Pulmonary Clinicopathological Correlation after Allogeneic Hematopoietic Stem Cell Transplantation: An Autopsy Series

Abstract

Pulmonary complications are a significant cause of morbidity, mortality, and resource utilization after hematopoietic stem cell transplantation (HSCT). The objective of this study was to compare antemortem clinical suspicion of pulmonary complications and postmortem findings in a modern HSCT cohort. All patients who underwent allogeneic HSCT at our institution (n = 1854) between January 1, 2000 and June 30, 2010 were reviewed and patients who died of any cause greater than 1 year after HSCT and had an unrestricted autopsy available for analysis were included. Presence of pulmonary graft-versus-host disease (GVHD) was assessed by a pathologist blinded to the autopsy report, as previously described by Yousem (1995). A total of 35 (1.9%) patients had autopsies available for review. Airway disease, vascular disease, and interstitial disease were all clinically under-recognized compared with the pathological findings on autopsy. Varying degrees of pathological changes were detected, including 10 (28.6%) patients having bronchiolitis obliterans (BO) and 12 (34.3%) patients having pulmonary veno-occlusive disease (PVOD). Pulmonary manifestations of chronic GVHD, particularly BO and PVOD, were clinically under-recognized in our cohort. Our results suggest that PVOD, which has traditionally been considered a rare complication, may be clinically and histologically under-recognized.

INTRODUCTION

Pulmonary complications after hematopoietic stem cell transplantation (HSCT) are a major cause of morbidity, mortality, and resource utilization [1]. They are estimated to occur in nearly one-half of patients and account for approximately 50% of transplantation-related deaths and diminished quality of life in those who survive [2].

There is a wide spectrum of pulmonary complications associated with HSCT and the etiologies are believed to be multifactorial, including pretransplantation chemo radiation, infection, and graft-versus-host disease (GVHD) [3]. Late-onset clinical syndromes affecting lungs in patients after HSCT also include bronchiolitis obliterans syndrome (BOS), cryptogenic organizing pneumonia (COP), and pulmonary veno-occlusive disease (PVOD) [4,5]. The histological spectrum of pulmonary GVHD was described by Yousem [6] and includes manifestations in the airways, vasculature, and interstitium. Airway features include lymphocytic bronchitis/bronchiolitis (LB), lymphocytic bronchitis/bronchiolitis with intraluminal granulation tissue, and cicatricial bronchiolitis obliterans (BO) [6]. Vasculature manifestations include veno-occlusive disease [6]. Pathological changes within the interstitium include cellular interstitial pneumonitis (CIP), airspace granulation tissue/organizing pneumonia (OP), and patchy interstitial fibrosis (IF) [6].

Among these, only BOS/BO is currently considered diagnostic of chronic GVHD, according to the 2005/2014 National Institutes of Health (NIH) consensus criteria [5,7].

The objective of this study was to compare antemortem clinical suspicion of pulmonary complications with postmortem findings in a retrospective autopsy cohort of patients status post HSCT who survived more than 1 year after transplantation.

METHODS

All patients who underwent allogeneic HSCT at our institution (n = 1854) between January 1, 2000 and June 30, 2010 and survived more than 1 year were reviewed for autopsies available for analysis. Histological findings were assessed by a pathologist blinded to the autopsy report, according to the categories of injury previously described by Yousem [6] as present or absent. Acute and chronic GVHD were graded by consensus grading criteria [8,9]. Lymphocytic infiltration and structural remodeling were scored separately for the airways, vasculature, and interstitium. International Society of Heart Lung Transplant criteria were used for grading venopathy [10,11]. Patients with extensive lung involvement by infection, malignancy, or other defined process (ie, aspiration injury) were excluded from analysis.

Clinical variables were collected prospectively as part of the usual care and stored in a centralized data repository. Premortem diagnosis of BOS, PVOD, and interstitial disease was based on treating physician’s clinical interpretation and diagnosis based on medical record review. Additional clinical variables not stored in the centralized data repository were collected retrospectively by review of the electronic medical record.

This study (protocol 09-316) was approved by the institutional review board at Dana Farber/Harvard Cancer Center, Boston, MA.

