Airway Inflammation Predicts Diffuse Alveolar Hemorrhage during Bone Marrow Transplantation in Patients with Hodgkin Disease1- 3
JOSEPH H. SISSON, AUSTIN B. THOMPSON, JAMES R. ANDERSON, RICHARD A. ROBBINS, JOHN R. SPURZEM, PAUL R. SPENCE, ELIZABETH C. REED, JAMES O. ARMITAGE, JULIE M. VOSE, MARK A. ARNESON, WILLIAM P. VAUGHAN, and STEPHEN I. RENNARD
Introduction
High-dose chemotherapy (HCTx) supported by bone marrow transplantation (BMT) has become an increasingly im-" portant curative-intent treatment modality for patients with malignancies resistant to conventional chemotherapy and radiation treatment. A major dose-limiting toxicity of this approach, however, is pulmonary failure. Between 10 and 200/0 of patients undergoing BMT die of pulmonary complications during their initial hospitalization (1). We recently described a diffuse alveolar hemorrhage syndrome (DAH) associated with an 800/0 mortality that occurred in approximately 200/0 of patients during autologous BMT (2). Although several factors such as age, temperature, platelet count, and oral mucositis score at the onset of the syndrome appeared to be associated with the development of DAH, no parameters were identified prior to the administration of HCTx and marrow infusion that were predictive of subsequent DAH and death. Because DAH was associated with mucositis at the time of the onset of the syndrome, we hypothesized that airway inflammation present before HCTx and BMT might predict for the development of DAH. We therefore prospectively examined a large number of patients at risk for DAH before HCTx and BMT by bronchoscopy with bronchoalveolar lavage (BAL) to identify factors in the lower respiratory tract that predicted for the development of DAH. Wereport that patients with Hodgkin's disease have evidenceof airwayinjury before the initiation of HCTx and BMT and that increased proportions of airway neutrophils and eosinophils are predictive for the occurrence of DAH and death.
SUMMARY We determined risk factors present In patients with Hodgkin disease that predicted the development of diffuse alveolar hemorrhage syndrome (DAH) during autologous bone marrow transplantation (BMT). One hundred twenty-three patients with Hodgkin disease prospectively underwent bronchoscopy with bronchoalveolar lavage (BAL) before receiving BMT.The bronchitis Index (BI) of the airways and bronchial and alveolar cell counts and differentials were determined In all patients and compared with 20 normal nonsmoking volunteers. logistic regression analysis was used to determine factors that predicted for the development of DAH. Visual evidence of bronchial Injury was observed regardless of smoking history (BI 7.8 ± 0.5 for BMT versus 2.3 ± 0.5 for volunteers, p 0.001). BMT patients who developed DAH (n 14) had significantly greater 109) patients numbers of bronchial neutrophlls and eoslnophlls compared with DAH-negatlve (n 0.006; bronchial (bronchial polymorphonuclear leukocytes (PMN), 33 ± 7% versus 14 ± 2%, P eoslnophlls,0.9 ± 0.3% versus 0.4% ± 0.07%, P = 0.02). logistic regression analysis revealed that the presence of bronchial PMN > 20% or bronchial eoslnophlls > zero% were predictive of DAH (p = 0.005 and 0.05, respectively). When both predictors were positive, the rate of DAH was 10 times greater than when both predictors were negative (43% versus 4% DAH occurrence). Survival was also sign Iflcantly reduced when these predictors were positive. This study demonstrates that bronchial Inflammation Is present with or without Intraluminal Inflammatory cells In the maJority of patients with Hodgkin disease before BMT. The subgroup of these patients with Increased bronchial Inflammatory cells are at greatly Increased risk for development of DAH and death.
=
=
=
=
=
AM REV RESPIR DIS 1992; 146:439-443
Methods Patients From January 1, 1987through March 31, 1990 a total of 143 patients with the diagnosis of Hodgkin disease were treated with HCT x and BMT at the University of Nebraska Medical Center, Omaha. Surveillance bronchoscopy with BAL was performed on 123of these patients prior to the initiation of HCTx and within 30 days of bone marrow infusion. Surveillance bronchoscopy, performed in the absence of clinical or radiographic evidence of active lung disease, has been shown to detect occult pulmonary abnormalities (3). The bronchoscopy findings of the BMT patients (age 32.25 ± 0.75 yr; mean ± SE) were compared with a group of 20 normal nonsmoking volunteers (age 33.1 ± 2.1 yr).
High Dose Chemotherapy and 'Iransplantation Protocol HCTx consisted of cyclophosphamide, etoposide (VP-16) and carmustine (BCNU) followed by autologous bone marrow infusion
(Received in original form August 19, 1991 and in revised form January 17, 1992) 1 From the Pulmonary/Critical Care and Oncology/Hematology Sections, Department of Medicine, and the Department of Preventive and Societal Medicine, University of Nebraska MedicalCenter, Omaha, Nebraska. 2 Correspondence and requests for reprints should be addressed to Joseph H. Sisson, M.D., Department of Internal Medicine, Pulmonary and Critical Care Medicine Section, University of Nebraska Medical Center, 600 South 42nd Street, Omaha, NE 68198-2465. 3 In addition to the above-named authors, the University of Nebraska Medical Center Bone Marrow Transplantation/Pulmonary Study Group includes the following individuals: P. Bierman, R. Cheney, A. Curseen, D. Darrington, K. Dicke, B. Edwards, J. Foley,T. Franco-Crowley,B. Frappier, J. Goldsmith, J. Gurney, W. Haire, W. Huerta, K. Jones, A. Kessinger, L. Klassen, J. Linder, J. Metcalf, J. Pierson, M. Raz, K. Rickard, D. Romberger, I. Rubinstein, K. Schidt-Pokorny, M. Shumway, J. Spinolo, M. Tempero, S. Von Essen, T. Walter, E. Wedergren, W. West, J. Wisecarver, R. Witte, G. Woods, and R. Yates.
439
440
SISSON, AUSTIN, THOMPSON, ET AL.
(4) in 116patients (94.3070). Six other patients (4.9070) received slightly different chemotherapy protocols followed by autologous bone marrow infusion. One additional patient received busulfan, cyclophosphamide, and VP16followed by allogeneic bone marrow infusion. All patients were included in the data analysis since exclusion of the seven patients with alternative chemotherapy protocols did not affect the results (data not shown).
