 227 patients who received azithromycin, 76 (33%) could be matched without replacement, up to 225 (99%) were matched when replacement was allowed. Balance was improved with an average SMD and c-index at 0.076 and 0.499, respectively without replacement, and at 0.10 and 0.504 with replacement. Estimates of exposure effects confirmed previous results, with an increased risk of SN (although based on the sample of 76 treated and 76 untreated this was not statistically significant), and no increased risk of relapse (Table 6).
Table 6:
Estimation of azithromycin (AZM) effect using matched samples on propensity score to receive azithromycin
| Relapse or subsequent neoplasm | Relapse | Subsequent neoplasm | ||||
|---|---|---|---|---|---|---|
| AZM exposure | OR (95%CI) | P value | OR (95%CI) | P value | OR (95%CI) | P value |
| With replacement | 1.21 (0.77–1.91) | 0.42 | 0.78 (0.42–1.43) | 0.43 | 2.25 (1.19–4.25) | 0.013 |
| Without replacement | 2.12 (0.67–3.07) | 0.36 | 0.88 (0.30–2.58) | 0.81 | 2.21 (0.78–6.32) | 0.14 |
OR: odd ratio; CI: confidence interval
Discussion
In this analysis of a large multi-site cohort of patients with BOS after HCT, exposure to azithromycin after BOS diagnosis was associated with an increased risk of developing a SN but not with risk for relapse of the original malignancy. These results were independent of chronic GVHD status and were further confirmed for each site independently. In the ALLOZITHRO trial, azithromycin was given early posttransplant when the risk of relapse is inherently high. In the current study, azithromycin was given many months to years after HCT for established BOS when the risk of relapse diminishes and the risk of SN increases in association with chronic GVHD and prolonged immunosuppressive treatment [22]. In this context, an increase in relapse associated with azithromycin would have required a much larger cohort for analysis, but an increased risk of SN associated with azithromycin is consistent with the natural history of long-term survivors. Thus, these results constitute a second signal suggesting the potential association of azithromycin with cancer in allogeneic HCT.
Antibiotic use has previously been associated with various cancers [23]. Azithromycin may directly interfere with antitumor immune-surveillance through an inhibitory effect on various cell types, including lymphocytes, dendritic cells and natural killer cells in a dose-dependent way [24–26]. Long-term, low-dose azithromycin was shown to be associated with downregulation of genes regulating antigen presentation, interferon and T-cell responses, and numerous inflammatory pathways in patients with neutrophilic chronic obstructive pulmonary disease [27]. There is growing evidence that antibiotics may alter immune functions that are important for surveillance and control of malignancy through the modification of gut microbiota [28] which composition was shown to be associated with tumorigenesis [29,30]. In the specific setting of HCT, alterations in gut microbiota within the first month following HCT was associated with both incidence and severity of GVHD and hematological relapse [31,32]. These data may explain the increase of relapse found in the ALLOZITHRO trial where patients received azithromycin prior to and during engraftment, which is an immunologically vulnerable period for the control of the hematological malignancy [33,34]. Similar mechanisms may be involved in the development of SN associated with azithromycin exposure later in the course of survivorship.
Long-term immunosuppressive treatments are known to be associated with the development of SN. Unfortunately, because of the retrospective design of our study, which included a long period of patient follow-up, complete data on immunosuppressive therapy for GVHD was not available. However, when taking into account in the statistical model the severity of GVHD, a reflection of the intensity of immunosuppressive treatment, the effect of azithromycin on SN persists. We could not identify whether patients with cancer predisposition syndromes were included in our cohort of patients. Notably, however, the association of azithromycin with SN also persists when adjusted for DRI, which likely reflects a predisposition to SN.
Patients who received azithromycin developed more SN but very few died of their SN. Most SN in azithromycin-exposed patients after BOS belonged to an intermediate prognostic group of malignancies [35]. We postulate that the duration of follow-up after cancer diagnosis may not have been long enough to assess the effect of the malignancy on mortality.
