Pulmonary Toxicity of High-Dose Chemotherapy for Breast Cancer: a Non-Invasive Approach to Diagnosis and Treatment

Pulmonary toxicity of high-dose chemotherapy for breast cancer: a non-invasive approach to diagnosis and treatment

L Chap1, R Shpiner2, M Levine2, L Norton1, M Lill1 and J Glaspy1

Divisions of 1Hematology/Oncology, and 2Pulmonary and Critical Care Medicine, UCLA Medical Center, Los Angeles, CA, USA

Summary: ive dyspnea on exertion (DOE), a nonproductive cough ± fever, restrictive changes in pulmonary function and Drug-induced pulmonary toxicity is one of the most fre- occasional radiographic infiltrates.2 The optimal approach quent non-hematologic toxicities in breast cancer to diagnosis and treatment, however, has not been eluci- patients receiving high-dose chemotherapy with cyclo- dated. phosphamide, cisplatin and BCNU (CY/CDDP/BCNU). Studies of new approaches for the prevention and early A non-invasive clinical scoring system was utilized in an diagnosis of pulmonary toxicity will depend upon the attempt to diagnose and treat early lung toxicity in 64 development of valid, reproducible and objective diagnostic consecutive breast cancer patients undergoing criteria. This study was intended to develop a noninvasive CY/CDDP/BCNU supported by peripheral blood pro- model to aid in the diagnosis and treatment of chemo- genitor cells. Following hospital discharge, patients who therapy-associated pulmonary toxicity by which points developed symptoms suggestive of lung toxicity were were assigned for abnormal results of pulmonary function evaluated with physical examination, DLCO, 2-min tests and physical examination. walking oximetry and a chest radiograph. Clinically weighted scores were assigned as follows: crackles on lung exam, 2; decrease in corrected DLCO by Ͼ10% Patients and methods from baseline, 3; decrease in O2 saturation by у4% with a 2-min walk, 3; and interstitial infiltrates on chest Patients radiograph, 3. Patients with scores у6 were treated Sixty-four consecutive breast cancer patients undergoing with prednisone (60 mg p.o. twice a day followed by a high-dose chemotherapy with CY/CDDP/BCNU supported 2-month taper). Treatment was instituted in 37 patients by peripheral blood progenitor cell (PBPC) transplantation (58%) a median of 56 days after high-dose chemo- between July 1993 and September 1994 were prospectively therapy. Steroid therapy was associated with rapid followed. Patients were eligible if they had high-risk breast clinical improvement in most patients. No fatal compli- cancer, defined as у10 involved lymph nodes or primary cations or chronic pulmonary fibrosis was seen. This tumors Ͼ5 cm with lymph node involvement, or metastatic non-invasive clinical scoring system can be utilized as a disease with a demonstrated response to standard-dose model for the early diagnosis of lung toxicity. Further chemotherapy. The high-risk breast cancer patients were investigation is warranted for the development of pre- treated with three or four courses of standard-dose doxorub- ventative measures against this syndrome. icin-based chemotherapy for induction, and patients with Keywords: high-dose chemotherapy; peripheral blood metastatic disease were treated with a variety of standard- progenitor cell; pulmonary toxicity; breast cancer dose chemotherapy agents until maximal response. Entry criteria included a minimum post-bronchodilator FEV1 of Ͼ60% predicted and corrected DLCO of Ͼ60% predicted. Following induction chemotherapy, patients were enrolled High-dose chemotherapy with autologous stem cell support in a randomized study to receive granulocyte colony-stimu- for subsets of patients with breast cancer is being used with lating factor (G-CSF) ± stem cell factor for the mobilization 1–3 increasing frequency in the USA. These patients are fre- of PBPCs; Ͼ4 × 108 mononuclear cells were collected dur- quently conditioned with a combination of cyclophospham- ing 3 days of leukapheresis. The high-dose chemotherapy ide, cisplatin and BCNU (CY/CDDP/BCNU). Pulmonary consisted of cyclophosphamide, 1875 mg/m2/day on days toxicity is one of the most frequent non-hematologic tox- −5to−3 ; cisplatin, 55 mg/m2/day on days −5to−3; and icities associated with this regimen, with a reported inci- BCNU, 600 mg/m2 as a single infusion on day −2. The 4 dence of 39%. The interstitial pneumonitis which occurs previously collected PBPCs were infused on days 0 to +2. can progress to fatal pulmonary fibrosis if left untreated