Dr. Esteva is supported by grant K23-CA82119 from the National Cancer Institute.
Assessment of histologic features and expression of biomarkers in predicting pathologic response to anthracycline-based neoadjuvant chemotherapy in patients with breast carcinoma
Article first published online: 12 JUN 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 12, pages 3107–3114, 15 June 2002
How to Cite
Wang, J., Buchholz, T. A., Middleton, L. P., Allred, D. C., Tucker, S. L., Kuerer, H. M., Esteva, F. J., Hortobagyi, G. N. and Sahin, A. A. (2002), Assessment of histologic features and expression of biomarkers in predicting pathologic response to anthracycline-based neoadjuvant chemotherapy in patients with breast carcinoma. Cancer, 94: 3107–3114. doi: 10.1002/cncr.10585
- Issue published online: 12 JUN 2002
- Article first published online: 12 JUN 2002
- Manuscript Accepted: 2 FEB 2002
- Manuscript Revised: 13 NOV 2001
- Manuscript Received: 24 JUL 2001
- Nellie B. Connally Breast Cancer Research Fund
- National Cancer Institute. Grant Number: K23-CA82119
- breast carcinoma;
- nuclear grade;
- mitotic activity;
- neoadjuvant chemotherapy
There is significant variability in the response of tumors to neoadjuvant chemotherapy, and the underlying mechanism for this variability is unknown. In this study, the authors investigated the roles of tumor nuclear grade, mitotic activity, and biomarker expression profiles in predicting the pathologic response of breast tumors to preoperative chemotherapy.
Eighty-two patients with breast carcinoma participated in two clinical trials and were treated with neoadjuvant chemotherapy, which consisted of either a conventional dose of fluorouracil, doxorubicin, and cyclophosphamide (FAC) or dose-escalated FAC. The mean age of the patients was 46 years (range, 24–69 years). Nuclear grade, mitotic activity, and biomarker profile (Her2-neu and mitosin expression patterns) in pretreatment tumors were correlated with the postchemotherapy pathologic response.
Twelve patients (15%) had a complete pathologic response (CPR), 23 patients (28%) had a near complete response (NCR), and 47 patients (57%) had significant residual disease present either at the primary site or in the axillary lymph nodes. The authors found that the nuclear grade and mitotic activity of pretreatment tumors were correlated significantly with CPR and NCR (P = 0.002 and P = 0.004). Mitosin also was correlated significantly with CPR and NCR (P = 0.028). A higher percentage of patients with Her2-neu-positive tumors had a CPR or an NCR (P = 0.152). CPR and NCR were not correlated significantly with disease stage (P = 0.186) or lymph node positivity (P = 0.498).
The current results indicate that tumor nuclear grade and tumor proliferative activity (mitotic activity and mitosin immunostaining) of pretreatment tumors in patients with breast carcinoma may serve as important indicators for the pathologic responsiveness of tumors to neoadjuvant, anthracycline-based chemotherapy. Cancer 2002;94:3107–14. © 2002 American Cancer Society.
Clinical studies have shown that the use of neoadjuvant chemotherapy for patients with locally advanced breast carcinoma can increase surgical resectability rates, and response to therapy is correlated with the patient's ultimate disease free survival.1–7 In addition, significant volume reduction in tumors after neoadjuvant chemotherapy may permit subsequent, successful breast-conserving surgical treatment.6, 8
There is significant variability in the histopathologic response of tumors to neoadjuvant chemotherapy, with approximately 15% of patients achieving a complete response, whereas, at the other end of the spectrum, 15% of patients display minimal change or progressive disease. Currently, the underlying mechanism for this variability is unknown. Contributing factors may include the diverse genetic background and hormonal environment of the tumor. Previous studies have focused on the correlation between the response of tumors to chemotherapy and various factors, such as histologic grade, DNA ploidy, cell kinetics, and estrogen receptor status of the primary tumor. However, those studies yielded inconsistent results.9–13
Currently, growing numbers of new and promising biomarkers of breast carcinoma are under evaluation for their role in predicting chemoresponsiveness to various forms of neoadjuvant chemotherapy regimens.14–20 Like previous studies, these studies have generated conflicting results regarding their positive or negative value in predicting chemoresponsiveness (e.g., the case of Her2-neu). Major contributing factors to these discordant results are various chemotherapy protocols, different dosage intensities, different patient selection criteria, and different methods of assessing of the biomarkers.