The paradoxical finding of more SN but decreased cause-specific hazard of death free of malignancy with azithromycin exposure may result from attenuating the progression of BOS, or by reducing the number of respiratory exacerbations related to the underlying obstructive lung disease. A steroid-sparing effect of azithromycin [5] may reduce infections or other life-limiting complications related to chronic corticosteroid use. Similarly, a recent retrospective analysis demonstrated a survival benefit of extracorporeal photopheresis in patients with BOS after HCT in the absence of an impact on FEV1 [36]. Our retrospective data did not allow for a full exploration of these hypotheses, as some long-term survivors are managed outside of the transplant center for non-malignancy related concerns and intercurrent infectious events are not comprehensively captured. Furthermore, the full effect of azithromycin on the respiratory function of patients with BOS was beyond the scope of our current analysis.
In addition to the limitations noted above, this study is limited by the retrospective design. However, a retrospective cohort analysis is the only way to address the serious concerns about azithromycin arising from the ALLOZITHRO trial and the EMA and FDA warnings in a timely fashion. This study represents the largest cohort of BOS patients analyzed to date, although the overall cohort remains relatively small due to the rarity of this complication. The study was designed to look specifically at azithromycin exposure after any designation of BOS. Despite differences in clinical practice between FHCRC and SLS, the effect of azithromycin was similar at each site, which further adds to the strength of our analysis. The era of transplant experienced by the study subjects span two decades, during which time transplant practices and indications have changed, although we did standardize disease risk and GVHD grading as much as possible. Significantly, the results were not modified when the year of transplant is included in the models. Nevertheless, as with any predictive analysis from observational cohort data, residual unobserved confounders (for instance in the reasons for administering azithromycin) cannot be excluded in our study.
We also recognized the potential for immortal time bias in our cohort, in which the exposed group may have an inherent survival bias, when comparing the effect of azithromycin with varying exposures over time. To address this bias, we utilized a cohort design with time-dependent Cox models in which the estimated hazard ratio represents the adjusted incidence rate ratio. This approach compares the risk of an event between exposed and non-exposed patients at each event time, and re-evaluates to which risk group each person belonged based on whether there had been an exposure by that time, and tends to result in estimates with lower bias and greater precision compared with a nested case–control design [16,17]. Given the potential confounding by indication due to observational data, causal inference methods based on propensity score matching were further used as sensitivity analyses. This confirmed an increased occurrence of SN in the azithromycin group; note that it was no longer statistically significant when replacement was not allowed, likely due to a lack of power given the limited sample size.
In the light of these results, a careful assessment of the potential risks and benefits should be performed for each patient with BOS to determine whether azithromycin treatment should be prescribed. The increased risk of SN must be weighed against the potential benefit of chronic exposure. Azithromycin has been generally considered by most practitioners to be safe and is in widespread use for respiratory infections and various chronic respiratory diseases. In the setting of HCT, although azithromycin has become standard of care for BOS treatment at many centers, robust data supporting its efficacy in ameliorating lung dysfunction are lacking. As in other chronic lung diseases, long-term use may reduce infectious morbidity, however this has never been demonstrated in BOS after HCT. In addition to known cardiovascular and hearing loss side effects, concerns regarding antibiotic resistance and changes in microbiome diversity related to chronic azithromycin use and its consequences have emerged [37]. Additional studies are needed both to determine whether azithromycin is beneficial for patients with BOS after HCT and to elucidate the underlying mechanisms of azithromycin-associated cancers.
Supplementary Material
Highlights.
Azithromycin is associated with increased second cancers in bronchiolitis obliterans
Azithromycin is not associated with increased relapse in bronchiolitis obliterans
Decreased cause-specific hazard of death free of malignancy with azithromycin exposure
ACKNOWLEDGEMENTS:
The authors wish to acknowledge Chris Davis, Gary Schoch, Jesse Hubbard, Kevin Bray, Aaron W, Johnson, Emmanuelle Bugnet, Elisabeth Grall and Stéphane Cassonnet for assistance in acquiring data. In addition, we thank Kelly Thibodeau for assistance with the manuscript.
Financial disclosure: The work on the Fred Hutchinson Cancer Research Center cohort was funded in part by grants from National Institutes of Health and the National Cancer Institute (CA018029 and the Cancer Center Support Grant P30 CA015704).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict-of-interest statement: The authors declare no competing financial interests.