or All patients received daily G-CSF beginning on day 0 until if the diagnosis is delayed.2,5 The pulmonary syndrome is hematologic recovery. Informed consent documents now well recognized and typically manifests with progress- approved by the UCLA Human Subjects Protection Com- mittee were signed by all patients. Correspondence: Dr LI Chap, Division of Hematology/Oncology, 10945 All high-risk patients received local-regional radiation LeConte Ave, 2333 PVUB, Los Angeles, California 90095, USA therapy (XRT) to the chest wall and regional lymph nodes Received 11 December 1996; accepted 21 August 1997 (45–50 Gy) following recovery from transplantation. Diagnosis of drug-induced lung toxicity L Chap et al 1064 Pulmonary evaluation Results

Pretreatment evaluation included a history and physical Clinical features examination, chest radiograph (PA and lateral), spirometry with DLCO (single breath diffusion capacity with standard Sixty-four consecutive female patients were studied. The Coles hemoglobin correction) and a 2-min flat walk oxime- patient characteristics are shown in Table 1. The median try. Following hospital discharge, patients were routinely age of the patients was 45 years and the majority (61%) followed-up at 1 and 2 weeks, monthly for 3 months and were treated for high-risk breast disease. Six of the patients every 2 months thereafter. Evaluation consisted of a symp- had pulmonary metastases; none of these demonstrated a tom history, physical examination and laboratory studies, lymphangitic pattern of disease. The majority of the including a complete blood count and chemistry panel. patients received radiation to the chest wall and regional Patients who reported symptoms suggestive of possible pul- lymph nodes following transplantation. Only eight of the monary toxicity (eg DOE, cough, fever or chest tightness) patients had a smoking history, and one patient had a his- underwent additional testing with DLCO, 2-min walking tory of sarcoidosis which was quiescent at the time of trans- oximetry and a chest radiograph. In nearly all instances, plantation. the DLCOs and oximetry were done on the same machine Treatment was instituted in 37 patients (58%) who and evaluated by the same technician. Patients who lived developed clinical evidence of pulmonary toxicity. Thirty too far away to travel to the transplant center for a full of these patients (81%) had a complete pulmonary evalu- pulmonary evaluation were encouraged to see their primary ation, as described above, and were assigned a clinically physicians and maintain contact by phone. All patients were weighted score. Twenty-six of the 30 patients (87%) had a clinical diagnostic score of у6, one had a score of 5 and followed for their response to treatment and outcome. A three had a score of 3 with symptoms (DOE) and a marked therapeutic response to treatment was defined as the resol- decrease from their baseline DLCOs (−30, −31 and −50%). ution of pulmonary symptoms and normalization of pul- Although the latter three patients did not meet the diagnos- monary function tests. The median follow-up was 28 tic criteria, they were treated on the basis of significant months (range, 22–34 months) post-transplantation. DOE and decrease in DLCO. The remaining seven patients were treated empirically, five based on symptoms, alone and two based on symptoms and a decreased DLCO. The clinical features of the patients who were treated are Diagnosis and treatment summarized in Table 2. The median onset of pulmonary toxicity was 56 days (range, 21–240 days) from day 0 of A clinically weighted point scoring system was developed transplantation. The most common symptom reported by to serve as a model for the diagnosis of pulmonary toxicity: affected patients was the subacute development of dyspnea crackles on lung exam, 2; decrease in corrected DLCO by on exertion (30 patients (81%)), usually evolving over sev- Ͼ10% from baseline, 3; decrease in O2 saturation by у4% eral days. Eight patients (22%) reported having a dry with a 2-min walk, 3; and interstitial infiltrates on chest cough, six as an isolated symptom, and three patients radiograph, 3. The individual scores assigned to these para- reported a fever Ͼ38.0°C, one as an isolated symptom and meters were based on the authors’ previous experience in two in conjunction with DOE. identifying this syndrome. Patients with a score у6 were Physical examination was abnormal in only three (8%) treated with prednisone; scores Ͻ6 were followed