We studied primary tumor specimens from 82 patients with breast carcinoma to evaluate the role of tumor nuclear grade, mitotic activity, mitosin immunostaining (a new marker of proliferative activity), and Her2-neu overexpression as predictors of pathologic response to doxorubicin-based neoadjuvant chemotherapy. All patients were evaluated, treated with the same chemotherapeutic agents, and underwent surgery after the completion of neoadjuvant chemotherapy at The University of Texas M. D. Anderson Cancer Center. We correlated pretherapeutic tumor histologic features and biomarker profiles with the histopathologic response.
MATERIALS AND METHODS
We studied primary tumor specimens from 82 patients with Stage IIA, IIB, IIIA, IIIB, or IV tumors (with ipsilateral supraclavicular lymph node involvement only), as defined according to the 1998 American Joint Committee on Cancer Classification System.21 All patients underwent either core needle biopsy, excisional biopsy, or fine-needle aspiration biopsy of any involved lymph nodes prior to any treatment. Patients with primary inflammatory carcinoma were not included in the study. Patients were registered in two prospective trials of neoadjuvant regimens (Regimens A and B) of 5-fluorouracil, doxorubicin, and cyclophosphamide (FAC) at The M. D. Anderson Cancer Center between 1989 and 1996. Sixty patients were enrolled in Regimen A, which consisted of 500 mg/m2 5-fluorouracil, 50 mg/m2 doxorubicin, and 500 mg/m2 cyclophosphamide. The remaining 22 patients were enrolled in Regimen B, which consisted of 600 mg/m2 5-fluorouracil, 60 mg/m2 doxorubicin, and 1000 mg/m2 cyclophosphamide. A total of 378 patients were treated on these two trials; 244 patients in these trials underwent fine-needle aspiration biopsy only prior to neoadjuvant chemotherapy. Among the remaining 134 patients, we were able to obtain adequate tissue material from 82 patients. Clinical features and outcomes of these patients were not significantly different than the features and outcomes of the entire group. After four cycles of chemotherapy, patient and tumor responses were evaluated by a multidisciplinary team. Chemotherapy was followed by modified radical mastectomy in 58 of 82 patients, and the other 24 patients underwent segmental resection with axillary lymph node dissection.
Pathology slides and reports for both prechemotherapy biopsies and postchemotherapy resection specimens were reviewed for all 82 patients. The histologic grade of pretherapeutic tumors was assessed according to the modified Black nuclear grading system (Grade 1, well differentiated; Grade 2, moderately differentiated; and Grade 3, poorly differentiated).22 Tumors were typed by using the World Health Organization classification criteria. Mitotic activity was counted as combined mitotic figures per 10 high-power fields (HPF). The median mitotic count was 6.1 mitotic figures per 10 HPF, and the mitotic activity was graded as inactive (≤ 6 mitotic figures per 10 HPF) and active (> 6 mitotic figures per 10 HPF). Pathologic response was graded as a complete pathologic response (CPR), defined as the absence of residual tumor in breast and axillary lymph nodes; a near complete response (NCR), defined by the presence of residual primary tumor < 1 cm3; or as a partial pathologic response (PPR), defined by the presence of residual primary tumor measuring > 1 cm3.
Immunohistochemical studies for mitosin (clone 7F11.2; Gene Tex; catalog no. MS-MIT11-PX1) and Her2-neu (clone Tab250; Zymed, San Francisco, CA; catalog no. 28-0003) were performed at dilutions of 10 μg/mL and 0.375 μg/mL, respectively, in all 82 specimens. Briefly, 4-μm tissue sections were cut from the same blocks of hematoxylin and eosin-stained slides, mounted on charged slides, deparaffinized in xylene, and rehydrated in descending grades of ethyl alcohol (from 100% to 70%). The slides used for mitosin staining were then subjected to heat-induced antigen retrieval by immersion in a 0.01 mol/L citrate buffer, pH 6.0; preheated to > 90 °C; and heated in an electric vegetable steamer (Black and Decker, Shelton, CT) for 54 minutes. Her2-neu antibody was applied without antigen retrieval. The endogenous peroxidase activity was blocked by a 5-minute treatment with 3% hydrogen peroxide in absolute methanol. Immunohistochemical analysis was then performed using the avidin-biotin technique (LSAB2 peroxidase kit; DAKO, Carpinteria, CA). The antigen-antibody immunoreaction was visualized using 3-3′-diaminobenzedine as the chromogen, and the slides were counterstained with Mayer hematoxylin.