Statistical Analysis

Data are presented as median and interquartile range for continuous variable and as percent for categorical variables. Categorical and continuous variables were compared between groups using 2-sided Fisher’s exact test and 2-sided Wilcoxon rank-sum test, respectively. The main analysis was performed using logistic regression where the outcome was airway disease, vascular disease, or interstitial disease. We first performed univariate logistic analysis with the following variables: patient age, gender, race, pretransplantation forced expiratory volume in 1 second as percent, pretransplantation carbon monoxide diffusion capacity as percent, patient cytomegalovirus (CMV) status at transplantation, high-risk disease, peripheral blood stem cell source, myeloablative conditioning regimen, busulfan-based conditioning regimen, presence of acute GVHD, presence of chronic GVHD, donor age, gender, donor-recipient gender mismatch, donor CMV status, related donor, and matched donor. A multivariable model was built using forward selection based upon their statistical significance (P < .05) on univariate analysis. All statistical analyses were performed using SAS9.3.

RESULTS

A total of 37 (2%) patients were noted to have an autopsy with lung specimens available for analysis. Two of these patients were noted to have diffuse parenchymal involvement by infection and were excluded, leaving a sample size of 35 for analysis. The median age of the cohort was 51 (42; 57) and the subjects survived for a median of 610 days (452; 892) after transplantation. Pulmonary complications were felt to contribute to death in 17 (49%) cases, and 32 (91%) of the patients had a diagnosis of chronic GVHD affecting organ systems other than the lungs before death (Table 1 and Table 2).

Table 1.

Underlying Disease at Time of Transplantation, Cause of Death, and Days after Transplantation of the Autopsy Cohort

Subject	Disease	Cause of Death as Determined by Autopsy	Days after Transplantation at Time of Death
1	AML	Hepatic failure secondary to probable VOD and a minor component of GVHD	438
2	MDS	Myocardial infarction and a cerebrovascular accident most likely as a result of thromboembolic events in the setting of myelodysplasia	497
3	CML	Renal and hepatic insufficiency in the setting of high-output congestive heart failure and diffuse serosanguinous effusions	836
4	CML	Pneumococcus sepsis	929
5	AML	Respiratory failure associated with marked DAD and superimposed acute bronchopneumonia	140
6	NHL	Sepsis in the setting of infective endocarditis, complicated by a large ring abscess, with myocardial seeding from septic emboli, ultimately ending in compromise of cardiac and respiratory function	2801
7	AML	Multiorgan system failure, including early DAD, renal failure, and centrilobular vascular congestion of the liver and hepatocyte dropout, that developed in the setting of relapse/persistence of AML, mild-to- moderate GVHD, and VOD with centrilobular venous congestion of the liver	391
8	CML	Candida albicans fungal pneumonia with superimposed Pseudomonas aeruginosa infection, complicated by acute renal failure and liver dysfunction	2667
9	ALL	Bronchopneumonia and sepsis in the setting of documented anaplasmosis	2666
10	CML	Progressive respiratory failure secondary to bilateral, mild acute and chronic GVHD with focal areas of airway obliteration consistent with BO with organizing pneumonia and organizing DAD	1433
11	CLL	Acute, bilateral uncal, and cerebellar tonsil herniation in the setting of diffuse cerebral edema after complications arising from hemorrhagic pancreatitis in the setting CLL status post bone marrow transplantation	603
12	NHL	Acute myocardial infarction in the setting of coronary artery disease secondary to chronic GVHD	598
13	CML	Respiratory failure with DAD and acute liver failure in the setting of chronic GVHD	638
14	NHL	Bilateral bronchopneumonia complicated by pleural fluid accumulation, multifocal healing myocardial microinfarcts, renal failure, and hemorrhagic colitis leading to multiorgan failure	3454
15	AML	DAD compounded by bacteremia and renal failure	136
16	HD	Acute and organizing DAD arising in the setting of extensive relapse of HD	374
17	NHL	Organizing DAD with focal acute pneumonitis in the setting of lung and liver GVHD	686
18	NHL	Hypoxemic respiratory failure due to bilateral organizing pneumonia with early organizing DAD	610
19	CLL	VOD in the setting of myeloablative stem cell transplantation for CLL	666
20	CLL	Bilateral acute bronchopneumonia	700
21	MDS	GVHD and organizing DAD, died in cardiorespiratory failure	452
22	HD	Extensive bilateral pulmonary emboli from a left femoral vein thrombosis, associated with severe acute bilateral bronchopneumonia superimposed on infarcted lung parenchyma	1379
23	CML	Progressive respiratory failure (clinical) secondary to severe bilateral upper lobe emphysema (COPD), pulmonary infarction associated with invasive zygomycosis, and pulmonary hypertension	602
24	ALL	Respiratory failure in the setting of acute and organizing DAD, subacute pulmonary embolism, and superimposed fungal infection	760
25	CLL	Liver failure due to GVHD, with polymicrobial sepsis as a contributing factor	509
26	CML	Acute respiratory distress syndrome with DAD and progressive respiratory failure	396
27	MDS	E. coli sepsis of gastrointestinal origin secondary to involvement of the bowel by PTLD	705
28	AML	Klebsiella pneumonia sepsis and pneumonia	432
29	MDS	Sepsis, suspect E. coli as etiology, with ischemic end-organ damage, including bowel	586
30	AML	Acute respiratory distress syndrome due to DAD and organizing pneumonia	485
31	NHL	Hypoxemic respiratory failure due to adenoviral pneumonia and DAD/diffuse alveolar hemorrhage	892
32	AML	Staphylococcus aureus bronchopneumonia with DAD in the setting of PVOD	486
33	MPD	Gram-negative bacteremia and sepsis, likely arising from a cecal ulcer, consistent with typhlitis, and secondary bronchopneumonia	794
34	MDS	Bilateral DAD and focal hemorrhage, left main pulmonary artery thromboembolism, left upper lobe hemorrhagic infarct with smaller thromboemboli and bilateral pleural effusions in the setting of MDS status post bone marrow transplantation complicated by chronic mucocutaneous GVHD	377
35	AML	Multiorgan involvement by AML (which included extensive pulmonary leukemic infiltrates), focal bronchopneumonia in the setting of emphysematous changes in the lung, and adrenal cortical hemorrhage consistent with Waterhouse-Friderichsen syndrome	449