Bronchoscopy and Bronchoalveolar Lavage Patients underwent surveillance bronchoscopy and BAL after giving written informed consent at the time of admission. Bronchoscopy and BAL were performed as previously described (5). Briefly, bronchoscopy was performed by inserting the end 0 f a bronchoscope (Olympus IT-R or PO-I0; Olympus Corp. of America, New Hyde Park, NY) transorally. A visual score for each of four characteristics (erythema, edema, secretions, and friability) was determined by assigning a value of zero to 3 (zero = normal, 3 = severely abnormal). This score was determined for each of the six major bronchi, and the bronchitis index (BI) represents the sum of these scores (6). The bronchoscope was gently advanced into a wedged position and five 20-ml aliquots of sterile saline were infused through the bronchoscope and immediately aspirated. Lavage was performed in each of three sites and usually included the right middle lobe, the lingula, and one of the lower lobes. The bronchial sample (first 20-ml aliquot) waskept separate from the alveolarsample (subsequent four 20 ml aliquots [5J). Bronchoalveolar Lavage Processing and Cytologic Examination Both the bronchial and alveolar samples were processedas previouslydescribed (5).The cells wereprepared for cytologicexamination by cytocentrifugation (Cytospin; Shandon Southern Instruments, Sewickley, PA). Celldifferentials were performed from the modified Wright-Giemsa-stained samples (Diff-Quik~; American Scientific, McGraw Park, IL). Papanicolaou stains and Gomori methenamine silver stained slides were also prepared for cytologic examination. Statistical Analysis Comparisons of the distribution of values of
continuous variables between subject groups the major bronchi, a semiquantitative BI were made using independent sample Stu- was determined (6). The BI was signifident's t tests, and the values were expressed cantly elevated in the patients undergoas means ± standard errors. The association ing BMT compared with normal nonof bronchoscopy and BAL data with the sub- smokers (7.9 ± 0.5 versus 2.3 ± 0.5, p = sequent occurrence of OAH was assessed using appropriate chi-squared tests for con- 0.(01) (figure 1). An elevated BI was prestingency tables. Multivariate assessment of ent regardless of the smoking status of the relationship of various bronchial and al- the patients at the time of admission for veolar percentages and the occurrence of BMT (8.8 ± 1.6 for smokers and 7.9 ± OAH was made through the use of logistic 0.6 for nonsmokers; p = 0.62). regression. Survival was defined as the time Cell Count and Differential Findings from transplantation to death from any cause. The distribution of survival times was esti- An elevated BI has been associated with mated using the product-limit method of the presence of inflammatory cells in the Kaplan and Meir (7). The statistical sig- airways of patients with chronic bronnificance of differences observed in survival chitis (6). We therefore examined the between patient subgroups was assessedusing the log-rank test (8). All p values reported are . bronchial and alveolar lavage fluid of two-sided. All data analyses were done using BMT patients for the presence of inflamstatistical analysis software (SAS Institute, matory cells (table 2). Bronchial neutrophilia (bronchial polymorphonuclear Inc., Cary, NC). leukocytes [PMN] > 20070 [6]) was present in 31 of 123 patients (25070). When Results all BMT patients were analyzed for bronDemographics chial neutrophilia, however, the mean One hundred forty-three patients with the bronchial neutrophil fraction was highdiagnosis of Hodgkin disease were ader than that observed in the nonsmoker mitted to the University of Nebraska control subjects, but the difference did Medical Center between January 1, 1987 not reach statistical significance (BMT and March 16,1990 for HCTx and BMT patients = 17.4 ± 2.0 versus normal sub(table 1). Twenty patients were excluded = 9.5 ± 1.0, p = 0.14). The degree jects from analysis because (1) active pulmoof bronchial neutrophilia did correlate nary disease was apparent clinically (n = with the visual BI (r = 0.46, 2), (2) their surveillance bronchoscopy positively 2 was performed outside the 30 day pre- r = 0.21, p = 0.0001). No other signifiBMT period (n = 5), (3) the bronchos- cant change in the proportion of cells recopy was performed after chemothera- covered was observed when all patients py was started (n = 5), or (4) bronchos- were analyzed (data not shown), with one copy was refused (n = 8). Thus, the exception, the fraction of alveolar macresults of 123 (86070) surveillance bron- rophages from the bronchial samples was choscopies were available for analysis. significantly lower in BMT patients than The study population comprised 73 men in normal nonsmoking control subjects (59070) and 50 women (41070), mean age, (p = 0.03). These findings were true 32.3 ± 0.75 yr. Bronchoscopy was per- regardless of whether the fraction of cells formed before the initiation of HCTx on or the absolute number of cells for each average approximately 1 wk before BMT cell type was compared.
(Day - 8.92 ± 0.40 days; range, Day - 29 to Day -5).
Bronchoscopy Findings To measure the degree of large airway inflammation present upon inspection of
TABLE 1 PATIENTS WITH BONE MARROW TRANSPLANTATION-HODGKIN DISEASE UNDERGOING SURVEILLANCE BRONCHOSCOPY Total patients for BMT Patients with surveillance bronchoscopy, n (0/0) Males, n (Ok) Females, n (0/0) Current smokers, n (Ok) Nonsmokers, n (Ok) Required ventilator during BMT, n (Ok) Developed DAH during BMT, n (0/0) Died during BMT, n (Ok) Age, yr (mean ± SE) Day of bronchoscopy relative to BMT (mean ± SE) Dsf/n/t/on of abbreviations: BMT
= bone marrow transplantation;
143 123 (100) 73 (59) 50 (41) 12 (10) 111 (90) 20 (16) 14 (11) 17 (14) 32.25 :t 0.75 - 8.92 :t 0.40 DAH
Autopsy Results of Patients Dying with DAB Because bronchial airway inflammation and not alveolar inflammation appeared to correlate with the development of DAH, we reviewed the available autopsies to ascertain whether a bleeding site could be determined. Autopsies were performed on four of nine patients (44070) who died with DAH. Extensive diffuse and patchy hemorrhage and alveolar damage was present in all four cases. No airway ulcerations were identified and no consistent site of bleeding was apparent.
Outcome Analysis (range, 17 to 54 yr) (range, - 29 to - 5 days)
= diffuse alveolar hemorrhage syndrome.
DAH is a common, life-threatening pulmonary complication encountered during autologous BMT (2). DAH occurred in 14 of 123 patients with an incidence
441
BRONCHITIS, BMT, AND DAH IN HODGKIN DISEASE PATIENTS
Fig. 1. Comparison of bronchitis index (BI) in normal volu nteers with bone marrow transplantation (BMT) patients prior to high-dose chemotherapy. The BI is a visual score of inflammatory findings at bronchoscopy. The vertical axis represents the sum of the visual score from the six major bronchi before bronchoalveolar lavage. The open circles represent the BI of normal volunteers (n = 25);the closed circles represent the BMT patients (n = 123) with Hodgkin's disease.
25
.
20
.....................................................................•..................
15
••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• ;~.~.:~~.~.~ •••••••••••.
•• • •••
...
~
G ~
••
! ~
u
. . ••::::••
0
.:::::.
10
~
m
o
5
••••••••••••
~
00 000
.
.
•••••• ••••••••••
O&X)
•••••••
ססoo
••••••••••
Normals
BMT-HD
O...L---oOOOC:X>OC::>----------.~......- -
of 11070. Nine of the 14patients with DAH developed respiratory failure requiring ventilatory support and died (64070 mortality). We assessed whether any of the parameters of airway inflammation pro- '". spectively recorded at the time of surveillance bronchoscopy, or other patient characteristics, were associated with the development of DAH (table 2). Wefound two variables that were predictive for the occurrence of DAH: bronchial neutrophil percentage (BrOJoPMN) and bronchial eosinophil percentage (BrOJoEos). Patients with a BrOJo PMN of 21 or more had a rate of DAH of 28070 (9/32) compared with a rate of 5070 (5/99) for the remaining patients (p = 0.003, table 3). Patients with a BrOJoEos of 1 or more had a rate of DAH of 24070 (8/34) compared with a rate of 6070 (6/97) for the remaining patients (p = 0.007, table 3). No alveolar cell parameters correlated with the development of DAH. Smoking status was also examined of the 14 patients with DAH. None werecurrent smokers, 12had
never smoked, and two had quit > 3 months before BMT. There was no correlation of smoking status with BI, inflammatory cell fractions, or the likelihood of developing DAH. Logistic regression analysis was performed to see how these variables would contribute to the prediction of the occurrence of DAH (table 4). When weconsidered a model including indicator variables of BrOJoPMN > 20 and BrOJoEos > zero, both variables contributed significantly to the prediction of DAH (BrOJoPMNs, p = 0.005; BrOJoEos, p = 0.05). Those patients with BrOJoPMN > 20070 and BrOJoEos = zero had a rate of DAH of 18070 (3/17). Those individuals with BrOJoPMN ~ 20070 and BrOJoEos ~ 1 had a rate of DAH of 10070 (2/19), but those with BrOJoPMN > 20070 and BrOJoEos ~ 1 had a rate of DAH of 43070 (6/14). The observed rate for those with bronchial PMN < 20070 and bronchial eosinophils = zero was 4070 (3/73). Exclusion of the seven patients receiving
TABLE 2 ASSOCIATION OF INFLAMMATORY VARIABLES WITH DIFFUSE ALVEOLAR HEMORRHAGE OCCURRENCE Normals (n = 19) Bronchial cell analysis Total cells x 10e/ml BAL Macrophages, 0/0 Ciliated,% Neutrophils, 0/0 Squamous, % Lymphocytes, 0/0 Eosinophils, 0/0 Alveolar cell analysis Total cells x 10e/ml BAL Macrophages, 0/0 Ciliated, 0/0 Neutrophils, 0/0 Squamous, 0/0 Lymphocytes, 0/0 Eosinophils, 0/0 Visual inflammation analysis Bronchitis Index
All BMT Pts (n = 123)
DAH+ (n = 14)
DAH(n = 109)
P Value (± DAH)
3.8 64.9 18.7 9.5 3.8 2.6 0.5
± ± ± ± ± ± ±
0.5 2.8 2.5 1.0 0.9 0.6 0.5
5.1 52.2 20.2 17.4 4.7 4.2 0.4
± ± ± ± ± ± ±
0.7 2.0 1.4 2.1 0.9 0.5 0.1
7.6 45.2 14.9 33.2 1.5 4.3 0.9
± ± ± ± ± ± ±
2.1 6.2 3.3 7.3 0.9 1.0 0.3
4.8 53.1 20.9 15.4 5.1 4.2 0.4
± ± ± ± ± ± ±
0.7 2.2 1.6 2.1 1.0 0.5 0.1
0.19 0.22 0.19 0.0061 0.19 0.96 0.019
25.2 80.4 3.8 3.0 0.0 12.3 0.5
± ± ± ± ± ± ±
3.1 2.9 0.9 0.5 0.4 2.5 0.2
27.7 74.8 6.8 5.4 0.9 10.9 0.6
± ± ± ± ± ± ±
2.2 1.7 1.0 0.9 0.4 1.0 0.2
29.1 73.2 5.1 6.2 0.2 14.6 0.6
± ± ± ± ± ± ±
5.4 3.7 1.0 1.3 0.2 3.8 0.3
27.5 75.1 7.0 5.3 0.9 10.4 0.6
± ± ± ± ± ± ±
2.4 1.8 1.1 1.1 0.5 1.0 0.2
0.82 0.73 0.53 0.76 0.59 0.18 0.87
7.5 ± 0.5
0.21
2.3 ± 0.5
7.8 ± 0.5
9.5 ± 1.5
Definitionof abbreviations: DAH = diffuse alveolarhemorrhage; SMT = bone marrowtransplantation; SAL lavage.
