weekly trastuzumab and paclitaxel therapy for metastatic breast cancer with analysis of efficacy by her2 immunophenotype and gene amplification Weekly Trastuzumab and Paclitaxel Therapy for Metastatic Breast Cancer With Analysis of Efficacy by HER2 Immunophenotype and Gene Amplification

Weekly Trastuzumab and Paclitaxel Therapy for Metastatic Breast Cancer With Analysis of Efficacy by HER2 Immunophenotype and Gene Amplification

  1. Clifford A. Hudis
  1. From the Breast Cancer Medicine Service, Department of Pathology, Department of Radiology, Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY; and Department of Breast Medical Oncology, The University of Texas M.D. Anderson Cancer Center, Houston, TX.
  1. Address reprint requests to Andrew D. Seidman, MD, Breast Cancer Medicine Service, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021; email: seidmana{at}mskcc.org
 
Next Section

Abstract

PURPOSE: This phase II study evaluated weekly trastuzumab and paclitaxel therapy in women with HER2-normal and HER2-overexpressing metastatic breast cancer. Efficacy was correlated with immunohistochemical and fluorescent in situ hybridization (FISH) assay results.

PATIENTS AND METHODS: Eligible patients had bidimensionally measurable metastatic breast cancer. Up to three prior chemotherapy regimens, including prior anthracycline and taxane therapy, were allowed. Trastuzumab 4 mg/kg and paclitaxel 90 mg/m2 were administered on week 1, with trastuzumab 2 mg/kg and paclitaxel 90 mg/m2 administered on subsequent weeks. HER2 status was evaluated using four different immunohistochemical assays and FISH.

RESULTS: Patients received a median of 25 weekly infusions (range, one to 85 infusions). Median delivered paclitaxel dose-intensity was 82 mg/m2/wk (range, 52 to 90 mg/m2/wk). The intent-to-treat response rate for all 95 patients enrolled was 56.8% (95% confidence interval, 47% to 67%). A response rate of 61.4% (4.5% complete response, 56.8% partial response) was observed in 88 fully assessable patients. In patients with HER2-overexpressing tumors, overall response rates ranged from 67% to 81% compared with 41% to 46% in patients with HER2-normal expression (ranges reflect the different assay methods used to assess HER2 status). Differences in response rates between patients with HER2-overexpressing tumors and those with normal HER2 expression were statistically significant for all assay methods, with CB11 and TAB250 antibodies and FISH having the strongest significance. Therapy was generally well tolerated, although three patients had serious cardiac complications.

CONCLUSION: Weekly trastuzumab and paclitaxel therapy is active in women with metastatic breast cancer. Therapy was relatively well tolerated; however, attention to cardiac function is necessary.

THE HER2 gene encodes a 185-kd transmembrane glycoprotein with tyrosine kinase activity that is a member of the epidermal growth factor receptor family.1,2 Overexpression of HER2 has important biologic implications. In preclinical studies, HER2 overexpression has been found to contribute to oncogenic transformation, tumorigenesis, and metastatic potential.3,4 HER2 overexpression in women with breast cancer is considered a negative prognostic factor, as several studies have found a correlation between HER2 overexpression and shorter disease-free and overall survival.5-8 Overexpression of HER2 has also been associated with potentially more aggressive tumors.

HER2 is overexpressed in 25% to 30% of human breast cancers.5,9,10 Evaluation and interpretation of tumor HER2 expression is of great clinical interest and the subject of much editorial.9,11-19 There are currently several different methods available to evaluate HER2 status, the most common being immunohistochemical (IHC) assays, which evaluate protein expression, and fluorescence in situ hybridization (FISH) assays, which evaluate gene amplification. There is inherent variability in the comparative results of different assay methods evaluating HER2 expression. Cellular or protein targets may differ among the IHC methods, and additional variability is introduced by operator technique and scoring criteria.12,14,15 There is also potential biologic heterogeneity between the degree of gene amplification and resulting protein expression.9,13,16,18-21 Thus, although FISH and IHC results are often in agreement, discrepancies are known to occur. Reminiscent of the development of optimal testing for estrogen receptor three decades ago, definition and standardization of a HER2 assay procedure that provides the most reliable information for prognostic and treatment purposes is an ongoing process.

