INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A number of chemotherapeutic agents have been associated with the development of a drug-induced interstitial pneumonitis. Interstitial pneumonitis has been commonly reported in the setting of both autologous and allogeneic bone marrow transplantation. The clinical symptoms can include a dry cough, dyspnea on exertion or at rest, and fever. Idiopathic pneumonia syndrome (IPS), a diffuse and severe lung injury without evidence of infection, is a serious complication of bone marrow transplant (1). Mortality can reach 65% (2). Steroids are often used to treat the interstitial pneumonitis, though the efficacy of such treatment is not proven. Following autologous bone marrow transplantation (ABMT), agents that have been associated with the development of IPS include carmustine (BCNU) (1, 3) and cyclophosphamide (CPA) (1, 3, 4, 6, 7). Following allogeneic bone marrow transplants, the development of IPS may be related to several factors, including the presence of severe graft-versus-host disease (GVHD), use of methotrexate to prevent GVHD, high doses of irradiation, older age, poorer performance ratings before transplantation, and infection with cytomegalovirus (CMV) (8, 9). In addition to IPS, an acute alveolar pulmonary hemorrhage has been described in patients receiving autologous and allogeneic marrow (10). In ABMT, this usually occurs earlier than IPS, typically within the first 2 wk following transplant (10). Idiopathic pneumonia syndrome, by contrast, is more typically delayed by several weeks to months following ABMT (1).

Carmustine-based dose-intensive chemotherapy regimens with ABMT for cellular support have been shown to improve overall survival and progression-free survival in patients with metastatic breast cancer as well as those with extensive axillary node involvement (13). The occurrence of pulmonary drug toxicity with this regimen is variable and has been reported to range from 39-53% (5). We sought to further characterize the incidence and severity of pulmonary toxicity in consecutive breast cancer patients treated with high-dose CPA/cDDP/BCNU and ABMT. In particular, we wanted to describe the incidence of symptoms, associated abnormalities in pulmonary function testing, and association with radiologic abnormalities noted on CT scans. We also sought to determine predictors for the development of pulmonary toxicity as well as the response to corticosteroid therapy.

Overall, we found a high incidence of pulmonary toxicity, though there was no related mortality due to lung injury. These results have led us to coin the term delayed pulmonary toxicity syndrome (DPTS) to better describe the post-transplant noninfectious complication seen in our patient population. Delayed pulmonary toxicity syndrome can involve the development of cough, dyspnea, and fever, and should include a significant decline in single-breath diffusing capacity (DL_CO). Delayed pulmonary toxicity syndrome, which appears to respond to steroid treatment, is to be distinguished from IPS, which carries significant mortality risk.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

We performed a retrospective chart review of 45 consecutive patients who underwent high-dose chemotherapy and autologous bone marrow transplant (HDC/ABMT) for primary breast cancer with involvement of four to nine lymph nodes. All patients were treated on research protocol approved by the Duke University Institutional Review Board (DORIS #92094) and gave informed consent. Eligibility criteria for this treatment protocol included the following: operable, histologically confirmed stage II/IIA breast cancer with four to nine involved lymph nodes at resection, less than 8 wk from last surgery, Karnofsky performance status 80-100%, no prior chemotherapy or radiation therapy, no other malignancy or comorbid disease, pulmonary function test (PFT) demonstrating DL_CO, FEV₁, FVC, and total lung capacity (TLC), to all be > 60% predicted, and negative bone marrow biopsies.