= bronchoalveolar
treatment regimens other than cyclophosphamide, VP-16, and BCNU did not significantly alter these findings (data not shown) . DAH is associated with a high mortality (2), and factors predictive of the occurrence of DAH would likely also predict death during the early post-BMT period. To examine this possibility, we analyzed survival using the same indicator variables found to predict the occurrence of DAH. The presence of BrOJoPMN > 20070 and'BrOJoEos ~ 1070 at the time of surveillance bronchoscopy predicted decreased survival after BMT (p = 0.001, figure 2). When survival from 3 months after BMT was examined, no significant difference in survival was apparent (p = 0.46, data not shown). Discussion
Bronchoscopy has become a commonly used procedure in the evaluation of immunocompromised hosts with clinical signs of pulmonary disease (9). In a previous report from our institution it was demonstrated that 12 of 19 (63070) BMT recipients with no clinical evidence of active pulmonary disease had significant abnormalities on bronchoscopy (3). The group of patients analyzed in that report included patients with multiple underlying malignancies and with different preBMT treatment regimens. The current study enlarges on that experience, focusing on a large group of patients with a specific diagnosis and similar treatment protocols. Surveillance bronchoscopy demonstrated a number of clinically unrecognized abnormalities, The most frequently observed findings were related to bronchial inflammation. A visual BI was elevated in the majority of BMT patients, with a score> 5 in 59070, > 10 in 33070, and> 15in 10070 of patients. In contrast, only 2 of 25 normal volunteers had a BI > 5. This suggests that mild to moderate bronchial inflammation is present in the majority of BMT patients before undergoing HCTx and BMT. The presence of intraluminal inflammatory cells is another measure of bronchial inflammation (6). Bronchial neutrophilia (> 20070 bronchial PMN) was present in a significant, but smaller, number of patients prior to BMT (31/123, 25070). With the exception of three cases, the patients with BrOJoPMN > 20 had an elevated BI. The visual index appears to identify two groups of patients with bronchial inflammation, those with associated increased bronchial inflammatory cells and those with normal bronchial cell proportions. There are several reasons why bronchial
442
SISSON, AUSTIN, THOMPSON, ET AL.
TABLE 3 ANALYSIS OF ASSOCIATION OF VARIABLES WITH DIFFUSE ALVEOLAR HEMORRHAGE OCCURRENCE No. of Patients
Range
Variable Bronchial cell, 0/0 Lymphocytes
o to 1 2 to 3 4 to 5
39 33 22 29 28 37 28 30 36 21 34 32 90 33 32 36 31 31 60 35 28
2 (5) 7 (21) 2 (9) 3 (10) 5 (18) 5 (14) 1 (4) 3 (10) 6 (17) 3 (14) 3 (9) 2 (6) 6 (7) 8 (24) 2 (6) 1 (3) 2 (6) 9 (29) 7 (12) 6 (17) 1 (4)
4 to 8 9 to 14 ~ 15 o to 1 2 to 3 4 to 9 ~ 10 o to 70 71 to 80 81 to 90 ~ 91 0 ~ 1 0 ~ 1
29 35 28 29 37 31 38 27 36 28 43 16 89 34 100 23
1 (3) 3 (9) 4 (14) 6 (21) 2 (5) 4 (13) 6 (16) 2 (7) 4 (11) 6 (21) 4 (9) 0(0) 9 (10) 5 (15) 13 (13) 1 (4)
< 100k 100 to 200k 200 to 300k 300 to 400k > 400k ~ 5k 5 to 7k 7 to 10k > 10k
15 16 24 18 20 34 25 14 29
2 (13) 3 (19) 3 (13) 3 (17) 1 (5) 4 (12) 1 (4) 3 (21) 4 (14)
~6
o to 7
Ciliated cells
8 to 18 19 to 30 ~ 31 o to 40 41 to 50 51 to 70 ~ 71 0 ~ 1 o to 2 3 to 7 8 to 20 ~ 21 0 1 to 3
Macrophages
Eosinophils Neutrophils
Squamous cells
~4
Alveolar cell, 0/0 Lymphocytes
o to 3
Ciliated cells
Macrophages
Neutrophils Squamous cells Hematologic Data Platelets, n = 93
WBC, n = 102
With DAH (%)
Definition of abbreviation: DAH
p Value
0.19
0.38
0.53 0.007
0.003
0.24
0.19
0.65
0.17 0.47 0.24
0.77
0.42
= diffuse alveolar hemorrhage.
TABLE 4 ANALYSIS OF DIFFUSE ALVEOLAR HEMORRHAGE BY PATIENT SUBGROUPS Bronchial PMN (0/0) ~
Expected DAH·
Observed DAHt
4 19 11 42
4 (3/73) 18 (3/17) 14 (2119) 43 (6/14)
Sensitivity (0/0)
Specificity (%)
Likelihood Ratio
9/14 (64) 8/14 (57)
87/109 (80) 84/109 (77)
3.2 2.5
6/14 (43)
10111 09 (93)
6.1
Bronchial Eosinophils (0/0)
20
> 20 ~
20
~
> 20
~
Test Positive Br%PMNS > 20 Br% Eos > 0 Br%PMNs > 20 and Br%Eos > 0
0 0 1 1
Definition of abbreviations: DAH = diffuse alveolar hemorrhage; PMN = polymorphonuclear leukocytes. • Predicted percentage from the logistical regression model. t Observed percentage (number with DAH/number of patients in subset).
