Kaplan-Meier recurrence-free survival curves for immunostaining.
Kaplan-Meier recurrence-free survival curves for fluorescence in situ hybridization.
Kaplan-Meier recurrence-free survival curves for all patients.
Kaplan-Meier recurrence-free survival curves for local recurrence.
Kaplan-Meier recurrence-free survival curves for distant recurrence.
Carr JA, Havstad S, Zarbo RJ, Divine G, Mackowiak P, Velanovich V. The Association of HER-2/neu Amplification With Breast Cancer Recurrence. Arch Surg. 2000;135(12):1469-1474. doi:10.1001/archsurg.135.12.1469
Copyright 2000 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2000
Amplification of the HER-2/neu oncogene in 25% of breast cancers is associated with a shortened disease-free survival.
Retrospective analysis of a patient population referred to a tertiary care facility for HER-2/neu testing. The mean follow-up was 56 months.
Large, urban, tertiary care hospital.
From 1995 to 1999, a consecutive sample of 190 patients with breast cancer had tissue samples tested for overexpression of the cell surface oncoprotein by immunostaining (IM) or amplification of the HER-2/neu oncogene by fluorescence in situ hybridization or both. Forty-nine subjects were excluded because they had tissue samples tested at our institution but received their treatment elsewhere. All patients tested for HER-2/neu after diagnosis with breast cancer in 1999 (n = 47) were excluded from analysis because of short follow-up time. One patient was excluded who had in situ ductal carcinoma. The remaining 93 patients were analyzed.
Of 93 patients, 40 (43%) had gene amplification. Overall, patients with oncogene amplification had a shorter median disease-free interval (22 months) compared with controls (40 months) (P = .003). Analysis by the Cox regression model showed that the HER-2/neu status remained significantly associated with time to relapse even after adjusting for age and tumor grade (P = .002; adjusted relative risk, 2.4; 95% confidence interval, 1.4-4.4). No association was found between gene amplification and tumor grade (P = .98), estrogen/progesterone receptor status (P = .29 and P = .43, respectively), or lymph node status (P = .98). Seventy-two patients (77%) eventually had disease recurrence, with 18 (25%) of these recurring locally.
The HER-2/neu oncogene is an independent prognostic indicator of a subset of breast cancers that are at high risk of early recurrence, regardless of tumor grade, estrogen/progesterone receptor status, and lymph node status. Patients amplifying the HER-2/neu oncogene have a shorter disease-free survival than patients without the oncogene.
IN 1985, the HER-2/neu oncogene was isolated independently by 2 separate groups, which they named HER-21 and c-erbB-2.2 Further analysis revealed the 2 genes to be the same,3,4 and it was renamed HER-2/neu. The gene encodes a transmembrane tyrosine kinase, which has homology with, but is distinct from epidermal growth factor receptor. The exact function of this kinase is unknown, but based on its close homology with epidermal growth factor receptor, it may function as a cellular receptor for an undiscovered growth factor. Great interest is now focused on determining the prognostic significance of the presence of this oncogene in breast tumors.
Several preliminary reports have documented that breast cancers with HER-2/neu amplification tend to have earlier relapse and shorter overall survival.5- 7 Retrospective data analysis support the incorporation of doxorubicin as part of adjuvant therapy for patients with breast cancer with HER-2/neu amplification,8,9 but it is still unknown if such therapy is beneficial. Moreover, it is unknown if closer surveillance and more frequent clinical examinations will improve survival.
Since HER-2/neu amplification is present in 25% to 40% of breast cancers,10- 12 the potential impact of this tumor marker as a prognostic tool would be profound if it can be shown to have a major effect on the clinical course of the tumor. To better understand its clinical role, a retrospective study of a consecutive sample population was performed at a single, large, tertiary care referral center to determine if there was a significant difference in the disease-free interval or recurrence rate between patients positive and negative for HER-2/neu.
From 1995 to 1999, 190 patients with breast cancer had tissue from the original surgical specimen tested for amplification of the HER-2/neu oncogene or overexpression of the gene product. Samples were directed for testing by the Breast Tumor Board Consensus at the Henry Ford Hospital, Detroit, Mich, or oncologists outside the institution for immunostaining (IM) to allow determination of protein overexpression and possible therapeutic use of trastuzumab (Herceptin; Genentech, San Francisco, Calif). Samples were also tested by fluorescence in situ hybridization (FISH) to gain prognostic information. Most patients had tissue samples initially tested for the HER-2/neu gene product by IM, and if results were strongly positive, no further testing was performed. If the initial staining was indeterminate, the sample was tested by FISH for further characterization. Fluorescence in situ hybridization was the only technique used in 31 cases during the institution's early experience. All samples were then interpreted by one surgical pathologist (R.J.Z.).