Chemotherapy/Bone Marrow Transplant Protocol

Patients received induction chemotherapy with adriamycin (80 mg/ m²) and 5-fluorouracil (800 mg/m²) given every 2 wk for three cycles. Patients then received recombinant granulocyte colony-stimulating factor (G-CSF) (filgrastim) for 5 d to prime peripheral progenitor cells (PBPC) for leukopheresis. Induction chemotherapy was followed by high-dose chemotherapy with cyclophosphamide (CPA) (1,875 mg/m² intravenously as a daily 1-h infusion for 3 d), cisplatin (cDDP) (165 mg/m² continuous intravenous infusion over 72 h) given on Days 6, 5, 4, and BCNU (600 mg/m² intravenously as a one-time infusion over 2 h) given on Day 3. Blood samples for measurement of BCNU elimination were taken at eight time points: one during and seven following infusion up to 210 min from the start of the infusion. High-dose chemotherapy was followed by ABMT at Day +1, and PBPC reinfusions on Days 1, 0, and +1. The area under the drug elimination curve (AUC) for each of the chemotherapeutic agents was determined from serial blood levels utilizing high performance liquid chromatography (HPLC) according to previously published methods (7, 17). The treatment protocol calls for standard local regional radiotherapy to be delivered to the chest wall and ipsilateral supraclavicular and internal mammary nodes at a minimum of 50.4 Gy, followed by a 10 Gy boost to the mastectomy scar. This is begun generally 6 wk following ABMT, but may be delayed by cytopenias or pulmonary toxicity.

Patient Follow-up

Patients had regular follow-up at the Duke University Bone Marrow Transplant Clinic. They were followed closely for the development of pulmonary toxicity by regularly scheduled pulmonary function testing done every 6 wk for the first 24 wk and every 12 wk thereafter. Additional PFTs were done when patients developed pulmonary complaints. If signs and/or symptoms of toxicity developed, the patients received prednisone at an initial dose of 60 mg/d for 2 wk followed by a gradual taper over the following 6 wk.

The clinicians in our Bone Marrow Transplant Clinic used the following guidelines for diagnosing DPTS and initiating prednisone therapy: (1) development of typical symptoms (nonproductive cough, dyspnea, with or without fever occurring several weeks to months following ABMT) and fall in DL_CO to less than 60% predicted; or (2) decline in DL_CO to less than 50% predicted with or without symptoms; (3) for asymptomatic patients with DL_CO 50-60% predicted, treatment was left to the discretion of the treating physician. Patients treated with prednisone also received concurrent pneumocystis prophylaxis with trimethoprim/sulfamethoxasole. Clinical features of these patients were recorded, including age, smoking history, history or prior lung disease, CMV immune status, pulmonary symptoms, and data relating to prednisone treatment and radiation therapy (XRT).

Pulmonary Function Tests

Pulmonary function testing was done before induction chemotherapy (baseline) and at regular intervals after HDC/AMBT, as noted above. Pulmonary function tests included spirometry, lung volumes, and diffusion capacities (DL_CO), and were performed according to American Thoracic Society standards (18, 19). The DL_CO was corrected for hemoglobin using the formula: (1.7 × hemoglobin)/(9.38 + hemoglobin) (19). To assess changes in pulmonary function, values for DL_CO, FEV₁, FVC, and TCL were normalized by expressing them as a percent of their baseline values (% BL). Baseline spirometry and DL_CO were available for all patients, but TLC was available at baseline in only 34 patients. We considered a decline from baseline value of 15% for FEV₁, FVC, and TLC and 20% for DL_CO to be significant (19, 20). Complete recovery of PFTs after development of pulmonary toxicity was defined as improvement to 85% of baseline for FEV₁, FVC, or TLC and to 80% of baseline DL_CO.

Radiologic Analysis

All patients had CT scans performed before HDC/ABMT and at regular intervals following HDC/ABMT as part of staging and surveillance for metastatic disease. We used these scans to evaluate radiologic manifestations of pulmonary toxicity. The CT scans were reviewed retrospectively by a chest radiologist (J.J.E.) who was blinded to the results of the PFTs and to the clinical status of the patients. The CT scans were scored using the method described by Owens and colleagues (21). Briefly, the images were assessed for presence and patterns of disease (ground glass opacification, consolidation, linear-nodular opacities, effusion) at five predefined levels; (1) origin of the great vessels; (2) mid-arch of the aorta; (3) main carina; (4) pulmonary venous confluence; and (5) 1 cm above the dome of the diaphragm. Extent of disease was derived by estimating the percentage of abnormal lung to the nearest 0.5% at each level. This was adjusted for the lung volume at each level as previously described (21). Adding the five adjusted figures gave an estimation of the overall percentage volume of abnormal lung, which is reported as the CT lung injury score (CT-LIS).