The HER2 protein is a viable therapeutic target. In preclinical studies, monoclonal antibodies directed against this protein demonstrated growth inhibition in HER2-overexpressing tumor cells.22 The humanized anti-HER2 monoclonal antibody trastuzumab (Herceptin; Genentech, South San Francisco, CA) has demonstrated activity in clinical trials in women with metastatic breast cancer overexpressing HER2. Response rates to the antibody given as a single agent have ranged from 12% to 27%.23-25

Clinical trials evaluating combination trastuzumab and chemotherapy have been initiated based on preclinical evidence of potentially enhanced antitumor activity when anti-HER2 antibodies were combined with traditional chemotherapeutic agents.26-28 In a phase II trial in women with refractory metastatic breast cancer, the use of anti-HER2 antibody in combination with cisplatin resulted in higher response rates than previously reported for cisplatin alone.29 In a recent randomized phase III trial, the efficacy and safety of trastuzumab combined with chemotherapy, either doxorubicin plus cyclophosphamide or single-agent paclitaxel, was compared with chemotherapy alone in women with HER2-overexpressing metastatic breast cancer.30,31 Patients who received trastuzumab in combination with chemotherapy had a significantly longer time to progression, higher overall response rate, and longer median overall survival compared with patients who had received chemotherapy alone. Although the combination therapy was relatively well tolerated, cardiac dysfunction was more common in patients treated with the antibody in combination with chemotherapy than in those receiving chemotherapy alone.32 Cardiac dysfunction was much more common and often more severe in patients who had received trastuzumab in combination with doxorubicin and cyclophosphamide compared with those who had received the antibody with paclitaxel.

Trastuzumab is administered on a weekly schedule, and coadministration with a weekly chemotherapy regimen had not yet been evaluated. Weekly paclitaxel therapy is effective and well tolerated in women with metastatic breast cancer.33-39 By using a weekly administration schedule for both agents, we sought to further exploit the response and survival benefits of the combination of trastuzumab and paclitaxel observed previously.30,31 Therefore, we initiated this phase II trial of weekly trastuzumab and paclitaxel therapy in women with metastatic breast cancer to evaluate the safety and efficacy of the combination on this schedule.

Previous clinical studies of trastuzumab in advanced breast cancer limited entry to women with HER2-overexpressing tumors. However, it should be recognized that HER2 expression is a continuous variable, and definitions of normal versus overexpression have been empirically derived. Normal HER2 protein expression does not indicate the absence of HER2 receptors, but rather an insufficient receptor density to translate into a positive reading by current IHC testing criteria. Thus, even tumors classified as HER2 normal express a potentially viable protein target. Therefore, we opted to evaluate the activity of combination trastuzumab and paclitaxel in women with HER2 overexpression and those with normal HER2 expression, and the study was designed to enroll an equal number of each. HER2 status was assessed using four different IHC assays and FISH to investigate assay concordance and the association between assay results and clinical outcome.

Previous SectionNext Section

PATIENTS AND METHODS

Eligibility

Women with histologically confirmed, bidimensionally measurable metastatic breast cancer were eligible for study. Patients had to be ≥ 18 years of age, have a Karnofsky performance status of ≥ 70, and provide written informed consent. Treatment with up to three prior chemotherapy regimens for breast cancer, including adjuvant, neoadjuvant, and metastatic therapy, was allowed. Prior anthracycline therapy was allowed, as was prior taxane therapy for metastatic disease provided that at least 1 year had elapsed since last taxane exposure. Any prior hormonal therapy must have been discontinued at least 3 weeks before study entry. Laboratory criteria included: absolute granulocyte count ≥ 1,500/μL, hemoglobin ≥ 8 g/dL, platelet count ≥ 100,000/μL, serum creatinine less than 2.0 mg/dL, serum bilirubin less than 1.5 mg/dL, serum calcium less than 11.0 mg/dL, and left ventricular ejection fraction (LVEF) ≥ 50%.

Patients with a history of significant cardiac disease, including serious arrhythmias or New York Heart Association Functional Class III/IV cardiac status, were excluded. Also excluded were patients with brain metastases, a history of grade 3/4 peripheral neuropathy of any etiology, and those who had received radiation to more than 50% of marrow-bearing bone. Patients who had received any prior monoclonal antibody treatment were excluded. Patients were ineligible if they had a history of other malignancy (except for carcinoma-in-situ of the cervix or nonmelanoma skin cancer) or had other serious medical illness. All patients must have recovered from any prior chemotherapy, radiotherapy, or surgery before study entry. Women of childbearing potential must have had a negative pregnancy test before enrollment and practice appropriate contraception while on study.