From the 190 original patients, 49 were excluded because they had tissue samples tested at our institution but received treatment elsewhere. All patients tested for HER-2/neu after diagnosis with breast cancer in 1999 (n = 47) were excluded from analysis because of short follow-up time. One patient was excluded who had only ductal carcinoma in situ. This left 93 patients who met study criteria. Patients were found to meet study criteria if they received their care at our institution and data were available to assess known prognostic variables (tumor size, grade, lymph node, and estrogen/progesterone receptor [ER/PR] status). Forty patients (43%) were determined to have amplification of the oncogene or overexpression of the gene product, while 53 patients (57%) did not. These 2 patient groups were similar with respect to age, race, histologic type of breast cancer, American Joint Committee on Cancer TNM stage, lymph node status, ER/PR status, tumor grade, and initial therapeutic modalities (Table 1). Only patients with invasive carcinoma were included. Since chemotherapeutic regimens are standardized at our institution, the initial chemotherapy given to all patients had minimal variation. The disease-free interval after initial therapy and the recurrence rate were determined for 83 of 93 patients (the exact length of remission could not be determined in 10 patients). The tumor size, lymph node status, ER/PR status, histologic type, and tumor grade were determined from the surgical resection specimen in 88 of 93 patients.
For determination of trastuzumab therapy candidates, IM was used on formalin-fixed, paraffin-embedded tissue samples to detect the HER-2/neu gene product using an immunohistochemical kit approved by the Food and Drug Administration, Washington, DC (DAKO Hercep Test; DAKO, Carpinteria, Calif). Findings were considered to be positive with scores of 2+ to 3+. Indeterminate samples or samples with a score of 1+ then underwent FISH for further gene characterization (INFORM HER-2/neu Gene Detection System; Oncor, Washington, DC). Because of the strong association of oncogene amplification with protein overexpression, the second test (FISH) was used to confirm the indeterminate or weakly positive cases. All samples were carefully scored as either amplifying the oncogene or without amplification (single copy) of the oncogene. Amplification greater than 4 copies by FISH was considered positive, although most of the samples testing positive expressed more than 8 copies.
Two databases were compiled based on the 2 different types of testing for determination of HER-2/neu status. The IM and FISH data sets were analyzed separately; but because of the strong association between the 2 methods,13- 15 the disease-free interval and recurrence rates were also determined from the combined data.
The χ2 and logistic regression analyses were performed to compare HER-2/neu amplification with tumor size, ER/PR status, lymph node status, and tumor grade. Overexpression of the gene product was related to disease-free survival period and recurrence using analysis by the Cox regression method and χ2 testing, respectively. Further analysis was accomplished comparing increasing number of copies of the HER-2/neu gene to decreased disease-free interval to determine if there was a linear relationship. Kaplan-Meier disease-free survival period estimates were plotted from overall recurrence data, as well as distant and local recurrence data. P≤.05 was considered significant.
Of 93 patients, 86 had invasive ductal carcinoma, and 7 had invasive lobular carcinoma. The mean follow-up period for all patients was 56 months. Eighty-eight patients had undergone either modified radical mastectomy or lumpectomy with axillary lymph node dissection. Five patients who had distant metastases (4 to bone, 1 to liver) at presentation never underwent a surgical resection. Only 1 of these 5 was found to amplify the oncogene.
Seventy-nine of 93 patients (85%) received adjuvant chemotherapy after surgery. Ten patients were given tamoxifen alone (6 negative for HER-2/neu and 4 positive) and 69 were given systemic chemotherapy (39 negative for HER-2/neu and 30 positive), consisting of either doxorubicin and cyclophosphamide or cyclophosphamide, methotrexate, and 5-fluorouracil. Thirty-eight patients received doxorubicin-based regimens, 22 of whom were negative for HER-2/neu, and 16 of whom were positive. Five patients had inflammatory breast cancer at initial examination (all positive for HER-2/neu) and underwent neoadjuvant therapy prior to resection, which prevented exact tumor size determination. Thirty-eight patients were treated with trastuzumab (Herceptin) at some point during their adjuvant therapy. All displayed 3+ IM except 7: 2 had only 2+ IM, and 5 were tested by FISH alone, all of whom had more than 10 copies of the oncogene. Of 38 patients, 14 (36%) had disease progression while on the drug.