Statistical Analysis

Comparisons between the groups of patients with and without pulmonary toxicity were performed using Student's t test or, when the two groups failed tests for equal variance, by the Mann-Whitney rank sum test. Comparison of proportions was performed using chi-square analysis. We used multiple linear regression to analyze the effects of radiation therapy and BCNU AUC on pulmonary function. A p value of < 0.05 was considered to be statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

The characteristics of the 45 patients are summarized in Table 1. The mean age was 43.2 ± 7.3 yr, and the majority of patients were white. Only three patients (6.7%) were actively smoking at the time of registration for the BMT program, and they had an average smoking history of 23 pack-years. All three stopped smoking prior to receiving HDC. Twenty-seven patients (60%) were never smokers. Fifteen (33%) were former smokers (mean 8.32 pack-years ± SD 6.54, range 1 to 20 pack-years).

Pulmonary Function Tests

Results of serial pulmonary function tests following HDC/ AMBT are summarized in Figure 1. Unless otherwise noted, values shown are mean ± SEM. On average, the DL_CO reaches a nadir of 58.2 ± 3.4% of baseline DL_CO at 15-18 wk following HDC/AMBT. The FEV₁, FVC, and TLC are reduced to a lesser extent. Forty-one patients (91%) met criteria for significant decline in DL_CO, and 11 (27%) of these completely recovered; 41 patients (91%) met criteria for significant decline in FEV₁, and 21 (51%) of these recovered; 39 (87%) met criteria for FVC and 16 (41%) of these recovered; 23 of 34 evaluable patients met criteria for decline in TLC, and 16 (70%) of these recovered. Only five patients (11%) had full recovery of all PFT parameters, whereas nine patients (20%) had recovery of none of the PFT parameters.

View larger version (22K): [in this window] [in a new window]   Figure 1.   Pulmonary function testing for 45 consecutive patients receiving HDC/ABMT. Median length of follow-up is 72 wk. Changes in PFT values are normalized by expressing them as percent of the baseline value. The baseline value is that measurement performed prior to receiving HDC/ABMT. The number of individual data points at each time point is shown below the corresponding diamond.

We compared pulmonary function testing in patients who developed symptoms of pulmonary toxicity to those without symptoms (Figure 2). Patients who developed symptoms had a lower nadir for DL_CO than those without symptoms (47% BL versus 64% BL; p = 0.0089) and took less time to reach that nadir (10.7 wk versus 16.3 wk; p = 0.0265). There were no significant differences for nadir values for FEV₁ (67% versus 76% BL, p = 0.0692), FVC (71% versus 78% BL, p = 0.0805), and TLC (83% versus 81% BL, p = 0.59), although this occurred significantly earlier in patients with symptoms (p = 0.0085 and 0.0038, respectively).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Comparison of pulmonary function testing in patients who develop symptoms of pulmonary toxicity to those without symptoms. Data for individual patients with and without pulmonary symptoms are shown as X and O, respectively. On average, patients who developed symptoms (dotted line) had a lower nadir for DLCO than those without symptoms (solid line) (47% BL versus 64% BL; p = 0.0089) and took less time to reach that nadir (10.7 wk versus 16.3 wk; p = 0.0265). There were no significant differences for nadir values for FEV1, FVC, and TLC (p = 0.0692, 0.0805, and 0.5902, respectively), though the number of days to nadir FEV1 and TLC were significantly less in patients with symptoms.

