Correlation between HER-2 expression and response to neoadjuvant chemotherapy with 5-fluorouracil, doxorubicin, and cyclophosphamide in patients with breast carcinoma




The objective of this study was to determine whether HER-2 overexpression is associated with improved response to neoadjuvant chemotherapy with 5-fluorouracil, doxorubicin, cyclophosphamide (FAC) in patients with breast carcinoma.


Ninety-seven patients with Stage I–III breast carcinoma were included. HER-2 expression was determined by routine clinical laboratory assessment, and tumors with 3 + immunohistochemistry staining intensity or gene amplification by fluorescent in situ hybridization were considered HER-2 positive. Response was assessed by physical examination, imaging assessment, and pathologic assessment at the time of surgery.


The median patient age was 45 years. At baseline, 68% of patients had lymph node positive disease, 87% had ≥ T2 tumors, and 28% of patients had HER-2 positive tumors. Eighty-four percent of patients received four courses of FAC, 8% of patients received 3 courses of FAC, and the remaining 8% of patients received 5–6 courses of FAC. The clinical response rate (complete response [CR] and partial response [PR]) was 78%, the imaging response rate (CR and PR) was 64%, and 15% of patients had a good pathologic response, defined as a CR or minimal residual disease (tumor measuring < 1 cm in greatest dimension and negative lymph nodes). Concordance between the three methods of response assessment (clinical, pathologic, and imaging) was modest and was best between clinical assessment and imaging assessment (64% concordance). HER-2 status did not correlate with pathologic or clinical response (assessed by physical examination or imaging), although a nonsignificant trend was noted toward better response in patients with breast tumors that overexpressed HER-2.


The authors found no significant correlation between HER-2 expression and clinical or pathologic response to neoadjuvant chemotherapy in patients with breast carcinoma. Cancer 2003;97:1758–65. © 2003 American Cancer Society.

DOI 10.1002/cncr.11245

The HER-2 gene encodes for a transmembrane tyrosine kinase growth factor receptor that is overexpressed in approximately 30% of breast carcinomas. HER-2 gene overexpression, as detected by immunohistochemistry (IHC) or fluorescent in situ hybridization (FISH), may influence the sensitivity of breast carcinoma to chemotherapy.1, 2 The effect of HER-2 overexpression on a patient's response to treatment may depend on the type of cytotoxic agent used. It has been proposed that patients who have tumors that overexpress HER-2 gain less benefit from adjuvant chemotherapy with cyclophosphamide, methotrexate, and 5-fluorouracil (5-FU) (CMF) compared with patients who have tumors with normal HER-2 status.3–5 Clinical observations also suggest that breast tumors that have HER-2 overexpression may be less sensitive to taxane therapy compared with tumors that have normal HER-2 status.6–8 Other data indicate that HER-2-overexpressing tumors may be particularly sensitive to anthracycline-based adjuvant therapy. This was reported first after retrospective analysis of the results of the Cancer and Leukemia Group B B8541 study, which investigated the role of dose intensity of doxorubicin in the cyclophosphamide, doxorubicin, and 5-FU (CAF) regimen. Those investigators concluded that patients with HER-2-overexpressing tumors benefited, in terms of improved disease-free survival (DFS) and overall survival, from higher doses of doxorubicin (60 mg/m2) compared with lower doses (30–40 mg/m2); they reported no similar dose-outcome correlation for patients who had tumors with normal HER-2 status.9, 10 Similar retrospective analyses from two other randomized studies—National Surgical Adjuvant Breast and Bowel Project (NSABP) Study B-11 (L-phenylalanine mustard, doxorubicin, and 5-FU vs. L-phenylalanine mustard and 5-FU) and South West Oncology Group Study 8814 (tamoxifen vs. tamoxifen plus CAF)—also indicated that the addition of an anthracycline to adjuvant therapy improved DFS only in patients with positive HER-2 status.11, 12 However, several other studies failed to reveal a statistically significant correlation between HER-2 overexpression and improved response to anthracycline therapy.13, 14 Retrospective evaluation of the HER-2 status of patients who participated in NSABP Study B-15 failed to show a statistically significantly better outcome with doxorubicin-based therapy in patients who had HER-2-overexpressing tumors compared with patients who had tumors with normal HER-2 status, although a trend toward improved DFS was observed.15 Similarly, a nonsignificant trend toward a better outcome in patients with HER-2 positive tumors who were treated with epirubicin was reported by Di Leo and his colleagues, who compared epirubicin plus cyclophosphamide with CMF adjuvant treatment.16 Preoperative/neoadjuvant chemotherapy for patients with breast carcinoma offers a unique opportunity to evaluate molecular predictors of response to treatment. The presence or absence of a marker can be correlated with response of the primary tumor and lymph nodes. Pathologic complete response (pCR) and, to a lesser extent, clinical complete response (CR) are associated strongly with prolonged long-term survival, independent of initial tumor stage and prognostic markers.17, 18 The importance of partial response (PR), as measured by physical examination or by mammogram, with regard to long-term outcome is less certain.19, 20 In the current retrospective study, we examined whether HER-2 overexpression, according to routine clinical diagnostic laboratory assessment, was correlated with the rate of objective tumor response to preoperative 5-FU, doxorubicin, cyclophosphamide (FAC) chemotherapy for patients with breast carcinoma.


