A Detailed Evaluation of Cardiac Toxicity: A Phase II Study of Doxorubicin and One- or Three-Hour-Infusion Paclitaxel in Patients with Metastatic Breast Cancer¹

Metastatic breast cancer remains a leading cause of mortality in the United States. Despite recent advances in the understanding of this disease and the development of novel therapies, 39,600 women are expected to die from breast cancer in 2002 (1). Thus, the development of more effective therapies for metastatic disease is a high priority in cancer research.

Of particular interest in recent years has been anthracycline and taxane combinations. In the pre-taxane era, the anthracyclines, such as doxorubicin, were the most effective treatment for breast cancer. Response rates of 35–50% were seen from single-agent doxorubicin therapy (2) and 60–80% from combination therapy with FAC⁴ (3). As front-line therapy, response rates after single-agent taxane therapy range from 32 to 62% (4, 5, 6, 7, 8). In anthracycline-resistant disease, objective responses are seen in 24–50% of patients (5, 6, 7, 8, 9, 10, 11, 12). With the excellent antitumor activity of both doxorubicin and paclitaxel, as well as their nonoverlapping mechanisms of action, testing their combination became an active area of research.

In early paclitaxel and doxorubicin combinations [from the University of Texas M. D. Anderson Cancer Center (13) and the National Cancer Institute (14)], both agents were given by prolonged infusion, either simultaneously or concomitantly. Encouraging activity and no significant cardiac toxicity were reported. Subsequent Phase I/II studies of combination doxorubicin and paclitaxel produced high response rates but unexpected cardiotoxicity. Gianni et al. (15) and Gehl et al. (16) reported high response rates (83% to 94%) with 3-h paclitaxel and bolus doxorubicin. However, 18–20% of patients developed clinical CHF. Other Phase II studies have found no increased cardiac risk, but median doxorubicin dose was only 220–240 mg/m² (17, 18).

The schedule dependence of doxorubicin and paclitaxel is likely an important factor in the etiology of cardiac failure. When paclitaxel precedes doxorubicin, the clearance of doxorubicin is reduced by one-third, and when administration is separated by <1 h, doxorubicin elimination may also be slowed (19, 20). Thus, the excess cardiotoxicity may be attributable to decreased elimination of doxorubicin. The cardiac toxicity of this regimen may be further enhanced by the accumulation of doxorubicin in the heart when it is given in combination with paclitaxel (21). Separating the administration of the two drugs and limiting the cumulative dose of doxorubicin limits cardiac toxicity. In two studies that separated doxorubicin and paclitaxel by at least 16 h, the rates of CHF were not higher than would be expected from anthracyclines alone, although the total dose of doxorubicin was limited to 400 mg/m² in both trials (22, 23).

Two Phase III trials of combination doxorubicin and paclitaxel for patients with metastatic breast cancer have been reported. In the Eastern Cooperative Oncology Group 1193 study, doxorubicin was followed 4 h later by paclitaxel given by 24-h infusion (24). No excess cardiac toxicity was reported. Similarly, Jassem et al. (23) found minimal cardiotoxicity in 134 patients who were treated with doxorubicin followed 24 h later by paclitaxel. In this study, the total dose of doxorubicin was limited to 400 mg/m². It remains unclear whether separating the administration of the two drugs, limiting the total dose of doxorubicin, or a combination of the two factors is responsible for the lower incidence of cardiac failure.

Our Phase II study was initiated after the publication of the Gianni trial to confirm the efficacy and toxicity of bolus doxorubicin and short-infusion paclitaxel. The regimen used in the Gianni study was chosen because it was reported to produce much higher response rates than those reported from our own early Phase I and II studies, which used somewhat lower doses and a different schedule of administration. Potential cardiac toxicity was of concern; therefore, initially the total doxorubicin dose was restricted to 360 mg/m². In addition, the protocol required patients to have prospective noninvasive monitoring of cardiac function. Endomyocardial biopsies have established value in predicting which patients may develop CHF; therefore, selected patients were to be monitored with endomyocardial biopsies. However, because of the invasive nature of this procedure, cardiac biopsies remained an optional component of the protocol. There was also interest in comparing the relative tolerance of paclitaxel administered by 1-h infusion compared with 3-h infusion. It was anticipated that in the absence of significant differences in toxicity, the 1-h schedule would be more convenient.