To evaluate the effect of AUC for BCNU on pulmonary function, we performed linear regression of the nadir DL_CO versus BCNU AUC (Figure 3). We could detect no significant correlation between systemic exposure to BCNU and the severity of the toxicity as measured by DL_CO. Multiple linear regression analysis for the AUC of each of the chemotherapy drugs revealed no correlation with decline in DL_CO. The power of these analyses was low due to small sample size.

View larger version (24K): [in this window] [in a new window]   Figure 3.   Relationship between systemic exposure to BCNU and pulmonary toxicity. We found no correlation between systemic exposure to BCNU and the nadir DLCO. Dotted lines represent 95% confidence intervals for linear regression. The mean and median AUC for BCNU were 1.012 mg min/ml and 0.744 mg min/ml, respectively. Two patients (neither of whom had pulmonary symptoms) had particularly high AUC for BCNU. The inset graph demonstrates that the inclusion of these two patients does not alter our conclusions.

Clinical Features

Clinical findings. Pulmonary symptoms included dyspnea with exertion and nonproductive cough, which often were associated with fever. Pulmonary symptoms developed in 26 patients (58%) with a mean onset of 10.3 wk (range 2.4 to 16.6 wk). We compared patients who developed pulmonary symptoms to those who remained free of pulmonary complaints (Table 2). There was no statistically significant difference between these groups for age, baseline pulmonary function, smoking history, or area under the curve for the chemotherapy agents. We also evaluated patients based on whether the toxicity occurred before or after XRT, and found no significant differences in drug exposures individually or in combination, though the power to detect such differences was low. There was no significant difference in the incidence of pulmonary toxicity between the never smokers and current/former smokers (p = 0.76), though the number of smokers was small, and therefore the power to detect a difference was only 0.06. 

Steroids. All of the 26 patients who developed pulmonary symptoms were treated with steroids. Median and mean duration of treatment were 58 and 75 d, respectively. Of these, three patients required a second course for recurrent pulmonary symptoms. Additionally, three patients who were asymptomatic were treated with prednisone because of the severity of their PFT abnormalities. Fifteen (33%) patients who had significant but less severe decline in PFTs never developed clinical symptoms and did not receive corticosteroids. One patient with severe decline in DL_CO (nadir 31% BL at Day 99 post-ABMT) but who remained asymptomatic did not receive prednisone. Her actual DL_CO at baseline was 133% predicted and fell to 41% predicted. Notably, there was marked long-term depression of her DL_CO, while the other patients with similar decrements in DL_CO who received corticosteroids had significant improvement (data not shown).

For patients receiving prednisone for pulmonary toxicity, we compared the DL_CO prior to treatment with the highest DL_CO occurring during or immediately following treatment. Patients who did not have a pretreatment DL_CO within 1 wk prior to starting prednisone were excluded from analysis. Results are shown in Figure 4. There was a statistically significant improvement in DL_CO (average improvement 17.1%; p = 0.0001; 95% CI 11.3-22.9). We also evaluated nine patients who received prednisone for thrombocytopenia immediately following HDC/ABMT. Interestingly, only two of these nine (22%) subsequently developed pulmonary symptoms. However, the number of patients in this group was too small to statistically detect whether the steroids exerted a protective effect against the development of pulmonary drug toxicity (p = 0.0608).

View larger version (30K): [in this window] [in a new window]   Figure 4.   Effect of prednisone on improving pulmonary function. For each episode in which patients received prednisone for a pulmonary indication (n = 26), we compared pre-prednisone DLCO with the highest DLCO occurring during or immediately following treatment. Patients who did not have a pre-treatment DLCO within 1 wk prior to starting prednisone were excluded from analysis (n = 8). Median and mean duration of treatment were 58 and 75 d, respectively. There was a statistically significant improvement in DLCO (average improvement 17.1%, p = 0.0001; 95% CI 11.3-22.9).