The Department of Breast Medical Oncology of the University of Texas M. D. Anderson Cancer Center (MDACC) maintains a prospective electronic clinical data base of patients who have received treatment through the Department since 1997. This data base was searched to identify patients who received preoperative chemotherapy with the FAC regimen from July, 1997 through April, 2001 and also had routine HER-2 status assessment performed by a clinical laboratory. Patients who received anthracycline plus taxane-based preoperative chemotherapy were excluded form the search. HER-2 receptor status was determined by IHC using an anti-HER-2 monoclonal antibody (AB8; Neomarker, Fremont, CA) or by FISH using the PathVision kit (Vysis, Downers Grove, IL). Tumors with a gene copy ratio of HER-2 to chromosome 17 centrosome > 2.0 or with an IHC staining intensity of 3 + (strong membranous staining in at least 10% of cells) were considered HER-2-overexpressing tumors for this analysis. The diagnosis of invasive breast carcinoma was established in all patients by fine-needle aspiration of lymph nodes and/or core needle biopsy of the primary breast lesion. All HER-2 measurements were performed at or reviewed and confirmed by the Department of Pathology of MDACC. Chemotherapy was given as standard clinical care or on clinical trials. FAC chemotherapy consisted of 5-FU (500 mg/m2 intravenously on Days 1 and 4), doxorubicin (50 mg/m2 in a 72-hour, continuous infusion through Days 1–3), and cyclophosphamide (500 mg/m2 intravenously on Day 1), with treatments repeated every 21 days. Physical examination was performed at baseline and was repeated before each treatment cycle to determine clinical response. Imaging of the primary breast lesion was performed with mammography and/or ultrasonogram at baseline and before surgery to determine imaging response. Clinical and imaging responses were categorized by the International Union Against Cancer criteria. A CR was defined as the complete disappearance of primary tumor and lymph node metastasis. A PR was defined as tumor reduction > 50%, and stable disease (SD) was defined as tumor reduction < 50% or tumor size increase < 25%. Progressive disease (PD) was defined as an increase > 25% in tumor size. If a patient had both a mammogram and an ultrasonogram performed, to qualify for CR, both the mammogram and the sonogram had to show the complete resolution of disease in all sites. After completion of preoperative chemotherapy, patients underwent either modified radical mastectomy or breast-conserving surgery, as determined appropriate by the surgeon. A pCR was defined as the disappearance of all invasive disease from the breast and lymph nodes. Minimal residual disease (MRD) was defined as invasive tumor measuring < 1 cm in the breast and negative lymph nodes. Invasive tumor measuring ≥ 1 cm in the breast or any positive lymph node in the axilla, regardless of the size of residual disease in the breast, was considered extensive residual disease (ERD). After patients were identified through the electronic data base, charts were reviewed by two authors for accuracy and to obtain clinical updates (F.Z. and L.P.). This retrospective chart review was approved by the Institutional Review Board of MDACC.