Prestudy evaluation included a complete physical examination and history, complete blood count with differential, platelet count, comprehensive serum chemistries, electrocardiogram, and multigated aquisition scan or echocardiogram. Chest x-rays, bone scans, and computed tomography imaging were performed as appropriate for disease assessment. HER2 status was determined at study entry using the TAB250 antibody (methodology below).

Treatment

On the first treatment week, trastuzumab 4 mg/kg was administered intravenously (IV) over 90 minutes on day 0 followed by paclitaxel 90 mg/m2 IV over 1 hour on day 1. On subsequent weeks, administration of paclitaxel 90 mg/m2 IV by 1-hour infusion was followed immediately by trastuzumab 2 mg/kg over 30 minutes. All therapy was delivered in the outpatient clinic. Treatment was administered every 7 days until disease progression or prohibitive toxicity. However, patients with tumors having normal HER2 expression who did not achieve at least a partial response by week 28 had trastuzumab therapy discontinued but could continue to receive weekly paclitaxel. In patients with HER2-overexpressing tumors treated to week 28 without disease progression, treating physicians at their discretion could discontinue paclitaxel therapy and maintain weekly trastuzumab treatment.

Premedications consisted of dexamethasone 10 mg IV, diphenhydramine 50 mg IV, and cimetidine 300 mg IV, administered 30 to 60 minutes before the paclitaxel infusion. If patients did not experience any manifestation of a hypersensitivity or allergic reaction after the first eight infusions, dexamethasone was dose reduced as follows: week 9 and 10, dexamethasone 8 mg IV; week 11 and 12, dexamethasone 6 mg IV; and week 13 and beyond, dexamethasone 4 mg IV.

Paclitaxel was dose reduced for a platelet nadir ≤ 50,000/μL or infection. Therapy was held pending recovery of absolute neutrophil count to 1,000 cells/μL, platelets more than 100,000/μL, and resolution of any nonhematologic toxicity to grade 2 or less. Paclitaxel was dose reduced in increments of 10 or 20 mg/m2 in the event of grade 2 or 3 nonhematologic toxicity, respectively. Paclitaxel was dose reduced in patients requiring more than 2 weeks to recover from any toxicity.

Response and Toxicity Assessment

Physical examinations, complete blood counts, and biochemistry profiles were repeated every other treatment week. Toxicities were graded according to the National Cancer Institute common toxicity scale. Repeat MUGA scans or echocardiograms were performed at weeks 8, 16, 28, 40, 52, every 12 weeks thereafter, and at any other time deemed necessary.

Radiographic evaluations for tumor response were performed at weeks 8 and 16, and then every 12 weeks thereafter. Responses were defined as follows: a complete response was the disappearance of all clinical and radiographic signs of tumor for at least 4 weeks; a partial response was a more than 50% reduction in the sum of products of the biperpendicular diameters of all measurable lesions with no increase in size of any lesion and no new lesions; a minor response was a 25% to 49% reduction in the sum of products of the biperpendicular diameters of all measurable lesions with no increase in size of any lesion and no new lesions; stable disease was a less than 25% reduction in the sum of products of the biperpendicular diameters of all measurable lesions with no increase in size of any lesion and no new lesions; and disease progression was a 25% or greater increase in size of any lesion or the appearance of any new lesion.

Analysis of HER2 Expression

The HER2 status of 80 primary and 15 metastatic tumors was assessed by IHC and FISH on formalin-fixed paraffin-embedded tissue. IHC was performed using four different antibodies, two of which were polyclonal and two of which were monoclonal: (1) DAKO HerceptTest (rabbit antihuman HER2/neu polyclonal antibody; DAKO Corporation, Carpinteria, CA); (2) Pab1 (rabbit antihuman polyclonal antibody; Zymed Laboratories, South San Francisco, CA); (3) TAB250 (mouse antihuman monoclonal antibody; Zymed); and (4) CB11 (mouse antihuman monoclonal antibody; Ventana Medical Systems Inc, Tucson, AZ) ( Fig 1). Assays were preformed according to the manufacturers’ instructions. The intensity of membrane staining was evaluated according to the following criteria set forth by the DAKO HerceptTest: score 0, no or up to 10% membrane staining; score 1+, partial and/or faint membrane staining present in more than 10% of tumor cells; score 2+, weak to moderate complete membrane staining present in more than 10% tumor cells; and score 3+, strong complete membrane staining present in more than 10% of tumor cells. Scores 0 and 1+ were considered normal (negative for overexpression) and scores 2+ and 3+ were considered positive for HER2 overexpression.