Infectious complications. Two patients were diagnosed with pneumonia: one at 6 wk and one at 69 wk following HDC. Neither patient was considered to have pulmonary drug toxicity. The first developed fever and right anterior chest pain, and a chest radiograph obtained by her local physician revealed right lung infiltrate. Blood cultures were negative. She improved following treatment with clarithromycin. The second patient had fever, chills, cough productive of purulent sputum, wheezing, and anterior chest pain with leukocytosis (WBC 18,000) and lobar infiltrate. Her symptoms improved and radiograph cleared on oral amoxicillin/clavulinic acid alone. No specific etiology was identified, but both improved with antibiotic therapy, and neither received steroids.

Of those patients treated with steroids for pulmonary toxicity, none developed infectious pneumonia during or immediately following steroid treatment. One patient developed pneumonia at Day 699, which was 19 mo after completion of steroid taper.

Deaths. There were two deaths; one patient who died 46 d post-ABMT of a suspected acute air embolus; the other died from metastatic breast carcinoma. There were no deaths resulting from pulmonary toxicity.

Radiologic Analysis

Computed tomography (CT) of the chest was obtained at regular intervals in all patients. In interpreting the CT-lung injury score, we examined the total score as well as the types and distribution of abnormalities noted. We used the following definitions: minimal injury = CT-LIS < 5%, mild injury = LIS 5- 10%, moderate = LIS 10-25%, severe = LIS > 25%. Minor lineage markings in dependent posterior lung fields are frequently seen in normal patients and represent subsegmental atelectasis (22, 23). If the only finding was subsegmental atelectasis and the LIS was < 5%, then the CT scan was considered normal. Baseline CT scans were available for 39 patients and were normal in all cases. Radiographic criteria for radiation-induced injury were the presence of ground glass, linear-nodular, consolidative, or mixed opacities restricted to the radiation port. In general, they were felt to be radiation-induced if anterior in location and unilateral, or if bilateral then involving the anterior-medial aspect of the contralateral lung (i.e., in the radiation port). Radiographic criteria for pulmonary drug toxicity (PDT) included the presence of bilateral ground glass, linear-nodular, consolidative, or mixed opacities in the absence of evidence for infection. If these changes occurred during or following XRT, they were found outside as well as inside the radiation port. If these were only minor abnormalities (CT-LIS < 5.0%), then we considered the CT to be only slightly suggestive of PDT.

CT findings at the time of pulmonary toxicity. Of the 28 patients treated with steroids for pulmonary toxicity, 18 had CT scans performed at the time of toxicity (16 patients 24 h prior to steroids, 1 patient 3 d prior, and 1 patient 6 d prior). Only five developed CT abnormalities strongly suggestive of PDT (mean CT-LIS, 25.8%, range 9.1-61.8%); two had minimal CT changes suggestive of PDT (CT-LIS < 5%); two had CT abnormalities only in the radiation port. The remaining 10 patients had normal CT scans at the time of pulmonary toxicity. Four of these never developed radiographic abnormalities. The other six patients developed subsequent XRT changes that were minimal for all but two patients with CT-LIS of 18.6% and 5.4%.

Types of CT abnormalities. The development of radiographic abnormalities did not necessarily correlate with symptoms or decline in pulmonary function. Abnormalities on CT scan developed simultaneously with symptoms or decline in DL_CO in only 13 (41%) of 32 patients who had CT at the time of symptoms/nadir DL_CO. Twelve patients (37%) later developed abnormalities restricted to the radiation port. The CT scans for seven patients (22%) remained normal despite the development of significant abnormalities in PFTs. Types of abnormalities present at symptoms/DL_CO nadir included ground glass opacities (n = 18), mixed ground glass and linear-nodular opacities (n = 8), line-nodular opacities (n = 1), consolidation (n = 2), and mixed ground glass opacities with consolidation (n = 3). Only one patient had an effusion that was moderate to large. In follow-up scans, ground glass opacities typically either resolved or were replaced by linear-nodular opacities. There was no difference between the median CT-LIS for those with symptoms versus without symptoms either for the maximum score for each patient (4.82% versus 2.76%; p = 0.146) or for score at time of DL_CO nadir (2.80% versus 0.23%; p = 0.096). Using a CT score of 5%, the sensitivity and specificity for the development of symptoms were 33% and 60%, respectively, though all abnormalities that developed in patients who remained asymptomatic were consistent with postradiation changes. The sensitivity of the CT for a decrease in DL_CO to < 85% BL was 30%.