inflammation may be present in this group of patients. First, many of these patients had received radiation therapy at some time during their treatment of Hodgkin disease. This is not a likely cause, however, because the findings were those of acute inflammation and none of the patients received radiotherapy for several months to years before BMT. Second, an infectious etiology might explain the signs of bronchitis. This is not likely inasmuch as all of the patients in this study were asymptomatic at the time of bronchoscopy and only 3 of 123patients (2.4070) from this serieshad any infectious pathogens recoveredon lavage.Third, the bronchial inflammation might be caused by cytotoxic chemotherapy. Indeed, all of these patients had receivedchemotherapy within weeks of the pre-BMT bronchoscopy. We think that this is the most likely explanation for the bronchial inflammation observed. Unfortunately, our data do not permit a quantitative correlation analysis of this factor. Outcome analysis revealed two indicators that were predictive of DAH. The fraction of bronchial neutrophils was twice as high in the patients who developed DAH than in those that did not (33.3070 versus 15.4070). The fraction of bronchial eosinophils was also higher in the patients who developed DAH than in those that did not (0.92070 versus 0.36070). Not surprisingly, these same two variables that predicted DAH were also predictive of early death. The absolute number of eosinophils in BAL fluid is usually small, and therefore the difference between zeroOJo and 1070 eosinophils is difficult to discern. In our laboratory, bronchial and alveolar cell differentials are counted for a total of 200 cells per determination and by each of two different technicians. Because our model would shift a patient from a low risk (zeroOJo BrEos) to a high risk (> 0070 BrEos) group with the presence of one or two eosinophils, werepeated the differential counts after our original data analysis by a technician unaware of the study findings (data not shown). Differential counts were performed on 13 patients selected with original bronchial neutrophils near the model threshold of 20070, including representative numbers of DAH + and DAH - patients. The repeat counts showed similar eosinophil counts and did not shift any patients from low to high risk who did not develop DAH (false positives) or from high to low risk who did develop DAH (false negatives). This analysis suggests that the finding of small numbers of bronchial eosinophils is reproducible
443
BRONCHITIS, BMT, AND DAH IN HODGKIN DISEASE PATIENTS
Fig. 2. Probability ofsurvival in BMT patients based on pretreatment bronchial Ok PMN and Ok eosinophils. Probability is displayedon the verticalaxis,and time in months after transplantation (day 0) is displayed on the horizontalaxis. The solid line represents the patients with BrOkPMN < 20 and BrOkEos < 1, the small dashed line represents the patients withBrOkPMN < 20 and BrOkEos .. 1, the mediumdashed line represents the patients with BrOkPMN .. 20 and BrOkEos < 1. and the large dashed line represents the patients with BrOkPMN .. 20 and BrOkEos .. 1. Small vertical lines on each horizontalline represent the survivingpatients at the time of last follow-up.
ing BMT. Further studies are indicated to determine whether these criteria predict for the occurrence of DAH in other groups of BMT candidates, and, if applied prospectively, whether intervention with anti-inflammatory drugs or other agents can alter the apparent risk before proceeding with BMT.
10 ':
08
1
l
>-
D
,g
I
'U.... ~ Jt.,-- - ,
0.6
o
cl:
~. ~
:..·tH ' 14H-f--i, 11_ ;........ ·I..· t·~~ ~~ ~~~~·~,
:..
1 L_l,
0.4
l_.,
··· ·1
'-------- f--- j
l
en
. ·..·
:
1
02
L---t - - - - - -I I i i
o
in this patient population and consistent with the concept of the role eosinophils are thought to play in the pathogenesis of acute lung injury (10). It is attractive to hypothesize that the presence of increased numbers of inflammatory cells in the airways indicates a heightened "inflammatory state" compared with those patients who have only visual signs of airway inflammation. This would imply that the inflammatory cell burden of the airways before HCTx contributes to the development of DAH. The concept that the neutrophil may be important in the pathogenesis of DAH is supported by several observations. First, Robbins and colleagues observed that DAH usually becomes clinically manifest at a time during marrow recovery in which circulating neutrophils first start to reappear after aplasia (2). Second, corticosteroids are anti-inflammatory compounds that are known to attenuate neutrophil function. Recent reports indicate that the administration of high-dose corticosteroids to patients with early DAH may favorably modify the natural history of the syndrome and appears to improve outcome (11, 12). Third, in adult respiratory distress syndrome, another diffuse lung injury syndrome with many clinical findings similar to DAH, the neutrophil has been implicated in the amplification of previous lung injury (13). Last, animal models of acute lung injury have demonstrated that the degree of clinical manifestations after initial lung injury are largely dependent on the circulating neutrophil pool for full expression of the injury (14). It is interesting that the bronchial and not the alveolar inflammatory cell population was elevated in those patients at risk for DAH. One interpretation of this
i
i
'
iii iii
10
i
I
I I
I i
j
il'
Iii iii i
20
'I 30
f
I i
I Iii
i,
I 40
Time (monlhs)
Acknowledgment The writers thank Robin Zotti and Theresa McKillip for their valuable assistance in reviewing patient records and entering data . References
observation is that the development of DAH involves bronchial inflammation before alveolar inflammation. Antecedent bronchial injury might therefore be important in the pathogenesis of this syndrome, even though alveolar abnormalities are lacking at the time of bronchoscopy. Whether this injury is caused by the direct effects of high-dose chemotherapy, infectious agents, or other factors is not known. Alternatively, it is possible that the site of bleeding in DAH is the small airway and not the alveolus. The bronchoscopic lavage technique we employ separates proximal ("bronchial") from distal ("alveolar") fluid, but cannot precisely locate either the site of bleeding in DAH or the source of neutrophils observed in the two fractions. It is conceivable that when DAH occurs, both the site of neutrophil accumulation and the source of hemorrhage are the small airways. The alveolar collection of blood observed during lavage may be simply the result of distal accumulation of blood and not the actual site of injury. Unfortunately, the study of lungs from DAH patients at autopsy has not revealed an obvious site of bleeding. In conclusion, our results confirm that bronchoscopy detects visual evidence of airway inflammation in a majority of patients and cytologic evidence of inflammation in a smaller subpopulation. In addition, the subpopulation of patients with increased inflammatory airway cells appears to be at significantly increased risk for the development of pulmonary complications, including DAH , respiratory failure, and death. The finding of airway inflammation before HCTx and bone marrow infusion appears to detect this high-risk group of patients among those with Hodgkin disease undergo-
I. Buckner CD, Meyers JD, Springmeyer SC. Pul monary complications of marrow transplantation, reviewof the Seattle experience. Exp Hematol1984; 12:1-5. 2. Robbins RA , Linder JL, Stahl MG, et at. Diffuse alveolar hemorrhage in autologous bone mar row transplantation. Am J Med 1989; 87:511-8. 3. Vaughan WP, Linder J , Robbins R, Arneson M, Rennard S1. Pulmonary surveillance using bronchoscopy and bronchoalveolar lavage during high dose antineoplastic therapy. Chest 1991; 99:105-11. 4. Jagannath S, Armitage JO, Dicke KA, e/ at. Prognostic factors for response and surv ival after high-dose cyclophosphamide, carmustine, and etopos ide with autologous bone marrow transplantation for relapsed Hodgkin's disease. J Clin Oncol 1989; 7:179-85. 5. Rennard SI, Ghafouri M, Thompson AB, et at. Fractional processing of sequential bronchoalveolar lavage to separate bronchial and alveolar samples. Am Rev Resp Dis 1990; 141:208-17. 6. Thompson AB, Daughton D, Robbins RA , Ghafouri MA, Oehlerking M, Rennard SI. Intraluminal airway inflammation in chronic bronchitis. Characterization and correlation with clinical parameters. Am Rev Resp Dis 1989; 140:1527-37. 7. Kaplan GL, Meir P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53:457-81. 8. Peto R, Pike M, Arm itage P, et at. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II : analysis and examples . Br J Cancer 1977; 35:1-39. 9. Emanuel D, Peppard J, Stover D, Gold J, Armstrong D, Hammerling U. Rapid immunodiagnosis of cytomegalovirus pneumonia by bronchoalveolar lavage using human and murine monoclonal antibodies. Ann Intern Med 1986; 104:476-81. 10. O'Brodovich H . The eosinophil and acute lung injury. Am Rev Respir Dis 1990; 142:1245-6. 11. Metcalf J, Armitage J, Arneson M, et al. The effect of glucocorticoids on survival and development of subsequent opportunistic infections in bone marrow transplant patients with diffuse alveolar hemorrhage. Am Rev Respir Dis 1991; 143:A474. 12. ChaoNJ, DuncanSR, LongGD, HorningSJ, Blume KG. Corticosteroid therapy for diffuse alveolar hemorrhage in autologous bone marrow recipients. Ann Intern Med 1991; 114:145-6. 13. Tate RM, Repine JE . Neutrophils and the adult respiratory distress syndrome. Am Rev Respir Dis 1983; 128:552-9. 14. Heflin ACJ , Brigham KL. Prevention by granulocyte depletion of increased vascular permeability of sheep lung following endotoxemia. J Clin Invest 1981; 68:1253-60 .