up of the patients. Two patients had bibasilar inspiratory weekly until stable over 6 weeks or treatment criteria were crackles and one had diffuse, end-expiratory wheezing. met. Prednisone therapy was instituted at a dose of 60 mg p.o. twice a day for 10 days, followed by 30 mg q day × 1 week, 20 mg q day × 1 week, 15 mg q day × 1 week and then tapered by 5 mg once a day, each week thereafter. Table 1 Patient characteristics (n = 64) Patients who were unable to undergo a full pulmonary evaluation, due to distance from the transplant center, were Characteristic No. (%) treated empirically on the basis of symptoms and when available, by DLCO. Age Ͻ50 years 44 (69) у50 years 20 (31) Median, 45 years (range, 32–64) Stage Statistics Locally advanced 39 (61) Metastatic 25 (39) Data on patient age, disease status, radiation treatment, his- Radiation treatment Pre-transplant 18 (28) tory of tobacco use and cytokine(s) used for PBPC mobiliz- Post-transplant 41a (64) ation were compared between the treated and non-treated None 10 (16) groups. A ␹2 test for proportions was used and a P value History of tobacco use 8 (13) of 0.05 or less was considered to indicate statistical signifi- History of lung disease 1 (2) cance. aOne patient received XRT to the spine only. Diagnosis of drug-induced lung toxicity L Chap et al 1065 Table 2 Characteristics of patients with pulmonary toxicity (n = 37) months. The remaining six patients experienced an exacerbation of their pulmonary symptoms after radiation was Onset of symptoms (days) instituted. Among the 27 unaffected patients, 17 received Median 56 post-transplant radiation (63%). Range 21–240 Symptoms (n = 37) (%) DOE 30 (81) Treatment of pulmonary toxicity Dry cough 8 (22) Fever Ͼ38.0°C 3 (8) Patients diagnosed as having pulmonary toxicity were Physical examination (n = 37) (%) treated with prednisone, as described above. Patients gener- Abnormal 3 (8) ally reported marked improvement of their symptoms DLCO decrease (n = 32) (five unavailable) (%) within 72 h of initiating treatment and demonstrated Ͼ10% 30 (94) improvement in their DLCOs over the next few weeks. Range 15–55 Eleven patients required prolonged treatment with steroids Ͻ10% 2 (6) (4–8 months), experiencing exacerbation of their symptoms Median decrease 27 when prednisone was decreased to 15–20 mg/day. Seven 2-min exercise oximetry (n = 28) (nine unavailable) (%) of these patients developed toxicity during their post- Decrease у4% 19 (68) transplant radiation treatment. Four of the 11 patients Range 5–11 underwent bronchoscopy. One was found to have Pneumo- Decrease Ͻ4% 9 (32) cystis carinii pneumonia (PCP) which responded to i.v. Median decrease 6 Chest X-ray (n = 30) (%) trimethoprim/sulfamethoxazole and the other three had Abnormal 13 (43) non-diagnostic findings. Two of the latter patients ulti- mately recovered and the third patient was found to have lymphangitic spread of carcinoma on open lung biopsy. Four of the 37 patients expired, two from progressive meta- Pulmonary function tests static, visceral disease (2 and 7 months following Thirty-two of the treated patients were evaluated with a transplantation), one from carcinomatous meningitis (at 4. corrected DLCO, of which 30 experienced Ͼ10% decrease months), and another from cytomegalovirus (CMV) pneu- from their pretransplant level, with a median decrease of monitis (at 3 months). The remaining patients were treated 27% from baseline. Twenty-eight of these patients also with prednisone for 2–3 months. At a median follow-up of underwent 2-min exercise oximetry, of which 19 had a 28 months, the majority of patients report full recovery with decrease in oxygen saturation of Ͼ4% with a median a few continuing to have mild DOE. No patient has decrease of 6%. Only two of the patients diagnosed with developed evidence of progressive pulmonary disease due pulmonary toxicity had a Ͼ4% decrease in oximetry with- to drug toxicity. out a concomitant fall in DLCO. In contrast, four of the patients had a fall in DLCO of Ͼ10% without a significant Comparisons with untreated patients decline in oximetry. Differences in disease status (locally advanced vs metastatic), radiation treatment or history of tobacco use Radiographic studies were not associated with significant differences in the proportion of patients who met criteria for the diagnosis of Of the 37 patients who developed clinical evidence of pul- pulmonary toxicity (Table 3). No differences in the rate of monary toxicity, 30 had a