REFERENCES
- 1.Rain J, Spooner A. Pharmacovigilance Risk Assessment Committee (PRAC). European Medicines Agency [Internet]/ 2018. May 17 EMA/PRAC/288660/2018. Available from: https://www.ema.europa.eu/documents/minutes/minutes-prac-meeting-9-12-april-2018_en.pdf. Accessed 28 February 2019.
- 2.Food and Drug Administration. FDA warns about increased risk of cancer relapse with long-term use of azithromycin (Zithromax, Zmax) antibiotic after donor stem cell transplant. Issued August 3, 2018. https://www.fda.gov/Drugs/DrugSafety/ucm614085.htm. Accessed 28 February 2019.
- 3.Bergeron A, Chevret S, Granata A, Chevallier P, Vincent L, Huynh A, Tabrizi R, Labussiere-Wallet H, Bernard M, Chantepie S, Bay JO, Thiebaut-Bertrand A, Thepot S, Contentin N, Fornecker LM, Maillard N, Risso K, Berceanu A, Blaise D, Peffault de La Tour R, Chien JW, Coiteux V, Socié G; ALLOZITHRO Study Investigators. Effect of Azithromycin on Airflow Decline-Free Survival After Allogeneic Hematopoietic Stem Cell Transplant: The ALLOZITHRO Randomized Clinical Trial. JAMA 2017;318:557–566. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Vos R, Vanaudenaerde BM, Verleden SE, De Vleeschawer SI, Willems-Widyastuti A, Van Raemdonck DE, Schoonis A, Nawrot TS, Dupont LJ, Verleden GM. A randomised controlled trial of azithromycin to prevent chronic rejection after lung transplantation. Eur Respir J 2011;37:164–172. [DOI] [PubMed] [Google Scholar]
- 5.Williams KM, Cheng GS, Pusic I, Jagasia M, Burns L, Ho VT, Pidala J, Palmer J, Johnston L, Mayer S, Chien JW, Jacobsohn DA, Pavletic SZ, Martin PJ, Storer BE, Inamoto Y, Chai X, Flowers MED, Lee SJ. Azithromycin, and Montelukast Treatment for New-Onset Bronchiolitis Obliterans Syndrome after Hematopoietic Cell Transplantation. Biol Blood Marrow Transplant 2016;22:710–716. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Meyer KC, Raghu G, Verleden GM, Corris PA, Aurora P, Wilson KC, Brozek J, Glanville AR, Committee IAEBTF, Committee IAEBTF. An international ISHLT/ATS/ERS clinical practice guideline: diagnosis and management of bronchiolitis obliterans syndrome. Eur Respir J 2014;44:1479–1503. [DOI] [PubMed] [Google Scholar]
- 7.Bergeron A, Cheng GS. Bronchiolitis Obliterans Syndrome and Other Late Pulmonary Complications After Allogeneic Hematopoietic Stem Cell Transplantation. Clin Chest Med 2017;38:607–621. [DOI] [PubMed] [Google Scholar]
- 8.Boyiadzis M, Arora M, Klein JP, Hassebroek A, Hemmer M, Urbano-Ispizua A, Antin JH, Bolwell BJ, Cahn JY, Cairo MS, Cutler CS, Flowers ME, Gale RP, Herzig R, Isola LM, Jacobsohn DA, Jagasia MH, Klumpp TR, Lee SJ, Petersdorf EW, Santarone S, Spellman SR, Schouten HC, Verdonck LF, Wingard JR, Weisdorf DJ, Horowitz MM, Pavletic SZ. Impact of Chronic Graft-versus-Host Disease on Late Relapse and Survival on 7,489 Patients after Myeloablative Allogeneic Hematopoietic Cell Transplantation for Leukemia. Clin Cancer Res 2015;21:2020–2028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Danylesko I, Shimoni A. Second Malignancies after Hematopoietic Stem Cell Transplantation. Curr Treat Options Oncol 2018;19:9. [DOI] [PubMed] [Google Scholar]