Statistical Analysis

This was a retrospective study, and sample size was determined by availability of data on patients treated between 1997 and 2001. We identified 97 patients who met the inclusion criteria, and 28% had HER-2 overexpression. It is estimated that a study of this size has a 50% power to show 20% difference in response rates between patients with normal HER-2 status and patients with amplified HER-2 status. Categorical data were compared using Fisher exact tests and chi-square tests. P values < 0.05 were considered statistically significant: All P values were two-tailed and were adjusted for multiple comparisons when indicated. Due to the limited number of events and patients in each clinical category, univariate analysis was used only to estimate the effects of clinical TNM stage, nuclear grade, hormone receptor status, and HER receptor status on response to therapy. The 95% confidence interval (95% CI) of the risk ratios for good response in patients who had HER-2-overexpressing tumors compared with patients who had tumors with normal HER-2 status were calculated by the asymptotic method of Katz et al.21 DFS was plotted by the Kaplan–Meier method.


Patient Characteristics

Ninety-seven patients met the search criteria of available results from routine HER-2 receptor assessment and treatment with preoperative FAC chemotherapy. During the same period, 152 additional patients received preoperative FAC treatment, but their HER-2 status was not determined. HER-2 assessment became part of routine pathologic evaluation only after trastuzumab was approved for the treatment of patients with metastatic breast carcinoma in 1998. Clinical characteristics of the patients who were included in the study are presented in Table 1. At the time of diagnosis, 87% of patients (n = 84 women) had ≥ T2 tumors, and 68% of patients (n = 66 women) had clinically positive lymph nodes. Seventy-five tumors were studied with IHC to determine HER-2 status (22 tumors had 3 + IHC staining, 9 tumors had 2 + IHC staining, 10 tumors had 1 + IHC staining, and 34 tumors had 0 IHC staining), and 48 tumors were analyzed with FISH (13 tumors showed amplification). Twenty-eight percent of tumors (n = 28 tumors) had HER-2 overexpression, defined as either 3 + IHC staining or gene amplification by FISH. Eighty-one patients (84%) received 4 cycles of chemotherapy, 8 patients (8%) received 3 cycles of chemotherapy, 7 patients (7%) received 6 cycles of chemotherapy, and 1 patient received 5 cycles of chemotherapy before surgery. The length of preoperative treatment was at the discretion of the treating physicians. Institutional treatment pathways call for four courses of preoperative chemotherapy. For patients who participated in clinical trials, treatment was determined by the study protocol. After completion of neoadjuvant therapy, 61 patients (63%) underwent modified radical mastectomy, and the remaining patients underwent breast-conserving surgery. Postoperatively, 78 patients received additional chemotherapy, 44 patients received 4–6 cycles of paclitaxel, 26 patients received 2–4 cycles of FAC, and 8 patients received 4 courses of docetaxel. All patients with positive estrogen receptor status received 5 years of tamoxifen, and radiation also was delivered after all breast-conserving surgery, or if the primary tumor measured ≥ 5 cm in greatest dimension, or if > 4 lymph nodes were involved with disease in the axilla.

Table 1. Patient Characteristics at Diagnosis before Preoperative Therapy
CategoryNo. of patients (%)
  • a

    HER-2 positive was defined as either 3 + staining by immunohistochemistry or gene amplification by fluorescence in situ hybridization.

No. of patients97
Age (yrs) 
 ≥ 5043 (44)
 < 5054 (56)
Tumor size 
 T113 (13)
 T251 (53)
 ≥ T333 (34)
Lymph node status 
 Negative32 (33)
 Positive65 (67)
Nuclear grading 
 15 (5)
 233 (34)
 359 (61)
 Invasive ductal carcinoma89 (92)
 Other histology8 (8)
 Estrogen receptor positive63 (65)
 Progesterone receptor positive54 (56)
 HER-2 positivea28 (28)