JOSEPH H. SISSON, AUSTIN B. THOMPSON, JAMES R. ANDERSON, RICHARD A. ROBBINS, JOHN R. SPURZEM, PAUL R. SPENCE, ELIZABETH C. REED, JAMES O. ARMITAGE, JULIE M. VOSE, MARK A. ARNESON, WILLIAM P. VAUGHAN, and STEPHEN I. RENNARD
Introduction
High-dose chemotherapy (HCTx) supported by bone marrow transplantation (BMT) has become an increasingly im-" portant curative-intent treatment modality for patients with malignancies resistant to conventional chemotherapy and radiation treatment. A major dose-limiting toxicity of this approach, however, is pulmonary failure. Between 10 and 200/0 of patients undergoing BMT die of pulmonary complications during their initial hospitalization (1). We recently described a diffuse alveolar hemorrhage syndrome (DAH) associated with an 800/0 mortality that occurred in approximately 200/0 of patients during autologous BMT (2). Although several factors such as age, temperature, platelet count, and oral mucositis score at the onset of the syndrome appeared to be associated with the development of DAH, no parameters were identified prior to the administration of HCTx and marrow infusion that were predictive of subsequent DAH and death. Because DAH was associated with mucositis at the time of the onset of the syndrome, we hypothesized that airway inflammation present before HCTx and BMT might predict for the development of DAH. We therefore prospectively examined a large number of patients at risk for DAH before HCTx and BMT by bronchoscopy with bronchoalveolar lavage (BAL) to identify factors in the lower respiratory tract that predicted for the development of DAH. Wereport that patients with Hodgkin's disease have evidenceof airwayinjury before the initiation of HCTx and BMT and that increased proportions of airway neutrophils and eosinophils are predictive for the occurrence of DAH and death.
SUMMARY We determined risk factors present In patients with Hodgkin disease that predicted the development of diffuse alveolar hemorrhage syndrome (DAH) during autologous bone marrow transplantation (BMT). One hundred twenty-three patients with Hodgkin disease prospectively underwent bronchoscopy with bronchoalveolar lavage (BAL) before receiving BMT.The bronchitis Index (BI) of the airways and bronchial and alveolar cell counts and differentials were determined In all patients and compared with 20 normal nonsmoking volunteers. logistic regression analysis was used to determine factors that predicted for the development of DAH. Visual evidence of bronchial Injury was observed regardless of smoking history (BI 7.8 ± 0.5 for BMT versus 2.3 ± 0.5 for volunteers, p 0.001). BMT patients who developed DAH (n 14) had significantly greater 109) patients numbers of bronchial neutrophlls and eoslnophlls compared with DAH-negatlve (n 0.006; bronchial (bronchial polymorphonuclear leukocytes (PMN), 33 ± 7% versus 14 ± 2%, P eoslnophlls,0.9 ± 0.3% versus 0.4% ± 0.07%, P = 0.02). logistic regression analysis revealed that the presence of bronchial PMN > 20% or bronchial eoslnophlls > zero% were predictive of DAH (p = 0.005 and 0.05, respectively). When both predictors were positive, the rate of DAH was 10 times greater than when both predictors were negative (43% versus 4% DAH occurrence). Survival was also sign Iflcantly reduced when these predictors were positive. This study demonstrates that bronchial Inflammation Is present with or without Intraluminal Inflammatory cells In the maJority of patients with Hodgkin disease before BMT. The subgroup of these patients with Increased bronchial Inflammatory cells are at greatly Increased risk for development of DAH and death.
=
=
=
=
=
AM REV RESPIR DIS 1992; 146:439-443
Methods Patients From January 1, 1987through March 31, 1990 a total of 143 patients with the diagnosis of Hodgkin disease were treated with HCT x and BMT at the University of Nebraska Medical Center, Omaha. Surveillance bronchoscopy with BAL was performed on 123of these patients prior to the initiation of HCTx and within 30 days of bone marrow infusion. Surveillance bronchoscopy, performed in the absence of clinical or radiographic evidence of active lung disease, has been shown to detect occult pulmonary abnormalities (3). The bronchoscopy findings of the BMT patients (age 32.25 ± 0.75 yr; mean ± SE) were compared with a group of 20 normal nonsmoking volunteers (age 33.1 ± 2.1 yr).
High Dose Chemotherapy and 'Iransplantation Protocol HCTx consisted of cyclophosphamide, etoposide (VP-16) and carmustine (BCNU) followed by autologous bone marrow infusion
(Received in original form August 19, 1991 and in revised form January 17, 1992) 1 From the Pulmonary/Critical Care and Oncology/Hematology Sections, Department of Medicine, and the Department of Preventive and Societal Medicine, University of Nebraska MedicalCenter, Omaha, Nebraska. 2 Correspondence and requests for reprints should be addressed to Joseph H. Sisson, M.D., Department of Internal Medicine, Pulmonary and Critical Care Medicine Section, University of Nebraska Medical Center, 600 South 42nd Street, Omaha, NE 68198-2465. 3 In addition to the above-named authors, the University of Nebraska Medical Center Bone Marrow Transplantation/Pulmonary Study Group includes the following individuals: P. Bierman, R. Cheney, A. Curseen, D. Darrington, K. Dicke, B. Edwards, J. Foley,T. Franco-Crowley,B. Frappier, J. Goldsmith, J. Gurney, W. Haire, W. Huerta, K. Jones, A. Kessinger, L. Klassen, J. Linder, J. Metcalf, J. Pierson, M. Raz, K. Rickard, D. Romberger, I. Rubinstein, K. Schidt-Pokorny, M. Shumway, J. Spinolo, M. Tempero, S. Von Essen, T. Walter, E. Wedergren, W. West, J. Wisecarver, R. Witte, G. Woods, and R. Yates.
439
440
SISSON, AUSTIN, THOMPSON, ET AL.
(4) in 116patients (94.3070). Six other patients (4.9070) received slightly different chemotherapy protocols followed by autologous bone marrow infusion. One additional patient received busulfan, cyclophosphamide, and VP16followed by allogeneic bone marrow infusion. All patients were included in the data analysis since exclusion of the seven patients with alternative chemotherapy protocols did not affect the results (data not shown).