Of the original 93 patients reviewed, 25 have died. The remaining 68 are still being followed. Twenty-one patients never developed a recurrence after initial therapy. Table 1 presents the results of the association of HER-2/neu with tumor size, receptor status, lymph node metastases, and tumor grade. Thirty subjects (32%) had T1 tumors; 37 (40%), T2; 10 (11%), T3; and 11 (12%), T4. Five (5%) of tumors were measured in the surgical specimen after neoadjuvant therapy and were classified as TX. Of 40 tumors that overexpressed the gene product and/or amplified the HER-2/neu oncogene, 11 (37%) were classified as T1; 17 (46%), T2; 2 (20%), T3; and 9 (82%), T4. Logistic regression analysis of tumor size and amplification reached statistical significance (P = .05). In particular, a T4 tumor had an increased odds ratio of being positive for HER-2/neu (Table 1). A test for trend did not find a statistically significant association of tumor size with HER-2/neu status (P = .08). This seems to be due to the unexpected small percentage of HER-2/neu positivity at the T3 tumor size, which interrupts the increasing pattern from 1 to 4.
All patients were tested for ER/PR status by immunohistochemical analysis. Forty-nine patients (53%) were ER-positive. Nineteen ER-positive patients (39%) had HER-2/neu amplification, as did 20 (50%) of ER-negative patients. Forty-seven (51%) of all patients were PR positive. Nineteen (40%) of PR-positive patients had HER-2/neu amplification, as did 20 (49%) of PR-negative patients. There was no statistical association between ER/PR status and HER-2/neu expression (P = .29 and P = .43, respectively).
A total of 58 (62%) of 93 patients had metastatic disease to the ipsilateral axillary lymph nodes at the time of surgery, without evidence of distant metastatic disease (by computed tomographic scan or magnetic resonance imaging of the brain, chest, and abdomen, and bone scan). HER-2/neu amplification was found in 25 (43%) of 58 lymph node–positive patients and 13 (43%) of 30 lymph node–negative patients (P = .98). The 5 patients treated with neoadjuvant therapy were classified as NX.
Seventy-four of the tumors were nuclear grade 2 (intermediate grade) or grade 3 (high grade), while 18 were not graded. HER-2/neu amplification did not vary with grade (G2, 11 [50%] of 22; G3, 25 [48%] of 52; P = .88).
Immunostaining and FISH were both used for determination of the patients' HER-2/neu status, as previously discussed. However, we sought to determine the prognostic value of each test independently. Using the IM technique alone to assess protein overexpression, the sample size is reduced to 36 patients staining positive and 20 staining negative. To best compare time to relapse, we used survival analysis. A significant difference was not found by the log-rank test (P = .26). The median time to relapse was 27 months in the HER-2/neu–positive group and 38 months in the negative group (Figure 1). The percent relapse in the HER-2/neu–positive group was 78% (28 patients) compared with 88% (18 patients) in the HER-2/neu–negative group. Using Cox regression, the unadjusted risk ratio (RR) estimate of HER-2/neu status to time to relapse was 1.5 (95% confidence interval [CI], 0.8-2.8).
Using the FISH data for gene amplification alone, the sample size is reduced to 17 patients in the positive group and 35 in the negative group. To best compare time to relapse, we again used survival analysis. A significant difference was found by the log-rank test (P = .03). The median time to relapse was 29 months in the HER-2/neu–positive group and 41 months in the negative group (Figure 2). The percent relapse in the HER-2/neu–positive group was 63% (11 patients) compared with 70% (25 patients) in the negative group. Using Cox regression, the unadjusted RR estimate of HER-2/neu status to time to relapse was 2.4 (95% CI, 1.1-5.5).
Using both HER-2/neu determination methods to enhance the accuracy, the average disease-free interval for those patients amplifying the HER-2/neu oncogene was 22 months, whereas, for the patients without amplification or overexpression, it was 40 months (Figure 3). This achieved statistical significance (P = .003) using the log-rank test. Using Cox regression analysis, the HER-2/neu status remained significantly associated with time to relapse even after adjusting for age and tumor grade (P = .002; adjusted RR, 2.4; 95% CI, 1.4-4.4). Also using Cox regression, the unadjusted RR estimate of HER-2/neu status to time to relapse was 2.2 (95% CI, 1.3-3.7). Of 72 patients who had disease recurrence, 18 (25%) recurred locally. Results of χ2 analysis showed that those patients who had amplification of HER-2/neu did not have a statistically significant increased incidence of local recurrence (50% [20 patients]) compared with patients without amplification (41% [22 patients]) (P = .49). Kaplan-Meier disease-free survival period estimates are shown for local and distant recurrence in Figure 4 and Figure 5.
The method and reproducibility of IM for the HER-2/neu gene product has been proven accurate in our laboratory and others,10 and is a US Food and Drug Administration–approved immunohistochemical assay to determine women likely to respond to treatment with trastuzumab (Herceptin). Prior work has shown that paraffin-embedded archival tissues saved for 2 to 5 years do not suffer from deterioration of antigenicity over time.16 Fluorescence in situ hybridization has also been proven to be an accurate and reproducible method for assessing HER-2/neu gene amplification in human breast cancer.6 Immunostaining for the HER-2/neu gene product and FISH to detect amplification of the oncogene have demonstrated excellent correlation in several studies.13- 15 Gene amplification detected by FISH is 92% concordant with immunohistochemical detection of overexpression of the oncoprotein.14 This allows the use of both tests to determine the presence of HER-2/neu in tumors from women with breast cancer.