Radiation-induced CT changes. Radiographic abnormalities limited to the radiation therapy port were a frequent finding. Of the 39 patients who did receive XTR, 13 developed pulmonary toxicity prior to receiving XRT (12 with both symptoms and PFT changes, one by DL_CO alone). Nine of these had CT scans performed at the time of toxicity, only four of which had any abnormalities on CT; these were consistent with inflammatory drug reaction and resolved in all cases. Following XRT, nine of these 13 patients developed abnormalities in the radiation port that were minimal (CT score < 2.0%) in all but two cases (CT scores, 4.9% and 18.6%). Ten of the patients developed pulmonary toxicity following XRT or late in the course of XRT, and eight of these had CT scans at the time of symptoms. Three of these had CT changes strongly suggestive of PDT (CT scores, 10.5%, 2.2%, 61.8%); two had abnormalities only in the radiation port (CT scores, 3.4%, 22.3%); three had normal CT findings and later developed minimal changes in the radiation port; the two patients who did not have CT scans at the time of symptoms also had mild radiation changes in subsequent CT scans. In the six patients who did not receive XRT, only one patient developed any CT abnormalities. These developed at the time of her symptoms and were consistent with PDT. For the 17 patients who did not have pulmonary toxicity (as defined by treatment for symptoms or low DL_CO), 16 patients received XRT (the one who did not died of suspected air embolus prior to planned XRT) and 14 had CT scans done at the time of nadir DL_CO. Only one of these developed any CT abnormalities suggestive of PDT (CT-LIS 4.35%) that were present at the time of DL_CO nadir. Ten patients had normal CT at the time of DL_CO nadir and eight of these developed subsequent radiation changes. Of the 17 patients without pulmonary toxicity, 14 developed CT abnormalities suggestive or radiation change (CT-LIS from 1.7-25.6%), and three never developed abnormalities (though one of these died after only short follow-up).

Effect of radiation on pulmonary function. To determine whether the timing of radiation therapy relative to HDC was important to changes in lung function, we compared pre-XRT DL_CO with subsequent changes in DL_CO. Thirty-eight patients (84%) received radiation therapy. Nine patients who did not have PFTs close to the time of starting XRT were excluded from this analysis. On average, there was a 13.8% reduction in DL_CO from the start to stop of XRT treatment.

We plotted time of initiation of XRT on the ordinate and the change in DL_CO over the course of XRT, and then performed linear regression (Figure 5). The slope of this regression line was not statistically different from zero. The results were unaffected by whether patients receiving prednisone during XRT were included or excluded from analysis. However, the power of the test was only 0.31. Multiple linear regression of the XRT start day and the AUC for each of the chemotherapy agents showed no correlation with decline in DL_CO, though again power was limited. We also divided those patients with toxicity into two groups: those who had toxicity before XRT and those who developed their toxicity during or after XRT to see if XRT acted synergistically with the agents to cause toxicity. There was no difference in the chemotherapy AUC for these groups.