Bronchoscopy and Bronchoalveolar Lavage Patients underwent surveillance bronchoscopy and BAL after giving written informed consent at the time of admission. Bronchoscopy and BAL were performed as previously described (5). Briefly, bronchoscopy was performed by inserting the end 0 f a bronchoscope (Olympus IT-R or PO-I0; Olympus Corp. of America, New Hyde Park, NY) transorally. A visual score for each of four characteristics (erythema, edema, secretions, and friability) was determined by assigning a value of zero to 3 (zero = normal, 3 = severely abnormal). This score was determined for each of the six major bronchi, and the bronchitis index (BI) represents the sum of these scores (6). The bronchoscope was gently advanced into a wedged position and five 20-ml aliquots of sterile saline were infused through the bronchoscope and immediately aspirated. Lavage was performed in each of three sites and usually included the right middle lobe, the lingula, and one of the lower lobes. The bronchial sample (first 20-ml aliquot) waskept separate from the alveolarsample (subsequent four 20 ml aliquots [5J). Bronchoalveolar Lavage Processing and Cytologic Examination Both the bronchial and alveolar samples were processedas previouslydescribed (5).The cells wereprepared for cytologicexamination by cytocentrifugation (Cytospin; Shandon Southern Instruments, Sewickley, PA). Celldifferentials were performed from the modified Wright-Giemsa-stained samples (Diff-Quik~; American Scientific, McGraw Park, IL). Papanicolaou stains and Gomori methenamine silver stained slides were also prepared for cytologic examination. Statistical Analysis Comparisons of the distribution of values of
continuous variables between subject groups the major bronchi, a semiquantitative BI were made using independent sample Stu- was determined (6). The BI was signifident's t tests, and the values were expressed cantly elevated in the patients undergoas means ± standard errors. The association ing BMT compared with normal nonof bronchoscopy and BAL data with the sub- smokers (7.9 ± 0.5 versus 2.3 ± 0.5, p = sequent occurrence of OAH was assessed using appropriate chi-squared tests for con- 0.(01) (figure 1). An elevated BI was prestingency tables. Multivariate assessment of ent regardless of the smoking status of the relationship of various bronchial and al- the patients at the time of admission for veolar percentages and the occurrence of BMT (8.8 ± 1.6 for smokers and 7.9 ± OAH was made through the use of logistic 0.6 for nonsmokers; p = 0.62). regression. Survival was defined as the time Cell Count and Differential Findings from transplantation to death from any cause. The distribution of survival times was esti- An elevated BI has been associated with mated using the product-limit method of the presence of inflammatory cells in the Kaplan and Meir (7). The statistical sig- airways of patients with chronic bronnificance of differences observed in survival chitis (6). We therefore examined the between patient subgroups was assessedusing the log-rank test (8). All p values reported are . bronchial and alveolar lavage fluid of two-sided. All data analyses were done using BMT patients for the presence of inflamstatistical analysis software (SAS Institute, matory cells (table 2). Bronchial neutrophilia (bronchial polymorphonuclear Inc., Cary, NC). leukocytes [PMN] > 20070 [6]) was present in 31 of 123 patients (25070). When Results all BMT patients were analyzed for bronDemographics chial neutrophilia, however, the mean One hundred forty-three patients with the bronchial neutrophil fraction was highdiagnosis of Hodgkin disease were ader than that observed in the nonsmoker mitted to the University of Nebraska control subjects, but the difference did Medical Center between January 1, 1987 not reach statistical significance (BMT and March 16,1990 for HCTx and BMT patients = 17.4 ± 2.0 versus normal sub(table 1). Twenty patients were excluded = 9.5 ± 1.0, p = 0.14). The degree jects from analysis because (1) active pulmoof bronchial neutrophilia did correlate nary disease was apparent clinically (n = with the visual BI (r = 0.46, 2), (2) their surveillance bronchoscopy positively 2 was performed outside the 30 day pre- r = 0.21, p = 0.0001). No other signifiBMT period (n = 5), (3) the bronchos- cant change in the proportion of cells recopy was performed after chemothera- covered was observed when all patients py was started (n = 5), or (4) bronchos- were analyzed (data not shown), with one copy was refused (n = 8). Thus, the exception, the fraction of alveolar macresults of 123 (86070) surveillance bron- rophages from the bronchial samples was choscopies were available for analysis. significantly lower in BMT patients than The study population comprised 73 men in normal nonsmoking control subjects (59070) and 50 women (41070), mean age, (p = 0.03). These findings were true 32.3 ± 0.75 yr. Bronchoscopy was per- regardless of whether the fraction of cells formed before the initiation of HCTx on or the absolute number of cells for each average approximately 1 wk before BMT cell type was compared.
(Day - 8.92 ± 0.40 days; range, Day - 29 to Day -5).
Bronchoscopy Findings To measure the degree of large airway inflammation present upon inspection of
TABLE 1 PATIENTS WITH BONE MARROW TRANSPLANTATION-HODGKIN DISEASE UNDERGOING SURVEILLANCE BRONCHOSCOPY Total patients for BMT Patients with surveillance bronchoscopy, n (0/0) Males, n (Ok) Females, n (0/0) Current smokers, n (Ok) Nonsmokers, n (Ok) Required ventilator during BMT, n (Ok) Developed DAH during BMT, n (0/0) Died during BMT, n (Ok) Age, yr (mean ± SE) Day of bronchoscopy relative to BMT (mean ± SE) Dsf/n/t/on of abbreviations: BMT
= bone marrow transplantation;
143 123 (100) 73 (59) 50 (41) 12 (10) 111 (90) 20 (16) 14 (11) 17 (14) 32.25 :t 0.75 - 8.92 :t 0.40 DAH
Autopsy Results of Patients Dying with DAB Because bronchial airway inflammation and not alveolar inflammation appeared to correlate with the development of DAH, we reviewed the available autopsies to ascertain whether a bleeding site could be determined. Autopsies were performed on four of nine patients (44070) who died with DAH. Extensive diffuse and patchy hemorrhage and alveolar damage was present in all four cases. No airway ulcerations were identified and no consistent site of bleeding was apparent.
Outcome Analysis (range, 17 to 54 yr) (range, - 29 to - 5 days)
= diffuse alveolar hemorrhage syndrome.
DAH is a common, life-threatening pulmonary complication encountered during autologous BMT (2). DAH occurred in 14 of 123 patients with an incidence
441
BRONCHITIS, BMT, AND DAH IN HODGKIN DISEASE PATIENTS
Fig. 1. Comparison of bronchitis index (BI) in normal volu nteers with bone marrow transplantation (BMT) patients prior to high-dose chemotherapy. The BI is a visual score of inflammatory findings at bronchoscopy. The vertical axis represents the sum of the visual score from the six major bronchi before bronchoalveolar lavage. The open circles represent the BI of normal volunteers (n = 25);the closed circles represent the BMT patients (n = 123) with Hodgkin's disease.
25
.
20
.....................................................................•..................
15
••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• ;~.~.:~~.~.~ •••••••••••.
•• • •••
...
~
G ~
••
! ~
u
. . ••::::••
0
.:::::.
10
~
m
o
5
••••••••••••
~
00 000
.
.
•••••• ••••••••••
O&X)
•••••••
ססoo
••••••••••
Normals
BMT-HD
O...L---oOOOC:X>OC::>----------.~......- -
of 11070. Nine of the 14patients with DAH developed respiratory failure requiring ventilatory support and died (64070 mortality). We assessed whether any of the parameters of airway inflammation pro- '". spectively recorded at the time of surveillance bronchoscopy, or other patient characteristics, were associated with the development of DAH (table 2). Wefound two variables that were predictive for the occurrence of DAH: bronchial neutrophil percentage (BrOJoPMN) and bronchial eosinophil percentage (BrOJoEos). Patients with a BrOJo PMN of 21 or more had a rate of DAH of 28070 (9/32) compared with a rate of 5070 (5/99) for the remaining patients (p = 0.003, table 3). Patients with a BrOJoEos of 1 or more had a rate of DAH of 24070 (8/34) compared with a rate of 6070 (6/97) for the remaining patients (p = 0.007, table 3). No alveolar cell parameters correlated with the development of DAH. Smoking status was also examined of the 14 patients with DAH. None werecurrent smokers, 12had
never smoked, and two had quit > 3 months before BMT. There was no correlation of smoking status with BI, inflammatory cell fractions, or the likelihood of developing DAH. Logistic regression analysis was performed to see how these variables would contribute to the prediction of the occurrence of DAH (table 4). When weconsidered a model including indicator variables of BrOJoPMN > 20 and BrOJoEos > zero, both variables contributed significantly to the prediction of DAH (BrOJoPMNs, p = 0.005; BrOJoEos, p = 0.05). Those patients with BrOJoPMN > 20070 and BrOJoEos = zero had a rate of DAH of 18070 (3/17). Those individuals with BrOJoPMN ~ 20070 and BrOJoEos ~ 1 had a rate of DAH of 10070 (2/19), but those with BrOJoPMN > 20070 and BrOJoEos ~ 1 had a rate of DAH of 43070 (6/14). The observed rate for those with bronchial PMN < 20070 and bronchial eosinophils = zero was 4070 (3/73). Exclusion of the seven patients receiving
TABLE 2 ASSOCIATION OF INFLAMMATORY VARIABLES WITH DIFFUSE ALVEOLAR HEMORRHAGE OCCURRENCE Normals (n = 19) Bronchial cell analysis Total cells x 10e/ml BAL Macrophages, 0/0 Ciliated,% Neutrophils, 0/0 Squamous, % Lymphocytes, 0/0 Eosinophils, 0/0 Alveolar cell analysis Total cells x 10e/ml BAL Macrophages, 0/0 Ciliated, 0/0 Neutrophils, 0/0 Squamous, 0/0 Lymphocytes, 0/0 Eosinophils, 0/0 Visual inflammation analysis Bronchitis Index
All BMT Pts (n = 123)
DAH+ (n = 14)
DAH(n = 109)
P Value (± DAH)
3.8 64.9 18.7 9.5 3.8 2.6 0.5
± ± ± ± ± ± ±
0.5 2.8 2.5 1.0 0.9 0.6 0.5
5.1 52.2 20.2 17.4 4.7 4.2 0.4
± ± ± ± ± ± ±
0.7 2.0 1.4 2.1 0.9 0.5 0.1
7.6 45.2 14.9 33.2 1.5 4.3 0.9
± ± ± ± ± ± ±
2.1 6.2 3.3 7.3 0.9 1.0 0.3
4.8 53.1 20.9 15.4 5.1 4.2 0.4
± ± ± ± ± ± ±
0.7 2.2 1.6 2.1 1.0 0.5 0.1
0.19 0.22 0.19 0.0061 0.19 0.96 0.019
25.2 80.4 3.8 3.0 0.0 12.3 0.5
± ± ± ± ± ± ±
3.1 2.9 0.9 0.5 0.4 2.5 0.2
27.7 74.8 6.8 5.4 0.9 10.9 0.6
± ± ± ± ± ± ±
2.2 1.7 1.0 0.9 0.4 1.0 0.2
29.1 73.2 5.1 6.2 0.2 14.6 0.6
± ± ± ± ± ± ±
5.4 3.7 1.0 1.3 0.2 3.8 0.3
27.5 75.1 7.0 5.3 0.9 10.4 0.6
± ± ± ± ± ± ±
2.4 1.8 1.1 1.1 0.5 1.0 0.2
0.82 0.73 0.53 0.76 0.59 0.18 0.87
7.5 ± 0.5
0.21
2.3 ± 0.5
7.8 ± 0.5
9.5 ± 1.5
Definitionof abbreviations: DAH = diffuse alveolarhemorrhage; SMT = bone marrowtransplantation; SAL lavage.