The protocol enrolled patients from December 1995 through August 1998. Patients who had histological proof of breast carcinoma and progressive, measurable, or evaluable metastatic disease were eligible for this Phase II prospective randomized study. All patients were required to be ≥16 years of age, have a Zubrod performance status ≤2, and have an estimated life expectancy of at least 12 weeks. No prior chemotherapy for metastatic disease and no prior anthracycline or taxane therapy were allowed. Prior adjuvant chemotherapy with non-anthracycline- and non-taxane-containing regimens was permitted. Women of childbearing potential had to practice adequate contraception during treatment. All patients signed written informed consent in accordance with institutional and federal guidelines.

Adequate bone marrow, renal, and hepatic function was required for study entry and was defined as follows: granulocytes ≥1,500/μl, platelets ≥100,000/μl, serum creatinine ≤2.0, bilirubin ≤1.5 mg/dl, and alanine aminotransferase ≤2× upper limit of normal. Cardiac eligibility requirements were ejection fraction ≥55% by MUGA scan, no myocardial infarction within the past 6 months, no documented history of severe, persistent cardiac arrhythmia, and no wall motion abnormalities or perfusion defects on baseline cardiac MUGA scan. Patients with blastic bone metastases as their only site of disease were excluded.

Patients were prospectively randomized into one of two treatment arms: 60 mg/m² doxorubicin given by 30 min i.v. infusion, immediately followed by 200 mg/m² paclitaxel administered as a 1-h infusion or the identical regimen with paclitaxel given as a 3-h infusion. Premedication consisted of 20 mg of dexamethasone p.o. 12 and 6 h before chemotherapy and 50 mg of diphenhydramine and 300 mg of cimetidine, both given as i.v. bolus immediately before chemotherapy administration. Paclitaxel was diluted in 500 ml of D5W or normal saline for both the 1- and 3-h infusions. Maximum doxorubicin dose was initially limited to 360 mg/m²; but when no excess cardiac toxicity was identified, 8 patients were treated with a cumulative doxorubicin dose of 420 mg/m². At this dose level, a high incidence of cardiac dysfunction was noted; therefore, all remaining patients had a cumulative doxorubicin dose limited to 360 mg/m². Responding patients were continued on single-agent paclitaxel after completing six cycles of the combination. Chemotherapy was scheduled at 21-day intervals if recovery from hematological and nonhematological toxicity was complete. If delay was attributable to granulocytopenia, granulocyte-colony stimulating factor was added to subsequent cycles. If recovery from nonhematological toxicity was incomplete, treatment was delayed until symptoms returned to grade ≤1. Doses were reduced if platelet nadir was <50,000 or if the patient had bleeding, infection, or delay in therapy while on granulocyte-colony stimulating factor. Doses were also reduced for nonhematological toxicity ≥ grade 3 other than nausea and vomiting as defined by the National Cancer Institute Common Toxicity Criteria.

Baseline evaluation involved a complete history and physical examination with determination of Zubrod performance status; a complete blood count with differential, hepatic and renal function tests, biochemical profile, and tumor markers; and radiological evaluation as appropriate to determine extent and location of metastatic disease. Every cycle, patients were evaluated with interim history and physical examination, performance status determination, and laboratory studies. Assessment of tumor response by radiological evaluation was done after the second cycle and, in responding patients, after every three courses thereafter. Complete response was defined as disappearance of all evidence of tumor for at least 4 weeks. Partial response was defined as 50% or greater decrease in the sum of the products of longest diameters of all measurable lesions persisting for at least 4 weeks. Stable disease was defined as no change or a decrease in a measurable lesion that was too small or too brief to qualify as partial response, or minor growth that did not qualify as progressive disease for the duration of treatment. Progressive disease was defined as an increase of ≥25% in the sum of the products of longest perpendicular diameters of any measurable lesion or the appearance of a new lesion. Response duration was measured from the time of response until there was evidence of progressive disease. Times to treatment failure and survival were measured from the time of randomization.