It is interesting to examine the subset analysis detailing the prognostic value of IM alone and HER-2/neu status. The Kaplan-Meier curves in Figure 1 show that those patients who are positive for HER-2/neu do display a worse trend toward earlier recurrence, although this did not reach statistical significance. We attribute this to the small sample size in this subset population. Immunostaining combined with FISH as well as FISH alone both showed a statistically significant decreased disease-free interval and earlier recurrence for patients positive for HER-2/neu.
Previous studies have shown that HER-2/neu overexpression is significantly associated with an increased risk of earlier relapse and death.7,17- 22 Other studies have not demonstrated this relationship.23- 26 There is no consensus in the literature concerning the clinical significance of the amplification of this oncogene. Our study supports the finding of an increased risk of early recurrence (shortened disease-free interval) for the HER-2/neu positive group; however, since most patients in this series were still alive at the conclusion of the study, an increased risk of early death could not be confirmed. In addition, the HER-2/neu positive group's statistically significant overall increased risk of recurrence was independent of tumor size, ER/PR status, and tumor grade. Most local recurrences occurred within the ipsilateral breast after lumpectomy (22% [4 recurrences]), or on the chest wall in proximity to the previous mastectomy scar (67% [12 recurrences]). All patients who underwent lumpectomy had negative findings confirmed in the margins; thus, resection was adequate. Additionally, they received postoperative radiation therapy to the tumor bed. Several other patients had enlarged supraclavicular lymph nodes (distant metastases) at initial examination. Most of these were detected on physical examination by the patient's surgeon at follow-up visits or by subsequent mammography. An increased incidence of recurrence may warrant closer follow-up with surveillance mammography after lumpectomy and frequent clinical examination to detect chest wall nodules.
The fact that these HER-2/neu– amplified tumors recur earlier and behave more aggressively has been used by some investigators to propose that high-dose chemotherapy be given to patients positive for HER-2/neu with uninvolved lymph nodes.17,27 Others have disputed this claim.16 Unfortunately, the present study was too small to assess the significance of node-negative patients who received chemotherapy. Thirty-eight patients in our series were treated with trastuzumab (Herceptin). However, the number of patients was too small to determine any therapeutic benefit of this monoclonal antibody compared with conventional chemotherapy. Despite the use of trastuzumab, which was only given to patients positive for HER-2/neu, their overall disease-free survival period was still significantly shorter. Phase II clinical trials of intravenous trastuzumab therapy have shown an overall response rate of 11.6% with monotherapy28 and 24.3% in combination with cisplatin.29 It seems that the current evidence would argue for the need for close follow-up of these patients with clinical examination and mammography to detect recurrence and allow the earliest possible intervention.
Prior investigations have also differed on whether there exists a significant association between the HER-2/neu status and tumor size, tumor grade, and ER/PR status. HER-2/neu amplification has been associated with low ER/PR expression,18,30,31 high tumor grade,18,32,33 and increased metastatic potential.30 However, most studies have found no statistically significant association between HER-2/neu amplification and tumor size, ER/PR content, and tumor grade.5,6,34,35 Our study supports the latter findings with the exception of an apparent association between increasing tumor size and gene amplification reaching statistical significance (P = .05). However, this finding could represent a type II statistical error due to sample size, or it may indicate an association. Those patients overexpressing the gene product had a shorter disease-free interval (22 vs 40 months) after an average follow-up of 56 months. However, an increasing number of copies of the oncogene by FISH did not decrease the disease-free interval in a linear fashion. Association was found neither between tumor grade and amplification nor ER/PR status and amplification. Lymph node status was not found to be related to HER-2/neu overexpression. This is consistent with our previous work which showed that only tumor size predicted the presence of lymph node metastases.36 This lack of an association between other known prognostic factors and the HER-2/neu oncogene favors its acceptance as an independent predictor of disease recurrence.
The HER-2/neu oncogene is an independent prognostic indicator of a subset of breast cancers that are at high risk of recurrence, regardless of tumor size, grade, and lymph node status. Patients amplifying the HER-2/neu oncogene have a shorter disease-free survival period than patients without the oncogene. These results warrant close follow-up with clinical breast and chest wall examination in this group of patients.
Presented at the 47th Annual Meeting of the Michigan Chapter of the American College of Surgeons, Acme, Mich, May 4-6, 2000.
Corresponding author and reprints: John Alfred Carr, MD, Department of General Surgery K-8, Henry Ford Hospital, 2799 W Grand Blvd, Detroit, MI 48202-2689.