Immunoreactivity for mitosin was scored by the percentage of positive tumor-cell nuclei. The median mitosin value was 15%, and the proliferation rate was designated as low or high, corresponding to ≤ 15% or > 15% for mitosin positive tumor cell nuclei, respectively. Her2-neu expression was scored as the percentage of cells with membranous staining, and intensity of signal and was graded as negative (0–10% of cells with weak, incomplete membranous staining), weakly positive or borderline (10–30% of cells with weak membranous staining), and strongly positive (> 30% of cells with strong membranous staining).
Fluorescent In Situ Hybridization
Fluorescent in situ hybridization (FISH) for Her2-neu gene amplification was performed on corresponding sections of weakly positive (borderline) tumors by immunohistochemistry using the PathVysion Her2-DNA Probe Kit (Vysis, Downers Grove, IL; no. 35-161060) according to the manufacturer's instructions. Briefly, the PathVysion kit applies two DNA probes directly labeled with different fluorescent dyes. The dyes are Spectrum Orange fluorophore-labeled LSI Her2-neu, which is specific for the Her2-neu gene locus on chromosome 17q12-21.32, and Spectrum Green fluorophore-labeled CEP 17, which is targeted to the α satellite DNA sequence located at the centromeric region of chromosome 17. Signal enumeration was performed following the criteria established by Hopman et al.23 The adjacent ductal epithelial cells served as the internal control. Signals were counted for 60 tumor nuclei within the area of interest, and the results were reported as the ratio of average Her2-neu copy number to the number of CEP 17 per nucleus (total sum of Her2-neu signals divided by the total sum of CEP 17 signals). The expected ratio of LSI Her2-neu to CEP was < 2.0 for normal or unamplified tumor tissue specimens. A ratio of > 2.0 was considered positive for Her2-neu gene amplification, and the results were reported as amplified with an estimated copy number or as unamplified, as recommended by the manufacturer (Vysis).
The association between frequency counts in 2 × 2 tables was assessed using the Fisher exact test. The Mann–Whitney test was used to determine whether the distributions of mitosin values and mitotic activity were different in different patient subgroups.
Table 1 summarizes the characteristics of prechemotherapy breast tumors by stage, nuclear grade, histologic type, mitotic count, proliferative activity by mitosin immunoreactivity, and Her2-neu gene expression/amplification status. Sixty-one percent of tumors had a high nuclear grade (poorly differentiated), and 43% and 41% of those tumors showed high mitotic counts (> 6 mitotic figures per 10 HPF) and high proliferative activity by mitosin immunoreactivity (> 15% tumor cell nuclei). The high nuclear grade was associated positively with the percentage of mitosin positive tumor nuclei (Fig. 1). Sixty-three tumors (77%) showed no evidence of Her2-neu overexpression, and 18% of tumors showed Her2-neu gene overexpression or amplification, which were revealed by immunohistochemistry or FISH (Fig. 2). The Her2-neu gene status for the remaining four tumors (5%) was unknown because of the inadequate number of tumor cells on examined sections.