The protocol required all patients to have an electrocardiogram and assessment of LVEF by multigated radionuclide angiography (MUGA scan) at baseline, before every other course starting after three courses, and at the time of withdrawal from study. If LVEF fell by ≥10%, MUGA scans were repeated after the next course of therapy. Cardiac biopsies were performed in the first 3 consenting patients in both arms after every other course, starting after 4 courses. Cardiac biopsies were to be performed in 6 consenting patients after cumulative doxorubicin dose of 360 mg/m². Cardiac biopsies were requested in patients whose LVEF fell below 50% or by ≥15% after a single course of therapy. Biopsy specimens were fixed in 2% buffered glutaraldehyde. From each biopsy, 8–20 2-mm tissue blocks were obtained. In each case, six plastic embedded tissue blocks were selected, thin-sectioned, double-stained with uranyl acetate and lead citrate, and examined with electron microscopy. Doxorubicin was discontinued if patients developed clinical signs of heart failure, ejection fraction fell to <45%, or endomyocardial biopsy was grade ≥1.5.

The primary end point of this study was objective response rate, with toxicity evaluated as a secondary end point. All patients were registered prospectively. Registration and randomization were performed electronically. Randomization was to compare the toxicity of 1-h versus 3-h paclitaxel. However, no difference in efficacy was anticipated between the two arms; therefore, the two arms were collapsed into one common database to determine efficacy. A two-phase Simon design was used (25), with a type I error rate of 5% and a type II error rate of 10%. Nineteen patients were to be entered in the first stage of the trial. If ≤12 of these patients responded, the trial was to be stopped. If >12 patients responded, the trial was to accrue a total of 80 patients. The χ² test, or Fisher’s exact test when appropriate, was used to compare categorical variables. Overall survival and time to progression were calculated by the Kaplan-Meier method.

A total of 82 patients were enrolled in the trial, with 41 patients in each arm. All patients were evaluable for toxicity, and 79 were considered evaluable for response. Three patients were not evaluable for response; 1 patient was removed from the study secondary to a protocol violation, 1 died of septic shock 9 days after registration, and 1 stopped paclitaxel because of an allergic reaction to the initial dose. Patient characteristics were well balanced between the two arms and are listed in Table 1. The median age of all patients was 53 years (range, 32–78 years). Ninety-five % of patients had a Zubrod performance status of 0 or 1. Overall, 19.5% of patients had received prior adjuvant chemotherapy, 23.2% had prior chest wall radiation, and 47.6% had prior hormonal therapy, either in the adjuvant setting or as treatment of metastatic disease. Significant differences between the two arms were identified in the number of patients who had prior chest wall radiation (7.3% versus 39.0%; P < 0.001) and who had received prior hormonal therapy (34.1% versus 61.9%; P = 0.015) with more patients in the 3-h paclitaxel arm having had prior therapies. Similar numbers of patients in both arms were hormone receptor positive; overall, 62% had estrogen receptor-positive tumors, and 51% had progesterone receptor-positive tumors. No patients had prior anthracycline or taxane therapy. The dominant site of disease in most patients was visceral (70.9%), followed by bone (22.0%) and soft tissue (7.3%). Visceral dominant disease was defined as visceral metastases ± bone metastases ± soft tissue metastases; bone dominant was bone metastases ± soft tissue; and soft tissue dominant was soft tissue disease only.

No differences were identified in response rates between the two arms (Table 2). Of the 79 evaluable patients, 46 (58.2%) had an objective response. Three patients (3.8%) achieved a complete response, and 43 patients (54.4%) had partial responses. Median follow-up of live patients was 37.5 months (range, 11–59 months). Median duration of response was 7.6 months (range, 2–49+ months), median time to progression was 7 months (range, 1–50+ months), and median overall survival was 31 months (range, 1–59+ months). Patients who had received prior adjuvant chemotherapy with cyclophosphamide, methotrexate, and 5-fluorouracil had significantly higher response rates than those who had no prior adjuvant therapy (81.2% versus 52.4%; P = 0.037). Response rates were higher in patients with nonvisceral disease (70.6%) than in patients with visceral metastases (51.8%), but this trend did not reach statistical significance (P = 0.07). Response rates did not differ between patients who had <3 versus ≥3 sites of disease.