View larger version (21K): [in this window] [in a new window]   Figure 5.   Effect of timing of XRT on changes in DLCO. Linear regression analysis of the number of days post-ABMT for the start of XRT versus subsequent changes in DLCO showed the slope not to be different from zero. Dotted lines represent 95% confidence intervals for linear regression. Note that on average there was a 13.8% reduction in DLCO from the start to stop of XRT treatment.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary toxicity is a recognized complication of cancer chemotherapy, especially of dose-intensive regimens, and has been the subject of several reviews (1, 24). The development of a severe interstitial pneumonitis as a complication of bone marrow transplantation has recently been termed idiopathic pneumonia syndrome (IPS). It has been observed that the diffuse lung injury that follows autologous bone marrow transplants seems to have a less severe clinical course than that which follows allogeneic transplants (1). We propose the introduction of the term delayed pulmonary toxicity syndrome (DPTS) to better describe the pulmonary toxicity that can occur in patients who undergo HDC/ABMT. We feel this new terminology is warranted because of its higher incidence, lower mortality, and apparent steroid responsiveness that distinguish it from IPS. Additionally, most studies referring to IPS are in patients following allogeneic marrow transplant, and factors contributing to pulmonary dysfunction in these patients, such as GVHD, high doses/total body irradiation, and use of methotrexate, do not apply to our patients.

The incidence of DPTS in our study is based primarily on clinical symptoms and pulmonary function abnormalities, particularly DL_CO. The pathology associated with pulmonary toxicity in these patients has been well described and can include type II pneumocyte dysplasia, interstitial and septal thickening with fibrosis, fibroblast proliferation, intraalveolar edema and fibrin deposition, and increased numbers of alveolar macrophages (6, 25, 26, 28). Pathologic specimens were not routinely obtained from the patients in our study. In the past, the diagnosis of PDT generally has required bronchoscopic or open lung biopsy to confirm the diagnosis and exclude other etiologies. Because of the high frequency of noninfectious pulmonary toxicity in our patient population (6), the clinicians at our institution generally accept noninvasive means to exclude infection in patients with a typical clinical presentation.

Possible risk factors for the development of pulmonary toxicity in patients receiving BCNU-based chemotherapy have been identified and include (1) pre-existing lung disease; (2) a history of smoking; (3) industrial exposures; and (4) thoracic irradiation (6, 26, 29, 30). In our study, the development or severity of DPTS did not correlate with differences in baseline pulmonary function, though all patients had baseline DL_CO 65% predicted. We were not able to correlate the risk of developing DPTS with tobacco use, although the power to do so was low, given that few patients were active or former smokers at the time of entry into treatment and all active smokers stopped smoking at the time of registration. Likewise, the small number of patients and the retrospective nature of the study make if difficult to assess the contribution of XRT to decline in lung function. Because patients who do not receive thoracic irradiation are at risk for local recurrence (6), prospective randomized trials examining the role XRT plays in exacerbating lung toxicity are not likely. We suspect that XRT may contribute to lung injury in some patients whose radiographic changes are localized to the radiation portal. Our data (Figure 4), as well as that of others, demonstrate lung injury following XRT (6).

Two of the chemotherapy agents in our HDC regimen, carmustine and cyclophosphamide, have been associated with pulmonary injury. There is considerable evidence that the pulmonary toxicity of carmustine is related to levels of systemic drug exposure. Several studies note the incidence of pulmonary toxicity to increase with the total dose of BCNU (29, 31, 32). Aronin and colleagues reported that symptomatic patients received a higher dose of BCNU than asymptomatic patients (1,146 mg/m² and 777 mg/m², respectively), and while the overall incidence of pulmonary toxicity was 20%, this increased to 50% when the cumulative dose exceeded 1,500 mg/ m² (29). Phillips and coworkers (31) noted a 9.5% incidence of fatal pulmonary toxicity when BCNU used as a single agent was administered in doses exceeding 1,200 mg/m². Overall, 16 of 83 (19.3%) patients in their study developed BCNU interstitial pneumonitis, and no pulmonary toxicity occurred at doses less than 1,000 mg/m² (32). Durant and associates reported no correlation with total dose of BCNU, although this was a retrospective record review of 794 patients who had received BCNU and other agents from which only nine cases of pulmonary toxicity were identified, and doses were only reported for those patients with toxicity (30). All patients in our study received a BCNU dose of 600 mg/m², a dose which is below that associated with pulmonary toxicity when BCNU is administered as a single agent. These data alone suggest that factors in addition to BCNU (e.g., cyclophosphamide, cisplatin, or other components of the induction regimen) may predispose the lung to DPTS.