= bronchoalveolar
treatment regimens other than cyclophosphamide, VP-16, and BCNU did not significantly alter these findings (data not shown) . DAH is associated with a high mortality (2), and factors predictive of the occurrence of DAH would likely also predict death during the early post-BMT period. To examine this possibility, we analyzed survival using the same indicator variables found to predict the occurrence of DAH. The presence of BrOJoPMN > 20070 and'BrOJoEos ~ 1070 at the time of surveillance bronchoscopy predicted decreased survival after BMT (p = 0.001, figure 2). When survival from 3 months after BMT was examined, no significant difference in survival was apparent (p = 0.46, data not shown). Discussion
Bronchoscopy has become a commonly used procedure in the evaluation of immunocompromised hosts with clinical signs of pulmonary disease (9). In a previous report from our institution it was demonstrated that 12 of 19 (63070) BMT recipients with no clinical evidence of active pulmonary disease had significant abnormalities on bronchoscopy (3). The group of patients analyzed in that report included patients with multiple underlying malignancies and with different preBMT treatment regimens. The current study enlarges on that experience, focusing on a large group of patients with a specific diagnosis and similar treatment protocols. Surveillance bronchoscopy demonstrated a number of clinically unrecognized abnormalities, The most frequently observed findings were related to bronchial inflammation. A visual BI was elevated in the majority of BMT patients, with a score> 5 in 59070, > 10 in 33070, and> 15in 10070 of patients. In contrast, only 2 of 25 normal volunteers had a BI > 5. This suggests that mild to moderate bronchial inflammation is present in the majority of BMT patients before undergoing HCTx and BMT. The presence of intraluminal inflammatory cells is another measure of bronchial inflammation (6). Bronchial neutrophilia (> 20070 bronchial PMN) was present in a significant, but smaller, number of patients prior to BMT (31/123, 25070). With the exception of three cases, the patients with BrOJoPMN > 20 had an elevated BI. The visual index appears to identify two groups of patients with bronchial inflammation, those with associated increased bronchial inflammatory cells and those with normal bronchial cell proportions. There are several reasons why bronchial
442
SISSON, AUSTIN, THOMPSON, ET AL.
TABLE 3 ANALYSIS OF ASSOCIATION OF VARIABLES WITH DIFFUSE ALVEOLAR HEMORRHAGE OCCURRENCE No. of Patients
Range
Variable Bronchial cell, 0/0 Lymphocytes
o to 1 2 to 3 4 to 5
39 33 22 29 28 37 28 30 36 21 34 32 90 33 32 36 31 31 60 35 28
2 (5) 7 (21) 2 (9) 3 (10) 5 (18) 5 (14) 1 (4) 3 (10) 6 (17) 3 (14) 3 (9) 2 (6) 6 (7) 8 (24) 2 (6) 1 (3) 2 (6) 9 (29) 7 (12) 6 (17) 1 (4)
4 to 8 9 to 14 ~ 15 o to 1 2 to 3 4 to 9 ~ 10 o to 70 71 to 80 81 to 90 ~ 91 0 ~ 1 0 ~ 1
29 35 28 29 37 31 38 27 36 28 43 16 89 34 100 23
1 (3) 3 (9) 4 (14) 6 (21) 2 (5) 4 (13) 6 (16) 2 (7) 4 (11) 6 (21) 4 (9) 0(0) 9 (10) 5 (15) 13 (13) 1 (4)
< 100k 100 to 200k 200 to 300k 300 to 400k > 400k ~ 5k 5 to 7k 7 to 10k > 10k
15 16 24 18 20 34 25 14 29
2 (13) 3 (19) 3 (13) 3 (17) 1 (5) 4 (12) 1 (4) 3 (21) 4 (14)
~6
o to 7
Ciliated cells
8 to 18 19 to 30 ~ 31 o to 40 41 to 50 51 to 70 ~ 71 0 ~ 1 o to 2 3 to 7 8 to 20 ~ 21 0 1 to 3
Macrophages
Eosinophils Neutrophils
Squamous cells
~4
Alveolar cell, 0/0 Lymphocytes
o to 3
Ciliated cells
Macrophages
Neutrophils Squamous cells Hematologic Data Platelets, n = 93
WBC, n = 102
With DAH (%)
Definition of abbreviation: DAH
p Value
0.19
0.38
0.53 0.007
0.003
0.24
0.19
0.65
0.17 0.47 0.24
0.77
0.42
= diffuse alveolar hemorrhage.
TABLE 4 ANALYSIS OF DIFFUSE ALVEOLAR HEMORRHAGE BY PATIENT SUBGROUPS Bronchial PMN (0/0) ~
Expected DAH·
Observed DAHt
4 19 11 42
4 (3/73) 18 (3/17) 14 (2119) 43 (6/14)
Sensitivity (0/0)
Specificity (%)
Likelihood Ratio
9/14 (64) 8/14 (57)
87/109 (80) 84/109 (77)
3.2 2.5
6/14 (43)
10111 09 (93)
6.1
Bronchial Eosinophils (0/0)
20
> 20 ~
20
~
> 20
~
Test Positive Br%PMNS > 20 Br% Eos > 0 Br%PMNs > 20 and Br%Eos > 0
0 0 1 1
Definition of abbreviations: DAH = diffuse alveolar hemorrhage; PMN = polymorphonuclear leukocytes. • Predicted percentage from the logistical regression model. t Observed percentage (number with DAH/number of patients in subset).