AML indicates acute myeloid leukemia; VOD, veno-occlusive disease; MDS, myelodysplastic syndrome; CML, chronic myeloid leukemia; DAD, diffuse alveolar damage; NHL, non-Hodgkin lymphoma; ALL, acute lymphocytic leukemia; CLL, chronic lymphocytic leukemia; HD, Hodgkin disease; COPD, chronic obstructive pulmonary disease; PTLD, post-transplantation lymphoproliferative disease; MPD, myeloproliferative disease.

Table 2.

Clinical Characteristics of Autopsy Cohort

Characteristics	All Patients N = 35	Airway Disease n = 23 (66%)	Venopathy n = 14 (40%)	Interstitial Disease n = 13 (37%)
Age, yr	51 (42–57)	49 (57–38)	47 (38–56)	56 (51–57)*
Male gender	21 (60%)	15 (65%)	10 (71%)	8 (62%)
White race	32 (91%)	2/0 (87%)	14 (100%)	12 (92%)
Days after Transplantation	610 (452–892)	686 (596–929)	606 (438–929)	610 (485–929)
Pretransplantation FEV1%	97 (84–109)	96 (83–105)	102 (95–109)	99 (94–109)
Pretransplantation DLCO%	82 (70–101)	81 (71–100)	89 (71–106)	74 (70–106)
CMV positive	16 (46%)	15 (65%)*	6 (43%)	8 (62%)
High-risk disease†	21 (60%)	16 (70%)	11 (79%)	8 (62%)
Cell source
PBSC	26 (74%)	16 (70%)	9 (64%)	11 (85%)
Conditioning
Myeloablative‡	20 (57%)	15 (66%)	12 (86%)*	7 (54%)
Busulfan based	14 (40%)	7 (30%)	2 (14%)*	6 (46%)
GVHD
Acute GVHD§	19 (54%)	14 (61%)	11 (79%)*	7 (54%)
Chronic GVHD§	32 (91%)	23 (100%)*	14 (100%)	12 (92%)
Donor
Age, yr	37 (29–43)	39 (29–45)	41 (33–45)	39 (33–43)
Donor male gender	21 (60%)	13 (57%)	6 (43%)	4 (31%)*
Gender mismatch	10 (29%)	8 (35)%)	7 (50)%	6 (46%)
Related	7 (20%)	4 (17%)	4 (29%)	4 (31%)
Matched	31 (89%)	19 (83%)	14 (100%)	11 (85%)
Donor CMV positive	10 (29%)	7 (30%)	4 (29%)	4 (31%)

FEV₁ indicates forced expiratory volume in 1 second; DLCO, carbon monoxide diffusion capacity; PBSC, peripheral blood stem cell.

^*P < .05.

^†All diseases other than CML first chronic phase, AML/ALL in first complete remission MDS refractory anemia with ringed sideroblasts, or aplastic anemia are considered high risk in this analysis.