Tumor Response

Tumor response was assessed by both physical examination and imaging with ultrasonogram or mammogram. All 97 patients had pretreatment and posttreatment physical examinations to determine best clinical response. Eighty patients (83%) experienced objective clinical responses, including 6 CRs (6%) and 74 PRs (76%). Two patients had PD (2%). Eighty-eight patients were evaluable for imaging response, and the remaining 9 patients did not have corresponding pretreatment and posttreatment imaging studies. Fifty-six patients (64%) demonstrated tumor response assessed by imaging studies, including 12 CRs (14%) and 44 PRs (50%), and 3 patients (4%) had PD. All patients were evaluated for pathologic response. Sixteen patients had no residual disease in the breast at the time of surgery; however, 11 of those 16 patients had at least 1 involved axillary lymph node and, thus, were not considered to have achieved a pCR for the purpose of this study. Five patients (4%) achieved a pCR, and 9 patients had MRD (9%).

Correspondence between the three methods of measuring tumor response was modest. Clinical response was concordant with imaging response (64%), but neither clinical response nor imaging response correlated well with overall pathologic response (Table 2). Correlation was modest across all response categories, even among patients who achieved a CR. For example, of the five patients who achieved a pCR, only one patient had CR on imaging studies. The other four patients had residual architectural abnormalities. The modest correlation was due at least in part to the manner in which pathologic response was defined in this study. ERD included tumors that may have been as small as 1 cm on pathologic examination, and many patients as those small tumors may have been categorized with a PR according to imaging studies or physical examinations. Assigning pathologic response categories other than PCR after neoadjuvant therapy is arbitrary, because the baseline pathologic size, by definition, is unknown. However, it is accepted generally that, the larger the residual disease, the worse the prognosis, which is the reason why we dichotomized residual disease into MRD (≤ 1 cm) and ERD (> 1 cm).

Table 2. Concordance between Three Types of Tumor Response Assessments
AssessmentType of assessment: No. of patients (%)Concordance rate (%)
  1. pCR: pathologic complete response; MRD: minimal residual disease; ERD: extensive residual disease; PR; partial response; CR: complete response.

 PR + CR14 (14)66 (68)
 Other0 (0)17 (18)32
 PR + CR10 (11)46 (53)
 Other4 (4)28 (32)43
 PR + CR48 (55)24 (27)
 Other8 (9)8 (9)64

Correlation between Objective Tumor Response and HER-2 Status

To correlate clinical, imaging, and pathologic response with HER-2 status, patients were divided into groups. Patients with good clinical and imaging response, including patients who achieved a CR or PR, were considered one group for analysis; whereas patients with SD or PD formed another group. Similarly, patients with a pCR or with MRD were analyzed together as good responders. It has been shown previously that patients who achieved a pCR or who have residual invasive disease measuring < 1 cm have a very good prognosis compared with patients who have ERD after neoadjuvant chemotherapy.17

There was a week trend toward better response to therapy in patients who had tumors with HER-2 overexpression, but this did not reach statistical significance for pathologic or clinical response (Table 3). A good pathologic response (pCR and MRD) was seen in 18% of patients who had tumors with HER-2 overexpression compared with 13% of patients who had tumors with negative HER-2 status (risk ratio, 1.3; P = 0.53). A good clinical response assessed by physical examination (CR and PR) was seen in 93% of patients who had tumors with HER-2 overexpression compared with 78% of patients who had tumors with normal HER-2 status (risk ratio, 1.2; P = 0.14). When response was assessed by imaging, 80% of patients who had tumors with HER-2 overexpression had an objective imaging response (CR or PR) compared with 57% of patients who had tumors with normal HER-2 status (risk ratio, 1.4; P = 0.05, Fischer exact test; P = 0.25, Bonferroni adjusted). This finding was not regarded as statistically significant because of the multiplicity of endpoints tested.

Table 3. Correlation between Tumor Response Assessed by Various Methods and HER-2 Expression
ResponseNo. of patients (%)P valueaRR95%CI
Total patientsHER-2 status
  • RR: risk ratio; 95% CI: 95% confidence interval; CR: complete response; PR: partial response; MRD: minimal residual disease (<1 cm residual invasive carcinoma in the breast and no lymph node involvement); PD: progressive disease; SD: stable disease; pCR: pathologic complete response;; ERD: extensive residual disease (> 1 cm invasive carcinoma in the breast or any positive lymph node after neoadjuvant therapy regardless of the size of residual disease in the breast).