chest X-ray examination; 13 of these were abnormal. The most comon finding was bilateral Table 3 Characteristics of patients with and without pulmonary interstitial infiltrates, in 9/13 studies, followed by unilateral toxicity interstitial infiltrates in the remainder. High resolution CTs obtained in four patients with per- With Without sistent symptoms, demonstrated patchy ground glass alveo- toxicity toxicity litis typical of drug toxicity. Three of these patients also n = 37 (%) n = 27 (%) had chest X-rays which revealed interstitial infiltrates; the remaining patient’s chest X-ray was within normal limits. Age (years) у50 7 (19) 13 (48) Ͻ50 30 (81) 14 (52) Radiation treatment Median 42 49 Disease status Twenty-four of the 37 patients (65%) who were treated for Locally advanced 23 (62) 16 (59) pulmonary toxicity received local-regional chest wall radi- Metastatic 14 (38) 11 (41) ation following transplantation. Of these 24 patients, eight Pulmonary 3 (8) 3 (11) developed pulmonary drug toxicity during their course of Radiation treatment radiation, generally 2–4 weeks after beginning radiation. Pre-transplant 12 (32) 6 (22) Ten patients were diagnosed with pulmonary toxicity prior Post-transplant 24 (65) 17 (63) to the onset of radiation therapy, with the majority having None 4 (11) 6 (22) already completed a course of prednisone therapy; only one History of tobacco use 5 (14) 3 (11) of these patients required prednisone treatment beyond 2 Diagnosis of drug-induced lung toxicity L Chap et al 1066 pulmonary toxicity were observed in patients receiving G- drug-induced lung toxicity and chose to use these tests in CSF as compared to G-CSF + SCF for PBPC mobilization. our predictive model. A significant decline (Ͼ10%) in cor- A higher proportion of patients р50 years than those over rected DLCO from baseline, as weighted by our scoring 50 years of age met criteria for the development of lung system, appeared to be the most sensitive predictor of pul- toxicity (68% vs 35%, P = 0.01). There was no apparent monary toxicity. The 2-min exercise oximetry was also coexisting variable, such as number of previous therapies, abnormal in the majority of treated patients, but was only to account for the increased incidence of toxicity in rarely abnormal in the setting of a normal DLCO. younger patients. Less than 50% of the patients had abnormal radiographs and one patient was found to have an abnormal high- resolution computerized tomography (HRCT) with a nor- Discussion mal radiograph. HRCT has been demonstrated to be more sensitive than radiographs in the diagnosis of chemo- In this study, we used a clinical point scoring system as therapy-induced lung disease, with the most frequent a non-invasive tool to diagnose chemotherapy–associated appearance on HRCT being fibrosis, with or without con- pulmonary toxicity in breast cancer patients treated with solidation, or patchy, ground-glass opacification.7 high-dose CY/CDDP/BCNU. Ninety per cent of the Pre- or post-transplant radiation did not predict for the patients were diagnosed on the basis of these criteria and development of pulmonary toxicity. However, patients treated with prednisone. The remaining patients were diag- receiving XRT were more likely to demonstrate a protrac- nosed and treated on the basis of DOE ± a decrease in ted course and a slower response to steroid treatment; DLCO. None of the patients in the study developed pro- attempts at withdrawing steroids during radiation often gressive lung disease. The 58% incidence observed in this resulted in a flare of symptoms. We recommend that ster- series of patients is higher than that reported by other oids be tapered with caution and/or be continued until the authors4,5 and may be explained by one or more variables. XRT is completed. For patients diagnosed with pulmonary Todd et al4 required new radiographic opacities for diag- toxicity prior to the onset of radiation, consideration should nosis whereas in our study, only 43% of the treated patients be given to delaying radiation until prednisone has been had radiographic abnormalities. The patients with normal discontinued. In our series, a higher rate of pulmonary tox- radiographs may have been diagnosed at an earlier stage of icity was observed in patients under 50 years of age. This pulmonary toxicity, with the development of symptoms and has not been reported previously and merits examination in a decline in DLCO and oximetry preceding the develop- future studies of this