^‡Myeloablative conditioning regimens: cyclophosphamide + total body irradiation 1400 cGY; high-dose busulfan cyclophosphamide.

^§Acute and chronic GVHD was graded by consensus grading criteria [8,9].

A total of 23 (66%) patients had evidence of airway disease, 14 (40%) subjects had venopathy, and 13 (37%) patients had disease of the interstitium. Of these, only 8 (34.8%), 1 (7.1%), and 5 (38.5%), respectively, were suspected before death.

Airway Disease

Of the 35 patients, 23 (66%) had evidence of airway disease (Figure 1), LB was present in 17 (48.6%), LB/bronchiolitis with intraluminal granulation tissue was found in 5 (14.3%), and BO was identified in 10 (28.6%) (Table 3). A total of 20.4% (7 of 23) of the subjects with airway disease had more than 1 manifestation of airway disease present at the time of autopsy. A total of 8.7% (2 of 23) patients had evidence of all 3 manifestations of airway disease present at the time of autopsy. Multivariate analysis revealed that CMV-positive status (P < .02) and presence of chronic GVHD (P < .01) were both independent factors associated with the development of airway disease.

Figure 1.

Airways disease. (A) Shows lymphocytic bronchitis/bronchiolitis and (B) shows bronchiolitis obliterans.

Table 3.

Autopsy Findings

Autopsy Findings: Pathological Findings
Airway disease, n = 23 (66%)
LB	17 of 35 (48.6%)
Lymphocytic bronchitis/bronchiolitis with intraluminal granulation tissue	5 of 35 (14.3%)
BO	10 of 35 (28.6%)
Pulmonary venous disease, n = 14 (40%)
Perivascular lymphoplasmacytic infiltrates	3 of 35 (8.6%)
Septal obliteration (PVOD)	12 of 35 (34.3%)
Interstitial disease, n = 13 (37%)
CIP	6 of 35 (17.1%)
OP	6 of 35 (17.1%)
IF	8 of 35 (22.9%)

Venopathy

Fourteen (40%) patients had venopathy (Figure 2), perivascular lymphoplasmacytic infiltrates were identified in 8.6% (3 of 35), and PVOD was identified in 34.3% (12 of 35) of the subjects (Table 3). Multivariate analysis revealed that myeloablative conditioning (P < .01) and presence of acute GHVD (P < .04) were independent factors associated with the development of venopathy. There was also a trend towards an association with airway disease and pulmonary venous disease (P = .07).

Figure 2.

Pulmonary venous disease. Septal edema and narrowing of septal veins by myxoid connective tissue.

Disease of the Interstitium

Thirteen (37%) patients had disease of the interstitium (Figure 3). CIP was found in 17.1% (6 of 35), OP in 17.1% (6 of 35), and IF in 22.9% (8 of 35). Multivariate analysis revealed only female donors remained as an independent factor associated with the development of disease of the interstitium (P < .01).

Figure 3.

Disease of the interstitium. (A) Shows cellular interstitial pneumonitis and (B) shows interstitial fibrosi.

Upon univariate analysis the presence of OP was associated with BO (P < .05). Upon univariate analysis the presence of CIP was associated with IF (P < .02).

DISCUSSION

Pulmonary complications and respiratory failure after HSCT are a major cause of post-transplantation morbidity, mortality, and resource utilization [1,12]. We sought to characterize the pathologic pulmonary findings in patients who survived more than 1 year after allogeneic HSCT. We found that airway disease, vascular disease, and interstitial disease were all clinically under-recognized in our cohort.

BOS is a form of irreversible airflow obstruction and is a late noninfectious pulmonary complication after HSCT that is associated with a significant increase in morbidity and mortality [12–15].