All patients were evaluable for toxicity. Sixty-eight patients (83%) completed four or more cycles of combination chemotherapy, 63 patients (77%) completed five of more cycles, and 54 patients (66%) completed six or more cycles. The median cumulative doxorubicin dose was 360 mg/m² in both arms. The median number of cycles of subsequent single-agent paclitaxel was 2 in the 1-h paclitaxel arm (range, 0–9) and 3 in the 3-h paclitaxel group (range, 0–14). There was no significant difference in cardiotoxicity between the 1- and 3-h paclitaxel arms. Data on cardiac function are presented in Table 3. All patients had baseline MUGA scans; the median baseline EF was 62.5%. At the cumulative doxorubicin dose of 60–180 mg/m², 75 of 82 (91.4%) patients had MUGA scans performed; at 200–300 mg/m², 62 of 68 (92.1%) had MUGAs; at 310–360 mg/m², 18 of 52 (34.6%) had MUGAs; and at 420 mg/m², 4 of 8 (50%) had MUGA scans. The median ejection fraction at these dose levels was 60.0, 57.5, 52.5, and 32.0%, respectively. One patient who was treated to 420 mg/m² had the ejection fraction fall documented by echocardiogram. A consistent decline in ejection fraction was seen with increasing cumulative doses of doxorubicin (Fig. 1). Ten patients had a drop in ejection fraction by ≥15% within the normal range. Of these, 4 patients received further doxorubicin. Two of these patients had their ejection fractions increase to within 15% of their baseline value despite further doxorubicin, one patient had a stable ejection fraction, and one patient had a marked fall in ejection fraction and developed CHF after one additional dose. Fifteen patients (18.3%) had a ≥15% decline in ejection fraction to an absolute value of <50%. In 4 of 15 (26.7%), the ejection fraction recovered over time to >50%. Ten of 74 (13.5%) patients treated with a cumulative doxorubicin dose of ≤360 mg/m² developed a decline in ejection fraction as compared with 5 of 8 (62.5%) patients who received 420 mg/m² (P = 0.004). Age was also a significant predictor of ejection fraction decline. Patients ≥50 years of age had a relative risk of 3.56 of developing a fall in ejection fraction (P = 0.048). None of 8 patients who received prior left chest wall radiotherapy had a significant decline in ejection fraction.

Of the 15 patients with a significant fall in ejection fraction, 8 patients (53%) developed symptomatic cardiac dysfunction. Six had New York Heart Association class II, 1 had class III, and 1 was unknown. No patients had New York Heart Association class IV CHF. None of the eight patients had received prior radiotherapy to the left chest wall. Of the 8 patients who developed CHF, 4 developed symptoms during the course of chemotherapy with doxorubicin. The other 4 patients were diagnosed with CHF 2, 7, 12, and 24 months after the final dose of doxorubicin. None of these patients received additional cardiotoxic therapies. Risk of CHF with increasing cumulative doxorubicin dose is shown in Fig. 2. Four of 74 (5.4%) patients who were treated with a cumulative doxorubicin dose of ≤360 mg/m² developed CHF compared with 4 of 8 (50%) patients treated at 420 mg/m² (P = 0.002). One death was possibly attributable to CHF. The patient developed CHF 2 years after completing study, but her ejection fraction normalized and her symptoms resolved while on digoxin and an angiotensin-converting enzyme inhibitor. Three years after completing the study, she died suddenly at home, and no autopsy was performed.

Endomyocardial biopsy specimens were graded according to the Modified Billingham Scale (Table 4; Ref. 26). Results of the 23 endomyocardial biopsies performed on 16 patients are shown in Table 5. Twelve biopsies were grade 0, 6 biopsies were grade 0.5, and 5 biopsies were grade 1.0. No patients had a biopsy grade of ≥1.5, which would have required cessation of doxorubicin therapy. All patients who were biopsied had normal ejection fractions and no symptoms of CHF. However, 4 patients were biopsied because they had developed a fall in EF by 15% within the normal range. Three of these patients had biopsy grade 0, and one had biopsy grade 0.5. One of the patients with biopsy grade 0 developed CHF 1 year after the biopsy.