  • a

    Fisher exact test.

 CR + PR8054 (78)26 (93)
 MRD, PD, + SD1715 (22)2 (7)–1.39
 pCR53 (4)2 (7)
 MRD96 (9)3 (11)
 ERD8360 (87)23 (82)0.701.7
 pCR + MRD149 (13)5 (18)
 ERD8360 (87)23 (82)0.531.40.54–3.67
 CR + PR5636 (57)20 (80)
 PD + SD3227 (43)5 (20)–1.86
Pathologic response in the breast only      
 pCR1611 (16)5 (18)
 No pCR8057 (84)23 (82)0.991.1

Correlations between HER-2 expression and clinical characteristics were also examined (data not shown). A significant correlation (P = 0.02) between nuclear grade and HER-2 overexpression was observed. There was no correlation between HER-2 expression and hormone receptor status, lymph node involvement, or tumor size. No significant association was found between good pathologic response and hormone receptor status, tumor size, or nuclear grade in univariate analysis (Table 4). Thirty-six patients underwent breast-conserving surgery. The rate of breast-conservation surgery was the same: 37% in both groups of those patients who had tumors with HER-2 overexpression and these who had tumors with normal HER-2 status.

Table 4. Pathologic Good Response Rate by Patient Characteristics
CharacteristicsNo. of patientspCR + MRDP valuea
  • pCR: pathologic complete response: MRD: minimal residual disease; ER: estrogen receptor.

  • a

    Fisher exact test.

HER-2 status    
ER status    
Nuclear grade    
Tumor size    

Survival Analysis

Twenty-four of 97 patients (25%) experienced disease recurrence, and 8 patients died of metastatic disease at a median follow-up of 33 months. Eight patients were lost to follow-up after they completed locoregional therapy. The median DFS was 48 months for the entire group. Kaplan–Meier survival analysis showed no difference in DFS between patients who had tumors with HER-2 overexpression and patients who had tumors with normal HER-2 status (Fig. 1). The two groups were well balanced for conventional prognostic markers for DFS. Sixty-six percent of patients who had tumors with normal HER-2 status had positive lymph nodes, and 68% of patients who had tumors with HER-2 overexpression had positive lymph nodes. The importance of a pCR was apparent even with this small sample size. None of the 5 patients who achieved a pCR experienced disease recurrence, whereas 22% of patients with MRD and 26% of patients with ERD developed recurrent disease.

Figure 1.

The graph shows the median disease-free survival (DFS) at 33 months of 97 patients with breast carcinoma according to HER-2 expression. The median DFS was 48 months, and there was no apparent difference in survival between patients with HER-2 overexpression and patients with normal HER-2 status who received neoadjuvant chemotherapy with 5-fluorouracil, doxorubicin, and cyclophosphamide.


In this study of 97 patients, we examined the interaction between HER-2 overexpression and response to a standard-dose, anthracycline-based, preoperative chemotherapy regimen for breast carcinoma. HER-2 overexpression was defined by current clinical standards as 3 + IHC staining or gene amplification by FISH. All pathology results were obtained or confirmed by the Department of Pathology at MDACC. We observed a weak trend toward a better objective tumor response to preoperative FAC chemotherapy in patients who had tumors with HER-2 overexpression. Risk ratios ranged from 1.1 to 1.4, depending on the type of response assessment. The association between a better response and HER-2 overexpression was not statistically significant. These findings are consistent with several other smaller studies that evaluated response to neoadjuvant, anthracycline-based therapy as a function of HER-2 status.