- 10.Jagasia MH, Greinix HT, Arora M, Williams KM, Wolff D, Cowen EW, Palmer J, Weisdorf D, Triester NS, Cheng GS, Kerr H, Stratton P, Duarte RF, McDonald GB, Inamoto Y, Vigorito A, Arai S, Datiles MB, Jacobsohn D, Heller T, Kitko CL, Mitchell SA, Martin PJ, Shulman H, Wu RS, Cutler CS, Vogelsang GB, Lee SJ, Pavletic SZ, Flowers ME. National Institutes of Health Consensus Development Project on Criteria for Clinical Trials in Chronic Graft-versus-Host Disease: I. The 2014 Diagnosis and Staging Working Group report. Biol Blood Marrow Transplant 2015;21:389–401 e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Bergeron A, Godet C, Chevret S, Lorillon G, Peffault de Latour R, de Revel T, Robin M, Ribaud P, Socie G, Tazi A. Bronchiolitis obliterans syndrome after allogeneic hematopoietic SCT: phenotypes and prognosis. Bone Marrow Transplant 2013;48:819–824. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bergeron A, Chevret S, Peffault de Latour R, Chagnon K, de Margerie-Mellon C, Riviere F, Robin M, Mani J, Lorillon G, Socie G, Tazi A. Noninfectious lung complications after allogeneic haematopoietic stem cell transplantation. Eur Respir J 2018;51:1702617. [DOI] [PubMed] [Google Scholar]
- 13.Armand P, Kim HT, Logan BR, Wang Z, Alyea EP, Kalaycio ME, Maziarz RT, Antin JH, Soiffer RJ, Weisdorf DJ, Rizzo JD, Horowitz MM, Saber W. Validation and refinement of the Disease Risk Index for allogeneic stem cell transplantation. Blood 2014;123:3664–3671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Przepiorka D, Weisdorf D, Martin P, Klingemann HG, Beatty P, Hows J, Thomas ED. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant 1995;15:825–828. [PubMed] [Google Scholar]
- 15.Glucksberg H, Storb R, Fefer A, Buckner CD, Neiman PE, Clift RA, Lerner KG, Thomas ED. Clinical manifestations of graft-versus-host disease in human recipients of marrow from HL-A-matched sibling donors. Transplantation 1974;18:295–304. [DOI] [PubMed] [Google Scholar]
- 16.Zhou Z, Rahme E, Abrahamowicz M, Pilote L. Survival bias associated with time-to-treatment initiation in drug effectiveness evaluation: a comparison of methods. Am J Epidemiol. 2005;162(10):1016–23. [DOI] [PubMed] [Google Scholar]
- 17.Austin PC. Generating survival times to simulate Cox proportional hazards models with time-varying covariates. Stat Med 2012;31:3946–3958. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gail M, Simon R. Testing for qualitative interactions between treatment effects and patient subsets. Biometrics 1985;41:361–372. [PubMed] [Google Scholar]
- 19.Beyersmann J, Allignol A, Schumacher M. Competing risks and multistate models with R. New York: Springer; 2012 [Google Scholar]
- 20.Franklin JM, Rassen JA, Ackermann D, Bartels DB, Schneeweiss S. Metrics for covariate balance in cohort studies of causal effects. Stat Med 2014;33:1685–1699. [DOI] [PubMed] [Google Scholar]
- 21.Austin PC. Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations. Biom J 2009;51:171–184. [DOI] [PubMed] [Google Scholar]
- 22.Inamoto Y, Shah NN, Savani BN, Shaw BE, Abraham AA, Ahmed IA, Akpek G, Atsuta Y, Baker KS, Basak GW, Bitan M, DeFilipp Z, Gregory TK, Greinix HT, Hamadani M, Hamilton BK, Hayashi RJ, Jacobsohn DA, Kamble RT, Kasow KA, Khera N, Lazarus HM, Malone AK, Lupo-Stanghellini MT, Margossian SP, Muffly LS, Norkin M, Ramanathan M, Salooja N, Schoemans H, Wingard JR, Wirk B, Wood WA, Yong A, Duncan CN, Flowers ME, Majhail NS. Secondary solid cancer screening following hematopoietic cell transplantation. Bone Marrow Transplant 2015;50:1013–1023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Boursi B, Mamtani R, Haynes K, Yang YX. Recurrent antibiotic exposure may promote cancer formation--Another step in understanding the role of the human microbiota? Eur J Cancer 2015;51:2655–2664. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lin SJ, Kuo ML, Hsiao HS, Lee PT. Azithromycin modulates immune response of human monocyte-derived dendritic cells and CD4(+) T cells. Int Immunopharmacol 2016;40:318–326. [DOI] [PubMed] [Google Scholar]