Figure
View larger version:

Fig 1. Diagram of HER2 receptor protein indicating antibody-binding domains.

FISH analysis was performed using the PathVysion HER2 probe kit (Vysis Inc, Downers Grove, IL) according to the manufacturer’s instructions. This system uses a spectrum orange fluorophore-labeled DNA probe for the HER2 gene locus, a spectrum green fluorophore-labeled DNA probe for chromosome 17, and a DAPI counterstain for nuclei. Spectrum orange and green signals were counted on 60 cells per case, with a ratio of orange:green signals ≥ 2 considered positive. Two pathologists evaluated FISH and IHC analyses independently and in a blinded fashion.

Statistical Methods

The study was designed to enroll 50 patients each with HER2 normal and HER2-overexpressing metastatic breast cancers. With this sample size, a 95% confidence interval (CI) of the probability of response could be calculated with a maximum width of +/− 14% for each group. The association between clinical response and assay results was evaluated by normal χ2. Kappa was estimated along with its 95% CI to evaluate concordance among HER2 assay methods. In general, a kappa value of 1 denotes complete agreement, values more than 0.75 denote excellent agreement, values between 0.4 and 0.75 denote fair to good agreement, values more than 0 but less than 0.4 denote poor agreement, and a kappa value of 0 indicates that the observed agreement is equal to chance.

Previous SectionNext Section

RESULTS

A total of 95 patients were enrolled at two institutions (71 at Memorial Sloan-Kettering Cancer Center, New York, NY, and 24 at M.D. Anderson Cancer Center, Houston, TX). Patient characteristics are listed in Table 1. Initially, we intended to enroll 50 patients whose tumors overexpressed HER2 by TAB 250 and 50 patients whose tumors lacked overexpression. A.D.S. permitted patients to enroll at M.D. Anderson Cancer Center based on local IHC results; subsequently, all pathology blocks were sent to Memorial Sloan-Kettering Cancer Center for the full panel of IHC and FISH testing. This resulted in a less than 50:50 division between HER2 overexpressors and nonoverexpressors as determined by the TAB 250 index assay. Primary tumors were the source of testing for 80 patients, metastatic sites were the source for 15 patients.

View this table:

Patient Characteristics (n = 95)

A total of 2,619 weekly infusions were delivered, with patients receiving a median of 25 weekly treatments (range, one to 85 treatments). The median delivered paclitaxel dose was 82 mg/m2/wk (range, 52 to 90 mg/m2). Therapy delays occurred in 7% of weekly treatments, 4% were a result of medical reasons and primarily neurotoxicity, and 3% occurred because of patient requests for social reasons.

Therapy was generally well tolerated ( Table 2). Serious hematologic toxicity was infrequent, with grade 3 or 4 neutropenia occurring in 6% of patients and grade 3 anemia occurring in 2% of patients. neuropathy was the most common nonhematologic toxicity. Significant neuropathy occurred in 27 (28%) of 95 patients, with grade 4 toxicity occurring in one patient. There was one death on study caused by a nonneutropenic infection.

View this table:

Toxicity Profile of Weekly Trastuzumab and Paclitaxel Therapy (n = 95)

infusion-related reactions were infrequent. The dose of dexamethasone was successfully de-escalated as planned in all eligible patients who received therapy beyond week 8.

Cardiac function was monitored every 8 weeks for the first two evaluations and subsequently every 12 weeks. The majority of patients were assessed by MUGA, although some patients (< 10%) were assessed by echocardiogram. The results of serial assessments of left ventricular contractility are depicted in Fig 2. The median LVEF for patients treated on study remained relatively stable throughout the first year. Median LVEF values were as follows: baseline, 64%; week 8, 64%; week 16, 61%; week 28, 61%; week 40, 60%; and week 52, 61%.

Figure
View larger version:

Fig 2. Plot of serial MUGA values in patients receiving weekly trastuzumab and paclitaxel therapy. Horizontal bars represent median values. Dotted lines represent LVEF values over time in two patients experiencing symptomatic congestive heart failure.