A recent report by Petit et al.22 evaluated 64 patients with available HER-2 status results who received neoadjuvant, epirubicin-based chemotherapy (5-FU, epirubicin, and cyclophosphamide [FEC]) at two epirubicin dose levels of 50 mg/m2 or 100 mg/m2. In that study, a significantly better clinical response and a nonsignificant trend toward a better pCR were noted in the group with HER-2-overexpressing tumors who were treated with the higher dose of epirubicin compared with the group who had tumors with normal HER-2 status. A smaller study of 35 patients reported a significantly increased pathologic response rate in patients who had tumors with amplified HER-2 expression compared with patients who had tumors with normal HER-2 status. In that study, the same correlation could not be detected when HER-2 was measured by IHC.23 In contrast, another study in which patients received mitoxantrone and methotrexate with or without mitomycin as preoperative therapy reported that a good clinical response was significantly less likely in patients with positive HER-2 status assessed by IHC.10 A fourth study that evaluated the response of 100 patients who were treated with neoadjuvant FEC (using an epirubicin dose of 100 mg/m2) failed to show any correlation between pathologic response and HER-2 overexpression.24 Similarly, Tulbah et al. also failed to detect any correlation between HER-2 overexpression assessed by HercepTest (Dako Corporation, Carpinteria, CA) and pathologic or clinical response to neoadjuvant chemotherapy.25 These data collectively suggest that HER-2 overexpression is not a robust or consistent predictor of objective tumor response to neoadjuvant therapy in patients with breast carcinoma. The current results also draw attention to the significant differences in the objective response rates when assessed by different methods: physical examination, imaging, and pathology. This variable may contribute to the conflicting results reported in the literature. Furthermore, the clinical importance of response to neoadjuvant therapy may also depend on the type of response assessment. There is robust evidence, based on large patient series with prolonged follow-up, indicating that pCR after neoadjuvant therapy predicts excellent long-term DFS regardless of initial tumor stage.10, 18, 26 The importance of clinical response measured by physical examination or imaging studies is less certain. The current study, due to the short follow-up and few events (25% of patients developed recurrent disease, and 8% of patients died at median follow-up of 33 months), failed to show any survival difference between the various response groups except for patients who achieved a pCR. None of the 5 patients who achieved a pCR experienced disease recurrence, whereas 22% of patients with MRD and 26% of patients with ERD developed recurrence of their diseases.

The biologic rationale explaining how HER-2 overexpression and increased signaling through this receptor may confer sensitivity to anthracycline therapy also remains enigmatic. Transfection of the HER-2 oncogene into breast carcinoma cell lines does not consistently change sensitivity to anthracyclines.27 Furthermore, inhibition of the HER-2 receptor and/or its downstream signaling pathways results in increased sensitivity to chemotherapy, including anthracyclines. This observation has been confirmed in the clinic using the anti-HER-2 humanized antibody, trastuzumab.8 Recently, it was suggested that HER-2 overexpression may be a surrogate for other molecular changes present in cells that influence response to therapy. This could explain the weak predictive power of HER-2 itself. Topoisomerase IIα, an important target of doxorubicin, occasionally is coamplified with HER-2 because of their proximity on chromosome 17.16, 28 HER-2 receptor activation also may increase intracellular topoisomerase II activity directly that, in turn, results in increased sensitivity to doxorubicin in vitro.29 It has been proposed that topoisomerase II amplification may be a more powerful predictor of response to preoperative anthracycline therapy than HER-2 status.23, 28 A recent study that was presented as an abstract did not support this suggestion.24 The low prevalence of topoisomerase II amplification in breast carcinoma makes it very difficult to assess reliably the contribution of this gene to sensitivity to anthracycline therapy. Topoisomerase II amplification has not been reported without HER-2 amplification; however, only a minority of HER-2-amplified tumors (30–40%) shows topoisomerase II amplification.29

It is important to remember that single-arm, neoadjuvant chemotherapy studies that utilize homogenous treatment do not address the question of optimal chemotherapy regimen selection for patients who have tumors with HER-2 overexpression. In particular, the current study did not address whether FAC is more or less efficacious compared with CMF or other nonanthracycline-containing, preoperative chemotherapy regimens for the treatment of patients who have tumors that overexpress HER-2. Based on our results, we can conclude that HER-2 expression is a weak, clinically nonsignificant predictor of objective tumor response to preoperative anthracycline therapy. Patients with normal HER-2 status also frequently benefit from anthracycline-based chemotherapy and should not be excluded from such treatment.