complication. ment of frank pulmonary infiltrates. This is supported by None of the patients in this series were diagnosed on the the absence of progression to pulmonary fibrosis in any of basis of invasive testing. Open lung biopsy findings for this the patients. In contrast, Todd et al diagnosed 23 patients syndrome are non-specific but consistent with drug toxicity, with pulmonary toxicity, 10 of whom underwent open lung with an increase in type II pneumocytes, vascular injury, biopsy. Three of these patients died 2–6 months following increased alveolar macrophages and hyaline membrane for- chemotherapy from severe drug-induced lung toxicity. mation with diffuse alveolar damage.4 Many of their patients also required high doses of sup- The process by which high-dose chemotherapy results in plemental oxygen. Another confounding variable is that pulmonary toxicity is unknown. Both cyclophosphamide five of our patients were treated empirically, based only on and BCNU are known to cause lung toxicity and when used the development of DOE, without formal pulmonary test- together, may have an adverse synergistic effect. In a dose- ing. If these patients are excluded from the analysis, how- escalation study of BCNU, a 24% incidence of interstitial ever, the incidence of pulmonary toxicity only decreases to pneumonitis with a 9.5% mortality rate was observed at a 54% (32/59). Although other diagnoses might have met the dose of 1200 mg/m2.6 Other studies have reported a 20– proposed diagnostic criteria for pulmonary toxicity, it is 30% incidence of pulmonary toxicity in patients treated unlikely amongst our clinical cohort of patients. It is with BCNU for gliomas.8,9 The incidence appears to be less improbable that any of these patients had occult pulmonary at lower doses but may be enhanced when used with high- hemorrhage, given the delayed and subacute onset of symp- dose cyclophosphamide.10 Pulmonary toxicity associated toms, as well as hematologic recovery; the lack of disease with cyclophosphamide has not been as well documented progression on steroids makes CMV pneumonitis or other by clinical studies but is supported by reports in humans infectious processes unlikely; and clinically, none of the as well as animals.11 patients presented with evidence of congestive heart failure. The underlying mechanism(s) by which these drugs The clinical presentation of these patients was most cause adverse pulmonary reactions remains to be fully notable for the subacute development of dyspnea on exer- elucidated. The proposed mechanism for carmustine pul- tion with the physical examination only rarely being help- monary damage is through decreased availability of ful. The median onset of symptoms was 56 days following reduced glutathione (GSH) by the inhibition of glutathione stem cell infusion but occurred as late as 240 days. With reductase.12 GSH is an important antioxidant defense mech- regard to non-invasive pulmonary testing, primarily restric- anism of lung tisue. Cyclophosphamide also depletes GSH tive lung changes with no evidence of obstruction have stores through direct binding with its metabolite, acrolein,13 been documented by some authors,4,6 whereas others have and has been shown to increase oxidant production by also observed a significant decline in FEV1.5 Based on our macrophages in rats treated with the drug.14 Thus, the pul- own prior experience, we found DLCO and exercise oxime- monary toxicity observed in patients treated with BCNU try to be the most sensitive pulmonary function tests for and cyclophosphamide may be the result of a combined Diagnosis of drug-induced lung toxicity L Chap et al 1067 impairment in the lung’s antioxidant defense system. Radi- combination chemotherapy and autologous bone marrow ation may contribute to the development or exacerbation transplantation. Am Rev Respir Dis 1993; 147: 1264–1270. of the lung toxicity through further oxidative stress on the 5 Seiden MV, Elias A, Ayash L et al. Pulmonary toxicity asso- lung tissue. ciated with high-dose chemotherapy in the treatment of solid The proposed diagnostic criteria for chemotherapy- tumors with autologous marrow transplant: an analysis of four chemotherapy regimens. Bone Marrow Transplant 1992; 10: induced lung toxicity was developed only for this particular 57–63. preparative regimen but may also be applicable to other 6 Phillips GL, Fay JW, Herzig GP et al. Intensive 1,3-bis(2- high-dose chemotherapy settings. A 15–20% incidence of chloroethyl)-1-nitrosourea (BCNU), NSC 4366650 and cryo- drug-induced interstitial pneumonitis has been reported preserved autologous marrow transplantation for refractory with CY/BCNU/etoposide for advanced Hodgkin’s dis- cancer. Cancer 1983; 52: 1792–1802. ease,15 TBI/CY for conditioning in leukemia and lym- 7 Padley SPG, Adler B, Hansell DM et al. High-resolution com- phoma,16 and CY/thiotepa for breast cancer.17 puted tomography of drug-induced lung disease. Clin Radiol Drug-associated pulmonary toxicity is a significant 1992; 46: 232–236. source of morbidity and potential mortality in patients 8 Selker RG, Jacobs SA, Moore PB et al. 1,3-bis(2-chloroethyl)- undergoing high-dose chemotherapy. As the use of high- 1-nitrosourea (BCNU)-induced pulmonary fibrosis. Neurosur- gery 1980; 7: 560–565. dose chemotherapy with autologous stem cell support for 9 Aronin PA, Mahaley MS, Rudnick SA et al. Prediction of breast cancer gains acceptance in oncologic centers, it is BCNU pulmonary toxicity in patients with malignant gliomas: imperative that both physicians and patients are educated an assessment of risk factors. New Engl J Med 1987; 303: as to the early manifestations and treatment of this syn- 183–188. drome. As noted by a recent National Heart, Lung and 10 Durant JR, Norgord MJ, Murad TM et al. Pulmonary toxicity Blood Institute Workshop summary, there is an urgent need associated with bis-chloroethyl nitrosourea (BCNU). Ann to further understand the pathogenesis of this pulmonary Intern Med 1979; 90: 191–194. syndrome and to develop new treatment strategies.18 We 11 Cooper JAD Jr, White DA, Matthay RA. Drug-induced pul- believe that our noninvasive scoring system can serve as a monary disease. Part I: cytotoxic drugs. Am Rev Respir Dis model for the early clinical diagnosis of lung toxicity both 1986; 133: 321–340. 12 Smith AC, Boyd MR. Preferential effects of 1,3-bis(2- in patients with and without symptoms. The initiation of chloroethyl)-1-nitrosourea (BCNU) on pulmonary glutathione corticosteroid therapy early in the course of the disease is reductase and glutathione/glutathione disulfide ratios: possible feasible and is associated with rapid resolution in most implications for lung toxicity. J Pharmacol Exp Ther 1984; patients. The therapeutic efficacy of such treatment, how- 229: 658–663. ever, needs to be clarified in a randomized group of 13 Gurtoo HL, Hipkens JH, Sharma SD. Role of glutathione in patients. We are further attempting to validate the scoring the metabolism-dependent toxicity and chemotherapy of system’s utility, as well as the integration of HRCT, in a cyclophosphamide. Cancer Res 1981; 41: 3584–3591. pilot study examining the role of prophylactic antioxidant 14 Cooper JAD Jr, Merrill WW, Reynolds HY. Cyclophospham- therapy in these patients. ide modulation of bronchoalveolar cellular populations and macrophage oxidative metabolism. Possible mechanisms of pulmonary pharmacotoxicity. Am Rev Resp Dis 1986; 134: 108–114. References 15 Sayegh A, Reece D, Barnett M et al. Interstitial pneumonitis (IP) following high-dose chemotherapy (CT) with cyclophos- 1 Rowlings PA, Horowitz MM, Armitage JO et al. Report from phamide, BCNU, etoposide ± cisplatin (CBV ± P) and auto- the International Bone Marrow Transplant Registry and the logous bone marrow transplantation (AuBMT) for advanced North American Autologous Bone Marrow Transplant Regis- Hodgkin’s disease (HD): incidence, risk factors and outcome. try. Clin Transplants 1993; 101–108. J Cell Biochem 1992; 205 (Suppl.): 16A (Abstr.). 2 Peters WP, Ross M, Vrendenburgh JJ et al. High-dose chemo- 16 Keane TJ, Van Dyk J, Rider WD. Idiopathic interstitial pneu- therapy and autologous bone marrow support as consolidation monia following autologous bone marrow transplantation. Int after standard-dose adjuvant therapy for high-risk primary J Radiat Oncol Biol Phys 1981; 7: 1365–1370. breast cancer. J Clin Oncol 1993; 11: 1132–1143. 17 Williams SF, Mick R, Dresser R et al. High-dose consoli- 3 Saez RA, Selby GB, Slease RB et al. Autologous bone mar- dation therapy with autologous stem cell rescue in stage IV row transplantation for metastatic breast cancer. J Ok State breast cancer. J Clin Oncol 1990; 7: 1824–1830. Med Assoc 1994; 87: 405–410. 18 Cherniak RM, Abrams J, Kalica AR. Pulmonary disease asso- 4 Todd NW, Peters WP, Ost AH et al. Pulmonary drug toxicity ciated with breast cancer therapy. Am Rev Resp Dis 1994; 150: in patients with primary breast cancer treated with high-dose 1169–1173.