We noted a relative decrease in LVEF of more than 20% from baseline in seven patients who remained clinically asymptomatic. Prior anthracycline exposure did not exceed 400 mg/m2 in any of these patients. Weekly trastuzumab and paclitaxel therapy was continued in all of these patients without dose modification. The LVEF returned to baseline value in three patients, whereas two additional patients had relative improvements in LVEF. Two of these patients remain on study at this time. The LVEF remained low in two patients, although none were clinically symptomatic at the time of last follow-up. Only two patients received medical management with angiotensin-converting enzyme inhibitors, diuretics, or digoxin. Treatment was discontinued in six of these patients because of disease progression or noncardiac related issues.

Serious cardiac complications occurred in three patients. The first patient had a prior cumulative doxorubicin exposure of 615 mg/m2, approximately half of which was given as adjuvant therapy and the remaining given in conjunction with dexrazoxane as therapy for metastatic disease. The doxorubicin-based therapy for metastatic disease was discontinued approximately 4 weeks before her receiving trastuzumab and paclitaxel. Her tumor was positive for HER2 overexpression. At study entry, her LVEF was below 50% and subsequently deteriorated to 10% after 7 weeks of therapy. She was removed from the trial, and after treatment with digoxin, a diuretic, and an angiotensin-converting enzyme inhibitor, her LVEF improved to 20% to 30%. She eventually died 1 year later from brain metastases. The second patient had previously received four courses of doxorubicin and cyclophosphamide as adjuvant therapy for breast cancer; her cumulative doxorubicin exposure was 240 mg/m2. She was diagnosed with metastatic breast cancer approximately 2.5 years later and enrolled on this study. After 8 months of weekly trastuzumab and paclitaxel therapy, she developed an acute myocardial infarction and underwent angioplasty. She had severe left ventricular dysfunction and a LVEF of 35% to 40% at the time of cardiac catheterization. Her cardiac dysfunction was then managed medically, and she was removed from study. Approximately 2 months later, she had a second myocardial infarction; her LVEF at the time was 30%. She recovered, but her LVEF remained low, consistent with ischemic cardiomyopathy. She received 3 months of therapy with weekly paclitaxel 80 mg/m2 and has maintained stable disease. A third patient had been on study for 6 months. She developed influenza A pneumonia and was hospitalized. Shortly after being intubated for respiratory failure secondary to pneumonia, she had a massive anterior wall myocardial infarction. She died of acute respiratory failure secondary to pneumonia. Her prior chemotherapy was four courses of paclitaxel followed by four courses of fluorouracil, doxorubicin, and cyclophosphamide as adjuvant therapy. One additional patient experienced grade 1 supraventricular tachycardia while on study but had no cardiac sequelae.

Eighty-eight of the 95 enrolled patients received at least 8 weeks of therapy and were fully assessable for response. Seven patients were withdrawn from study before the first planned response evaluation: two patients had occult CNS metastases that manifested shortly after enrollment; four patients developed early toxicity; and one patient missed therapy for 2 consecutive weeks for social reasons. The overall response rate in the intent-to-treat population of 95 patients was 56.8% (95% CI, 47% to 67%) ( Table 3). The overall response rate in the 88 assessable patients was 61.48%, with four complete responses (4.5%) and 50 partial responses (56.8%). Minor responses or stable disease were observed in 30% of assessable patents. Only 8% of patients had direct disease progression without any period of at least stable disease while on protocol therapy. The median response duration was 7 months (range, 2 to 21 months).

View this table:

Response to Therapy

Results of the TAB250 assay were available for 94 patients, with HercepTest, Pab1, and CB11 results available for 90 patients. FISH assay results were available for 79 patients. When assay results were unavailable, it was due primarily to insufficient tissue to compete all assays or lack of availability of additional tissue samples. Figure 3 depicts the percent of patients scored at each staining level for the four IHC assays. The percent of patents with normal HER2 expression (scores of 0 or 1) and those with HER2 overexpression (scores of 2 or 3) by the IHC tests are compared with the results of FISH analysis in Fig 4. Of note, the percent of patients with HER2 overexpression by the TAB 250 antibody, the antibody used to assign HER2 status at study entry was 39%, despite our original intention to enter an equal number of patients with HER2-overexpressing tumors based on TAB 250 IHC staining. This imbalance resulted from interinstitutional differences in interpretation of immunohistochemical assay results (24 HercepTest-positive cases from M.D. Anderson Cancer Center were submitted for the full IHC panel performed at Memorial Sloan-Kettering Cancer Center) and interassay discordance (ie, polyclonal HercepTest overexpressors which lacked overexpression by the monoclonal TAB 250 antibody). However, by FISH analysis, the study population was equally balanced between patients with HER2-amplified and -nonamplified tumors.