Noncardiac toxicity data are presented in Table 6. No significant differences in grade 3 or 4 toxicity were identified between the 1-h and 3-h paclitaxel groups. Overall, grade 3 or 4 granulocytopenia was seen in 63 patients (76.8%), grade 3 or 4 thrombocytopenia in 18 patients (22.0%), and grade 3 or 4 anemia in 15 patients (18.3%). There were no grade 4 infectious complications; however, 21 patients (25.6%) had grade 3 infections, 7 patients (8.5%) had grade 3 neutropenic fever, and 2 patients (2.4%) had grade 3 neutropenic infections. One patient in the 3-h paclitaxel arm died of nonneutropenic sepsis secondary to typhlitis. Of interest, 33 patients (80.5%) developed any grade thrombocytopenia in the 1-h paclitaxel arm as compared with 10 patients (24.4%) in the 3-h paclitaxel arm (P < 0.001). Nonhematological toxicity ≥ grade 3 that was seen in ≥5% of patients included fatigue (61.0%), myalgia (31.7%), nausea (29.3%), neurosensory (28.0%), stomatitis (26.8%), vomiting (14.6%), diarrhea (8.5%), skin reactions (6.1%), and headache (6.1%). No patients developed evidence of ethanol intoxication. In summary, there was no difference in toxicity between the two treatment schedules for any grade or grade 3 or 4 toxicity with the exception of thrombocytopenia, which was seen more frequently in patients treated with 1-h paclitaxel.

We performed a randomized Phase II study of bolus doxorubicin combined with either 1- or 3-h infusion paclitaxel. Both infusion schedules of paclitaxel were well tolerated, and the only significant difference in toxicity was an increased incidence of thrombocytopenia in the 1-h paclitaxel arm. Of note, the incidence of grade 3 or 4 thrombocytopenia was not significantly different between the two arms. Most patients on the study had clinical benefit from this combination of drugs, with 58% of patients achieving an objective response and over one-third with disease stabilization. Cardiac toxicity was a significant concern when this regimen was designed because ∼20% of patients treated with bolus doxorubicin/paclitaxel had been reported to develop CHF. This study did confirm the cumulative dose-dependent risk of cardiac toxicity of this regimen, both in its effects on ejection fraction and on the development of CHF. We found a continuous and marked decline in ejection fraction with increasing cumulative dose of doxorubicin. However, we recognize that these data must be interpreted with caution because not all patients had the required MUGA scans, and those patients who were especially at risk for a decline in cardiac function may have been more likely to have had scans performed. Despite these limitations in the data, we believe that this decline in ejection fraction reflects an important toxicity of this combination of doxorubicin/paclitaxel. The validity of these data are supported by similar findings of a fall in ejection fraction with increasing doxorubicin dose reported by other investigators (15, 16). Although many patients had some degree of decline in ejection fraction over the course of the study, almost 20% of patients had a substantial decline. One-quarter of affected patients had a documented recovery of cardiac function. Women who had a fall in ejection fraction were clearly at increased cardiac risk because 53% of these patients developed CHF.

We also confirmed that limiting the cumulative dose of doxorubicin to 360 mg/m² results in a low incidence of clinically relevant cardiac toxicity, as reported previously by Gianni et al. (27). The risk of developing CHF was clearly linked to the cumulative dose of doxorubicin. A sharp increase in the incidence of CHF was seen in patients who received >360 mg/m² of doxorubicin, with half of these patients developing CHF. For patients who received <360 mg/m² of doxorubicin, the risk of developing CHF was 5%, which is not significantly higher than expected with doxorubicin alone. At comparable doxorubicin doses, the expected rate of CHF has been reported to be 2–8% (28, 29, 30, 31). Our data indicate that the risk from paclitaxel/doxorubicin combination therapy falls safely within that range when the total dose of doxorubicin is restricted to 360 mg/m².

Cardiac biopsies were requested as part of this study to help define which patients were at increased risk of developing cardiac toxicity so that potentially cardiotoxic therapy could be discontinued before any serious sequelae. Myocardial biopsies are done on an outpatient basis and have been shown to be quite safe. At Stanford, over 10,000 biopsies have been done with no fatal complications, although 5 patients had cardiac perforation (32). At our institution, we have performed over 2000 endomyocardial biopsies, with a similar safety record. Biopsy tissue specimens are evaluated by both routine light microscopy and electron microscopy to identify changes associated with anthracycline toxicity (26, 33). Patients who have biopsy grade ≥1.5 have a >20% chance of cardiac failure with continued therapy (26). Unfortunately, few patients consented to undergo cardiac biopsy, and no patients with cardiac toxicity were biopsied. Therefore, the data collected are insufficient to draw any conclusions regarding the predictive value of cardiac biopsy for patients treated with both paclitaxel and doxorubicin.