inflammation may be present in this group of patients. First, many of these patients had received radiation therapy at some time during their treatment of Hodgkin disease. This is not a likely cause, however, because the findings were those of acute inflammation and none of the patients received radiotherapy for several months to years before BMT. Second, an infectious etiology might explain the signs of bronchitis. This is not likely inasmuch as all of the patients in this study were asymptomatic at the time of bronchoscopy and only 3 of 123patients (2.4070) from this serieshad any infectious pathogens recoveredon lavage.Third, the bronchial inflammation might be caused by cytotoxic chemotherapy. Indeed, all of these patients had receivedchemotherapy within weeks of the pre-BMT bronchoscopy. We think that this is the most likely explanation for the bronchial inflammation observed. Unfortunately, our data do not permit a quantitative correlation analysis of this factor. Outcome analysis revealed two indicators that were predictive of DAH. The fraction of bronchial neutrophils was twice as high in the patients who developed DAH than in those that did not (33.3070 versus 15.4070). The fraction of bronchial eosinophils was also higher in the patients who developed DAH than in those that did not (0.92070 versus 0.36070). Not surprisingly, these same two variables that predicted DAH were also predictive of early death. The absolute number of eosinophils in BAL fluid is usually small, and therefore the difference between zeroOJo and 1070 eosinophils is difficult to discern. In our laboratory, bronchial and alveolar cell differentials are counted for a total of 200 cells per determination and by each of two different technicians. Because our model would shift a patient from a low risk (zeroOJo BrEos) to a high risk (> 0070 BrEos) group with the presence of one or two eosinophils, werepeated the differential counts after our original data analysis by a technician unaware of the study findings (data not shown). Differential counts were performed on 13 patients selected with original bronchial neutrophils near the model threshold of 20070, including representative numbers of DAH + and DAH - patients. The repeat counts showed similar eosinophil counts and did not shift any patients from low to high risk who did not develop DAH (false positives) or from high to low risk who did develop DAH (false negatives). This analysis suggests that the finding of small numbers of bronchial eosinophils is reproducible
443
BRONCHITIS, BMT, AND DAH IN HODGKIN DISEASE PATIENTS
Fig. 2. Probability ofsurvival in BMT patients based on pretreatment bronchial Ok PMN and Ok eosinophils. Probability is displayedon the verticalaxis,and time in months after transplantation (day 0) is displayed on the horizontalaxis. The solid line represents the patients with BrOkPMN < 20 and BrOkEos < 1, the small dashed line represents the patients withBrOkPMN < 20 and BrOkEos .. 1, the mediumdashed line represents the patients with BrOkPMN .. 20 and BrOkEos < 1. and the large dashed line represents the patients with BrOkPMN .. 20 and BrOkEos .. 1. Small vertical lines on each horizontalline represent the survivingpatients at the time of last follow-up.
ing BMT. Further studies are indicated to determine whether these criteria predict for the occurrence of DAH in other groups of BMT candidates, and, if applied prospectively, whether intervention with anti-inflammatory drugs or other agents can alter the apparent risk before proceeding with BMT.
10 ':
08
1
l
>-
D
,g
I
'U.... ~ Jt.,-- - ,
0.6
o
cl:
~. ~
:..·tH ' 14H-f--i, 11_ ;........ ·I..· t·~~ ~~ ~~~~·~,
:..
1 L_l,
0.4
l_.,
··· ·1
'-------- f--- j
l
en
. ·..·
:
1
02
L---t - - - - - -I I i i
o
in this patient population and consistent with the concept of the role eosinophils are thought to play in the pathogenesis of acute lung injury (10). It is attractive to hypothesize that the presence of increased numbers of inflammatory cells in the airways indicates a heightened "inflammatory state" compared with those patients who have only visual signs of airway inflammation. This would imply that the inflammatory cell burden of the airways before HCTx contributes to the development of DAH. The concept that the neutrophil may be important in the pathogenesis of DAH is supported by several observations. First, Robbins and colleagues observed that DAH usually becomes clinically manifest at a time during marrow recovery in which circulating neutrophils first start to reappear after aplasia (2). Second, corticosteroids are anti-inflammatory compounds that are known to attenuate neutrophil function. Recent reports indicate that the administration of high-dose corticosteroids to patients with early DAH may favorably modify the natural history of the syndrome and appears to improve outcome (11, 12). Third, in adult respiratory distress syndrome, another diffuse lung injury syndrome with many clinical findings similar to DAH, the neutrophil has been implicated in the amplification of previous lung injury (13). Last, animal models of acute lung injury have demonstrated that the degree of clinical manifestations after initial lung injury are largely dependent on the circulating neutrophil pool for full expression of the injury (14). It is interesting that the bronchial and not the alveolar inflammatory cell population was elevated in those patients at risk for DAH. One interpretation of this
i
i
'
iii iii
10
i
I
I I
I i
j
il'
Iii iii i
20
'I 30
f
I i
I Iii
i,
I 40
Time (monlhs)
Acknowledgment The writers thank Robin Zotti and Theresa McKillip for their valuable assistance in reviewing patient records and entering data . References
observation is that the development of DAH involves bronchial inflammation before alveolar inflammation. Antecedent bronchial injury might therefore be important in the pathogenesis of this syndrome, even though alveolar abnormalities are lacking at the time of bronchoscopy. Whether this injury is caused by the direct effects of high-dose chemotherapy, infectious agents, or other factors is not known. Alternatively, it is possible that the site of bleeding in DAH is the small airway and not the alveolus. The bronchoscopic lavage technique we employ separates proximal ("bronchial") from distal ("alveolar") fluid, but cannot precisely locate either the site of bleeding in DAH or the source of neutrophils observed in the two fractions. It is conceivable that when DAH occurs, both the site of neutrophil accumulation and the source of hemorrhage are the small airways. The alveolar collection of blood observed during lavage may be simply the result of distal accumulation of blood and not the actual site of injury. Unfortunately, the study of lungs from DAH patients at autopsy has not revealed an obvious site of bleeding. In conclusion, our results confirm that bronchoscopy detects visual evidence of airway inflammation in a majority of patients and cytologic evidence of inflammation in a smaller subpopulation. In addition, the subpopulation of patients with increased inflammatory airway cells appears to be at significantly increased risk for the development of pulmonary complications, including DAH , respiratory failure, and death. The finding of airway inflammation before HCTx and bone marrow infusion appears to detect this high-risk group of patients among those with Hodgkin disease undergo-
I. Buckner CD, Meyers JD, Springmeyer SC. Pul monary complications of marrow transplantation, reviewof the Seattle experience. Exp Hematol1984; 12:1-5. 2. Robbins RA , Linder JL, Stahl MG, et at. Diffuse alveolar hemorrhage in autologous bone mar row transplantation. Am J Med 1989; 87:511-8. 3. Vaughan WP, Linder J , Robbins R, Arneson M, Rennard S1. Pulmonary surveillance using bronchoscopy and bronchoalveolar lavage during high dose antineoplastic therapy. Chest 1991; 99:105-11. 4. Jagannath S, Armitage JO, Dicke KA, e/ at. Prognostic factors for response and surv ival after high-dose cyclophosphamide, carmustine, and etopos ide with autologous bone marrow transplantation for relapsed Hodgkin's disease. J Clin Oncol 1989; 7:179-85. 5. Rennard SI, Ghafouri M, Thompson AB, et at. Fractional processing of sequential bronchoalveolar lavage to separate bronchial and alveolar samples. Am Rev Resp Dis 1990; 141:208-17. 6. Thompson AB, Daughton D, Robbins RA , Ghafouri MA, Oehlerking M, Rennard SI. Intraluminal airway inflammation in chronic bronchitis. Characterization and correlation with clinical parameters. Am Rev Resp Dis 1989; 140:1527-37. 7. Kaplan GL, Meir P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53:457-81. 8. Peto R, Pike M, Arm itage P, et at. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II : analysis and examples . Br J Cancer 1977; 35:1-39. 9. Emanuel D, Peppard J, Stover D, Gold J, Armstrong D, Hammerling U. Rapid immunodiagnosis of cytomegalovirus pneumonia by bronchoalveolar lavage using human and murine monoclonal antibodies. Ann Intern Med 1986; 104:476-81. 10. O'Brodovich H . The eosinophil and acute lung injury. Am Rev Respir Dis 1990; 142:1245-6. 11. Metcalf J, Armitage J, Arneson M, et al. The effect of glucocorticoids on survival and development of subsequent opportunistic infections in bone marrow transplant patients with diffuse alveolar hemorrhage. Am Rev Respir Dis 1991; 143:A474. 12. ChaoNJ, DuncanSR, LongGD, HorningSJ, Blume KG. Corticosteroid therapy for diffuse alveolar hemorrhage in autologous bone marrow recipients. Ann Intern Med 1991; 114:145-6. 13. Tate RM, Repine JE . Neutrophils and the adult respiratory distress syndrome. Am Rev Respir Dis 1983; 128:552-9. 14. Heflin ACJ , Brigham KL. Prevention by granulocyte depletion of increased vascular permeability of sheep lung following endotoxemia. J Clin Invest 1981; 68:1253-60 .