Blood levels of pharmacologic agents may vary considerably in multidrug regiments (33). Jones and colleagues (33) used pharmacokinetic modeling to determine the area under the drug elimination curve (AUC) for BCNU. They found that an AUC for BCNU > 600 µg min/ml correlated significantly with the incidence of pulmonary toxicity in a cohort of 44 patients treated with CPA/cDDP/BCNU (7). Using these same methods, however, we found no such correlation. Thus, despite minor differences in drug administration, our pharmacokinetic data further support our hypothesis that factors in addition to or synergistic with BCNU may contribute to the development of DPTS.

Computed tomography assessments have been correlated to clinical and pulmonary function abnormalities in sarcoidosis and with lung injury score in ARDS (21, 34, 35). In the current study, the observation that CT abnormalities may be absent or may frequently occur later than symptoms and/or the development of pulmonary function abnormalities prevented a similar correlation. Several other investigators have noted that radiographic abnormalities may often be delayed relative to onset of symptoms or may not develop at all (27, 29, 30, 32). Additionally, several patients in our study developed significant CT abnormalities in the absence of concomitant or subsequent pulmonary symptoms. These opacities were attributable to XRT and resolved or improved without treatment. We conclude from these observations that routine standard chest CT is unlikely to be a useful tool for predicting the development of DPTS.

Controversy exists in the literature regarding the steroid responsiveness of pulmonary toxicity associated with BCNU-based regimens. Early reports particularly note variable or poor response to steroids (27, 29, 30). More recently, and with a greater awareness of pulmonary toxicity, Jones and coworkers reported symptomatic improvement within 1 wk in 12 of 13 patients treated with prednisone for pulmonary toxicity following HDC/ABMT for breast cancer (7). Similarly, Todd and associates noted that most patients in their series of 10 patients with pulmonary toxicity from HCT/ABMT had rapid symptomatic improvement with steroids (6). A recent National Heart, Lung, and Blood Institute (NHLBI) Workshop Summary noted that patients with interstitial pneumonitis following ABMT usually respond to high-dose steroids (1). However, to our knowledge, no well-controlled prospective trials exist for the treatment of pulmonary toxicity related to either autologous or allogeneic transplantation. The patients in our study who received steroid treatment of DPTS had a 17% improvement in DL_CO. We attribute our low level of mortality to early recognition of DPTS and rapid institution of treatment with corticosteroids. Furthermore, patients receiving prednisone early following HDC, usually for thrombocytopenia, appeared less likely to develop pulmonary symptoms, though the numbers were small and did not quite reach statistical significance. Controlled prospective trials are needed before the routine use of steroids for prophylaxis of DPTS can be recommended.

Our study of patients receiving high-dose chemotherapy and autologous bone marrow support for breast cancer revealed a high incidence of DPTS but no mortality due to DPTS. Conventional lung CT scans correlated poorly with development of pulmonary symptoms or decline in DL_CO and lacked both sensitivity and specificity. We found that patients who developed pulmonary symptoms had a more rapid and more severe decline in DL_CO. It is possible that the slope of decline in pulmonary function might be used to predict which patients may be at particular risk for developing DTPS. However, this would need to be tested prospectively using more frequent pulmonary function testing in the immediate post-HDC/ABMT period. Neither age, baseline pulmonary function, smoking history, or AUC for the chemotherapy agents were predictive of which patients would develop DPTS. In conclusion, our data suggest that early diagnosis of DPTS with concomitant institution of corticosteroids will result in significant improvement in pulmonary function and prompt resolution of pulmonary symptoms.