Yousem described GVHD of the airways to follow a spectrum from LB to BO. He speculated that LB represents an early manifestation of airway disease and may be amenable to therapy before the onset of fibrosis and BO [6]. This theory is supported by the findings of Holbro et al. [16], who demonstrated that patients with LB were more likely to respond to therapy and had a significant improvement in overall survival compared with patients with constrictive BO. Our results also suggest that airway disease is a spectrum, all 3 forms of airway disease can coexist in the same subject and, airway disease is clinically under-recognized in our HSCT cohort. Our study was performed before the initiation of a standard screening protocol at our institution. We previously published the prevalence of BOS in this cohort to be 4.8% [17]. Previous estimates of the prevalence of BOS range from 2% to 30% and up to 30% of patients are asymptomatic at the time of meeting NIH criteria [13–15,18–22]. In the absence of routine spirometric screening, reports of disease prevalence likely underestimate the true burden of BOS, since symptomatic patients are typically already suffering from moderate-to-severe airflow obstruction [19]. Kuzmina et al. found the cumulative incidence of BOS to be 27% at 3 years using a strict 3-month screening protocol, suggesting we may be vastly underestimating the true prevalence of BOS [21]. Within our autopsy series, only 34.3% (12 of 35) subjects had any post-transplantation pulmonary function test performed. Thus, our results suggest that airway disease and BO may have been clinically under-recognized in our previously published retrospective cohort [17].

The development of BOS has been found to be highly associated with the presence of chronic GVHD at other sites and is currently considered to be a form of chronic GVHD of the lungs, according to NIH consensus criteria [5,7,12–15,19,23–25]. Other risk factors reported to be associated with BOS include acute GVHD, busulfan-based conditioning regimens, use of methotrexate for GVHD prophylaxis, CMV, respiratory viral infections within the first 100 days after transplantation, peripheral blood stem cell source, history of pneumonitis, low immunoglobulin levels, and reduced values on pretransplantation pulmonary function tests [13,14,17–20,23–27]. Our results confirm that chronic GVHD and CMV status are associated with the development of airway disease.

PVOD is considered a rare complication of HSCT. PVOD and is pathologically identified by intimal proliferation and fibrosis of the pulmonary venules and small veins leading to progressive vascular obstructions. PVOD can lead to severe pulmonary arterial hypertension; radiographic evidence of pulmonary edema and a normal pulmonary artery wedge pressure are the clinical criteria for diagnosis of PVOD [28,29]. The etiology has been speculated to be the result of regimen-related toxicity or possibly related to the inflammatory nature of GVHD [29–32]. Our data suggests that both regimen-related toxicity and GVHD may play a role in the development of vasculopathy and PVOD. We demonstrated that myeloablative conditioning regimen was associated with PVOD. Cyclophosphamide/total body irradiation was the standard myeloablative regimen used in all patients in our cohort. This is consistent with prior reports suggesting an association between cyclophosphamide and PVOD [28,32]. A striking finding was that the chart review revealed that clinically only 7% (1 of 14) patients with PVOD were identified before death. This may suggest that PVOD is either clinically under-recognized or the pathological findings may vary in clinical significance.

COP is a disorder that involves the alveoli, alveolar ducts, and bronchioles, which become filled with buds of granulation tissue consisting of fibroblasts and an associated matrix of loose connective tissue [33]. Currently consensus diagnostic criteria are lacking for the diagnosis of COP after HSCT. The incidence of COP is estimated at 1% to 2% based upon single-center reports [34,35]. Our data suggests that OP is also under-recognized in our HSCT cohort. In addition we found an association between OP and BO, which is considered part of the NIH diagnostic criteria for BOS and may represent a manifestation of chronic GVHD [7]. However, OP often seen with BO may not reflect COP, which is often amenable to steroid treatment. Further study between the association of OP, BO, and COP is warranted.

Within our cohort, we also found an association between CIP and IF, suggesting a possible correlation. Pulmonary fibrosis after HSCT has been described and it has been postulated to be a manifestation of late toxicity of total body irradiation or chemotherapy or the result of interstitial pneumonitis secondary to infection or GVHD [36]. Similar to LB progressing to airway fibrosis and BOS, a similar process with CIP and the subsequent development of pulmonary fibrosis is suggested by our results.

Our data consist of a retrospective assessment of clinically acquired data. The premortem clinical diagnoses were at the discretion of the treating physicians and obtained from chart review. As such, it likely biased in nature to those patients who underwent clinical evaluation or an autopsy. In addition, diffuse alveolar damage was present to some degree in the majority of autopsy cases (15 of 35; 43%), often as a secondary result of the immediate cause of death and, thus, may confound our results. We made every effort to score these samples away from areas of diffuse alveolar damage, but the possibility remains that this process may lead to a bias towards overcalling pathologic findings. Finally, although the pathologic findings were present on autopsy, their clinical significance is not clear and the pathological findings may have been truly under-recognized or of an unclear clinical significance.

Given the limitations, our study suggests that pulmonary pathology of the airways, vasculature, and interstitium after HSCT may be clinically under-recognized and warrants further study.