Figure
View larger version:

Fig 3. Results of IHC staining with antibodies directed against HER2 protein. Numbers above bars represent actual percentage of patients.

Figure
View larger version:

Fig 4. Comparative results of FISH and IHC assays. Numbers above bars represent actual percentage of patients.

Concordance of the results between HercepTest and FISH for the 78 patients in whom data from both assays were available is listed in Table 4. The kappa value of 0.72 (95% CI, 0.56 to 0.87) suggests that there was good concordance between FISH and HercepTest in our population. There was a higher degree of concordance between FISH and the CB11 antibody, with a kappa of 0.80 (95% CI, 0.66 to 0.93), indicating excellent agreement ( Table 5).

View this table:

Concordance of HER2 Assay Results: FISH Versus HercepTest

View this table:

Concordance of HER2 Assay Results: FISH Versus CB11

We also evaluated the concordance of the polyclonal antibody used in the HerceptTest with that of the monoclonal antibody CB11. There was a high degree of concordance (kappa = 0.78) between these IHC tests ( Table 6). Of interest, there were 10 patients for whom assay results differed, and all had overexpression according to the HercepTest and normal expression by CB11.

View this table:

Concordance of HER2 Assay Results: Monoclonal Antibody (CB-11) Versus Polyclonal Antibody (HercepTest)

Clinical response rates relative to HER2 assay results for the four IHC assays and FISH are listed in Table 7. The overall response rate in patients with normal HER2 expression ranged from 41% to 46%, whereas responses in HER2-overexpressing patients ranged from 67% to 81%. The differences in response rates between patients with normal HER2 expression and HER2 overexpression were statistically significant for all assay methods used. The strongest correlation between assay results and response occurred with FISH and the monoclonal antibodies CB11 and TAB250, with a log difference in the significance level compared with the polyclonal antibodies HercepTest and Pab1.

View this table:

Correlation of Overall Response With HER2 Assay Results

Previous SectionNext Section

DISCUSSION

The evaluation and interpretation of HER2 expression and the clinical application of this information continues to evolve in women with metastatic breast cancer. In this phase II trial, we evaluated the safety and efficacy of trastuzumab and paclitaxel administered weekly in women with both HER2-overexpressing and HER2-normal tumors. We also compared the concordance and predictive value of FISH and four different IHC assays in this population.

Therapy with combined weekly trastuzumab and paclitaxel was active and generally well tolerated. We intended to enroll an equal number of patients with normal and overexpression of HER2, thus the percent of patients in our study with HER2 overexpression is higher than would be expected in the general population. The overall response rate in the intent-to-treat population was 56.8%. However, when responses were evaluated by HER2 status, there was a statistically significant difference in overall response rate favoring patients with HER2 overexpression. Depending on the assay method used to evaluate HER2 expression, the overall response rate in patients with HER2-overexpressing tumors ranged from 67% to 81%, compared with 41% to 46% in patients with tumors having normal HER2 expression. These differences were statistically significant regardless of the assay method used to assess HER2 status. However, this trial was exploratory in this regard and not designed to address the influence of HER2 overexpression on response rate. The relative benefit or lack thereof for paclitaxel with or without trastuzumab in this population is an important question currently under prospective investigation in a large randomized trial, CALGB 9840 ( Fig 5).

Figure
View larger version:

Fig 5. Schema of CALGB 9840.

In previous clinical studies, paclitaxel administered as a single agent in doses similar to that used in this study produced response rates ranging from 21% to 68% in women with metastatic breast cancer.33-39 The clinical trial reported by Slamon et al30 in women with HER2-overexpressing tumors found an overall response rate of 57% for combination paclitaxel and trastuzumab compared with 25% for paclitaxel alone.30 Our data indicate a very high response rate for the combination of weekly paclitaxel and trastuzumab in women with HER2-overexpressing tumors and is suggestive of a higher response rate than would be expected from paclitaxel alone. The relative efficacy, safety, and impact on quality of life of the conventional every 3-week dosing of paclitaxel at a dose of 175 mg/m2 by 3-hour infusion with weekly trastuzumab compared with the weekly regimen evaluated in this study is presently being studied in CALGB 9840.