- 25.Ratzinger F, Haslacher H, Poeppl W, Hoermann G, Kovarik JJ, Jutz S, steinberger P, Burgmann H, Pickl WF, Schmetterer KG. Azithromycin suppresses CD4(+) T-cell activation by direct modulation of mTOR activity. Sci Rep 2014;4:7438. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Lin SJ, Yan DC, Lee WI, Kuo ML, Hsiao HS, Lee PY. Effect of azithromycin on natural killer cell function. Int Immunopharmacol 2012;13:8–14. [DOI] [PubMed] [Google Scholar]
- 27.Baines KJ, Wright TK, Gibson PG, Powell H, Hansbro PM, Simpson JL. Azithromycin treatment modifies airway and blood gene expression networks in neutrophilic COPD. ERJ Open Res 2018;4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Galloway-Pena JR, Jenq RR, Shelburne SA. Can Consideration of the Microbiome Improve Antimicrobial Utilization and Treatment Outcomes in the Oncology Patient? Clin Cancer Res 2017;23:3263–3268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Vivarelli S, Salemi R, Candido S, Falzone L, Santagati M, Stefani S, Torino F, Banna GL, Tonini G, Libra M. Gut Microbiota and Cancer: From Pathogenesis to Therapy. Cancers (Basel) 2019;11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Goodman B, Gardner H. The microbiome and cancer. J Pathol 2018;244:667–676. [DOI] [PubMed] [Google Scholar]
- 31.Noor F, Kaysen A, Wilmes P, Schneider JG. The Gut Microbiota and Hematopoietic Stem Cell Transplantation: Challenges and Potentials. J Innate Immun 2018:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Peled JU, Devlin SM, Staffas A, Lumish M, Khanin R, Littmann ER, Ling L, Kosuri S, Maloy M, Slingerland JB, Ahr KF, Porosnicu Rodriguez KA, Shono Y, Slingerland AE, Docampo MD, Sung AD, Weber D, Alousi AM, Gyurkocza B, Ponce DM, Barker JN, Perales MA, Giralt SA, Taur Y, Pamer EG, Jenq RR, van den Brink MRM. Intestinal Microbiota and Relapse After Hematopoietic-Cell Transplantation. J Clin Oncol 2017;35:1650–1659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bejanyan N, Weisdorf DJ, Logan BR, Wang HL, Devine SM, de Lima M, Bunjes DW, Zhang MJ. Survival of patients with acute myeloid leukemia relapsing after allogeneic hematopoietic cell transplantation: a center for international blood and marrow transplant research study. Biol Blood Marrow Transplant 2015;21:454–459. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Schmid C, de Wreede LC, van Biezen A, et al. Outcome after relapse of myelodysplastic syndrome and secondary acute myeloid leukemia following allogeneic stem cell transplantation: a retrospective registry analysis on 698 patients by the Chronic Malignancies Working Party of the European Society of Blood and Marrow Transplantation. Haematologica 2018;103:237–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Tichelli A, Beohou E, Labopin M, Socie G, Rovo A, Bagoglio M, van Biezen A, Bader P, Duarte RF, Basak G, Salooja N, Transplant Complications Working Party of the EBMT. Evaluation of Second Solid Cancers After Hematopoietic Stem Cell Transplantation in European Patients. JAMA Oncol 2019;5:229–235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hefazi M, Langer KJ, Khera N, Adamski J, Roy V, Winters JL, Gastineau DA, Jacob EK, Kreuter JD, Gandhi MJ, Hogan WJ, Litzow MR, Hashmi SK, Yadav H, Iyer VN, Scott JP, Wylam ME, Cartin-Ceba R, Patnaik MM. Extracorporeal Photopheresis Improves Survival in Hematopoietic Cell Transplant Patients with Bronchiolitis Obliterans Syndrome without Significantly Impacting Measured Pulmonary Functions. Biol Blood Marrow Transplant 2018;24:1906–1913. [DOI] [PubMed] [Google Scholar]
- 37.Azithromycin Welte T., the Holy Grail to Prevent Exacerbations in Chronic Respiratory Disease. Am J Respir Crit Care Med 2019; 200:269–270. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.