Although the Gianni et al. (15) and Gehl et al. (16) studies had suggested that up to 90% of patients treated with combination doxorubicin/paclitaxel would respond, other Phase II studies have found more modest overall response rates ranging from 43 to 80% (15, 16, 17, 18, 22, 34, 35, 36), and the two Phase III studies reported have shown response rates of 68 and 46% (23, 24). Jassem et al. (23) compared FAC to doxorubicin and paclitaxel as front-line therapy for patients with metastatic breast cancer. They found improved overall response rates in the patients treated with doxorubicin/paclitaxel (68% versus 55%; P = 0.032). In addition, this study demonstrated a 5-month survival benefit for these patients (23.3 months versus 18.3 months; P = 0.013). However, only 24% of patients on the FAC arm received a taxane upon progression. Therefore, although doxorubicin/paclitaxel appears to be superior to FAC as front-line therapy, this study did not resolve whether doxorubicin/paclitaxel combination therapy is better than sequential therapy with an anthracycline followed by a taxane. In fact, the one large study that addresses this question has found equivalent survival between combined versus sequential therapy. Sledge et al. (24) presented the data from the Eastern Cooperative Oncology Group 1193 study, which compared doxorubicin, paclitaxel, and combination doxorubicin/paclitaxel. The patients treated with single-agent paclitaxel or doxorubicin were crossed over to the other therapy at time of progression. This study reported an overall response rate of 46% for combination therapy, which was significantly higher than the response rate for either agent alone. However, overall survival for patients treated with combination therapy was 22.4 months, which was not superior to survival for patients who received sequential therapy with the same two drugs. Our overall response rate of 58% is comparable with the response rates seen in these Phase II and III studies. We did find a longer overall survival than the range that has been reported previously. Our median overall survival was 31 months compared with 22–23 months in previous studies. The significance of this lengthened survival is unclear because this may simply reflect a select patient population from a single-institution study.

The high response rates and possible survival benefit of combination paclitaxel and doxorubicin have generated enthusiasm for other taxane/anthracycline combinations that may have less cardiac toxicity. Docetaxel, in contrast to paclitaxel, does not appear to inhibit the clearance of doxorubicin (37, 38); and regimens of docetaxel and doxorubicin have produced response rates ranging from 57 to 77% without significant cardiac toxicity (39, 40, 41). A Phase III study comparing doxorubicin/docetaxel to doxorubicin/cyclophosphamide reported higher response rates and longer time to progression in the doxorubicin/docetaxel arm without excess cardiac toxicity (42). Thus, docetaxel-based regimens are efficacious with less risk of CHF.

The incidence of cardiac damage may be further diminished by the use of other anthracyclines such as epirubicin or liposomal doxorubicin, which may have less cardiotoxicity than doxorubicin. The combination of epirubicin and paclitaxel has been evaluated in multiple studies and has produced response rates of 50–70% (43, 44, 45, 46, 47, 48). Pharmacokinetic studies show that giving paclitaxel and epirubicin together does not result in an increase in active metabolites, as is the case with paclitaxel and doxorubicin, and thus this combination does not produce excess cardiotoxicity (43, 44, 45, 46, 47, 48, 49). The optimal anthracycline/taxane regimen with the best therapeutic ratio has yet to be determined.

Other approaches to limiting cardiac toxicity have included separating the taxane and anthracycline administration (22, 23) and the addition of dexrazoxane, a cardioprotectant, to anthracycline-containing regimens (29, 50). Twenty-five patients were treated with doxorubicin, paclitaxel, and dexrazoxane; none had significant falls in ejection fraction or CHF (51).

In conclusion, combination doxorubicin and paclitaxel is a safe and effective regimen for metastatic breast cancer when cumulative doxorubicin dose is limited to 360 mg/m². This regimen may even have utility in the adjuvant or neoadjuvant setting, because other authors have found that this combination can produce high response rates without excess cardiac toxicity when used as a short neoadjuvant course (52). Ultimately, the clinical application of this regimen will depend on multiple factors, including the presence of cardiac risk factors, the need for the highest response rate because of high-risk primary tumors, and the patients’ understanding of the risk:benefit ratio for this and other combinations.