When patients with minor responses and stable disease are considered, the vast majority of patients, 85%, obtained at least some clinical benefit from the combination of trastuzumab and paclitaxel. In the intent-to-treat population, only 7% of patients had disease progression as their best response. Of note, two patients had rapid CNS progression and were withdrawn from study. The CNS was the first and only site of disease progression in eight patients in this trial. These patients, after being removed from study secondary to disease progression, continued trastuzumab therapy during whole brain irradiation, with six of eight patients experiencing at least 4 additional months without evidence of other systemic progression outside the CNS. Neither trastuzumab nor paclitaxel penetrates well into the CNS.40,41 The CNS may be a sanctuary site for disease in patients treated with trastuzumab and paclitaxel.

Therapy was generally well tolerated. Hematologic toxicity was limited to a 6% incidence of grade 3/4 neutropenia and a 2% incidence of grade 3 anemia, similar to what would be expected with single-agent paclitaxel. Four patients (4.2%) received transfusion of a total of 7 units of packed RBCs. Erythropoietin was used subsequently in these patients, as well as in three other patients who were not transfused. Aside from neuropathy, other grade 3/4 nonhematologic toxicities were infrequent.

Cardiac contractility was serially monitored in all patients receiving weekly trastuzumab and paclitaxel therapy. In the majority of patients, LVEF was maintained during study treatment, reflected by the relatively consistent median LVEF we observed through 1 year of therapy. There were three patients who had serious cardiac complications. Because there is the potential for cardiac dysfunction in patients treated with trastuzumab,23,32 we recommend close monitoring of cardiac contractility in patients receiving the antibody. The safety of trastuzumab therapy in patients with baseline cardiac dysfunction is unknown, and the possible benefits of trastuzumab therapy must be seriously weighed against potential cardiac risk in these patients. Clinicopathologic studies are underway to identify the molecular mechanisms of trastuzumab-associated cardiac toxicity. It is hoped that this information may eventually assist clinicians in defining patients at risk for cardiac complications.

Regarding the comparative results of the IHC and FISH assays, we found the concordance between the monoclonal antibody CB11 and FISH to be higher than that of the HercepTest and FISH. It has been noted that the HercepTest can yield false-positive results when compared with FISH, and these are attributed to a lower degree of specificity of the HercepTest as conducted according to the manufacturer’s instructions.13 It is interesting that in our comparison of the assay results of the monoclonal CB11 antibody and the polyclonal antibody of the HercepTest, 10 patients had tumors categorized as HER2 overexpressing by the HercepTest and HER2 normal by the CB11 antibody. Testing with less specific antibodies can lead to the treatment of patients who may be unlikely to derive benefit from trastuzumab, whereas testing with less sensitive antibodies can lead to missed opportunities to intervene therapeutically with this agent. The optimal test to select patients for trastuzumab-based therapies will have high predictive value (sensitivity and specificity); we are prospectively addressing this issue via IHC (Lyndsay Harris) and fluorescent in situ hybridization techniques (Lynn Dressler) in parallel with CALGB 9840.

There is currently much discussion and investigation concerning the HER2 assay method that will be most predictive for response.12,13,19,42 We found that each of the four IHC assays and the FISH assay was significantly predictive for response to therapy. However, because this clinical trial administered trastuzumab in combination with paclitaxel, this study cannot be considered a pure test of HER2 expression and response to either trastuzumab or paclitaxel. Also, it is important to note that all of the IHC assays we evaluated had some degree of discordance with FISH results. Continued efforts to define and standardize a testing method that will provide the most reliable information for prognostic and treatment purposes are warranted.

The weekly administration of both paclitaxel and trastuzumab offers the potential to further improve on the clinical benefit observed with tri-weekly paclitaxel administration with trastuzumab. At the same time that this possibility is presently being prospectively explored in CALGB 9840 (Fig 5), the evaluation of this active doublet is ongoing in the adjuvant setting in a large randomized intergroup adjuvant trial (NCCTG 9831) for patients with axillary lymph node–positive, HER2-overexpressing breast cancer. In parallel with these and other clinical trials of trastuzumab, clinicopathologic correlative science should define the optimal method of HER2 assessment.