|Category||No. of patients||%|
|Tumor clinical size|
|1 (well differentiated)||5||6|
|2 (moderately differentiated)||27||33|
|3 (poorly differentiated)||50||61|
|Invasive ductal carcinoma||72||87|
|Invasive lobular carcinoma||7||8|
|Low (≤ 6/10 hpf)||47||57|
|High (> 6/10 hpf)||35||43|
|Low (≤ 15%)||48||58|
|High (> 15%)||34||42|
Pathologic examination of postchemotherapy mastectomy and segmental mastectomy specimens showed that 12 of 82 patients (15%) had a CPR to preoperative chemotherapy, 28% had an NCR, and 57% had a PPR or a minimal response. The distribution of chemoresponsiveness according to nuclear grade (NG) showed that NG3 tumors (poorly differentiated) demonstrated significantly better responses (24% CPR and 32% NCR) compared with NG1 tumors (well differentiated) and NG2 tumors (moderately differentiated) together (0% CPR and 25% NCR; P = 0.002; Fisher exact test). All 12 patients who achieved a CPR had NG3 tumors (Table 2). There was no correlation between histologic type and response to therapy; however, the great majority were classified as ductal. High proliferative activity, according to either mitotic count (> 6 mitotic figures per 10 HPF) or mitosin immunostaining (> 15% positive tumor cell nuclei), was correlated significantly with a good pathologic response to preoperative chemotherapy when the CPR and NCR groups were combined (P = 0.004 and P = 0.028, respectively; Fisher exact test) (Tables 3, 4). A higher percentage of Her2-neu positive tumors showed good pathologic response (CPR or NCR). Nine of 15 patients (60%) with Her2-neu positive tumors achieved a CPR or an NCR, and 24 of 63 patients (38%) with Her2-neu negative tumors achieved a CPR or an NCR (P = 0.152, Fisher exact test) (Table 5). There was no significant difference in either the mitotic activity or mitosin values between Her2-neu positive and negative tumors (P = 0.904 and P = 0.745, respectively; Mann–Whitney test).
|Nuclear grade||CR (%)||NCR (%)||PR (%)||Total|
|1||0 (0)||0 (0)||5 (100)||5|
|2||0 (0)||7 (26)||20 (74)||27|
|3||12 (24)||16 (32)||22 (44)||50|
|Total||12 (15)||23 (28)||47 (57)||82|
|Mitotic activity||CR (%)||NCR (%)||PR (%)||Total|
|Low (≤ 6/10 hpf)||2 (4.0)||12 (26)||33 (70)||47|
|High (> 6/10 hpf)||10 (28.6)||11 (31)||14 (40)||35|
|Total||12 (15.0)||23 (28)||47 (57)||82|
|Mitosin||CR (%)||NCR (%)||PR (%)||Total|
|Low (≤ 15%)||3 (6)||11 (23)||34 (71)||48|
|High (> 15%)||9 (27)||12 (35)||13 (38)||34|
|Total||12 (15)||23 (28)||47 (57)||82|
|Her2-neu over-expressiona||CR (%)||NCR (%)||PR (%)||Total|
|Negative||10 (16)||14 (22)||39 (62)||63|
|Positive||2 (13)||7 (47)||6 (40)||15|
|Total||12 (15)||21 (27)||45 (58)||78|
In addition, we observed that patient age was correlated significantly with complete response (P = 0.006; Fisher exact test): Younger patients had a higher incidence of CPR and NCR compared with older patients. However, pathologic response was not associated significantly with tumor stage (P = 0.186), lymph node positivity (P = 0.498), or dose intensity of the FAC protocol (P = 0.491).
Among a significant number of patients in the current study, doxorubicin-based neoadjuvant chemotherapy either completely eliminated or markedly reduced the tumor volume of the primary breast carcinoma. Patients who achieved a CPR or an NCR were more likely to have tumors with a high nuclear grade and proliferative activity. Although the difference was not significant statistically, a higher percentage of patients with Her2-neu positive tumors achieved a CPR or an NCR. Because a CPR to neoadjuvant chemotherapy is correlated with prolonged disease free survival and overall survival, the identification of features that predict tumor responsiveness to chemotherapy would be useful in selecting patients for this multidisciplinary strategy. Despite intensive research for the predictive clinical and biologic markers for response to neoadjuvant chemotherapy, no single clinical or biologic parameter has been identified as a reliable predictor of response to chemotherapy.
The reported histopathologic response rates to preoperative chemotherapy range from 3% to 46% for complete responses and from 30% to 90% for partial responses.2, 4, 7, 9, 24–26 The response rates in our series (15% CPR, 28% NCR, and 57% PPR/no response) are in agreement with those reported in most studies. Although most studies showed a strong correlation between high nuclear grade tumors and significant response to neoadjuvant chemotherapy,14, 17, 25–27 isolated reports have suggested the contrary.12, 13 In our series, patients who had high nuclear grade tumors showed significantly higher percentages of CPRs (24%) and NCRs (32%) compared with patients who had low and intermediate nuclear grade tumors (0% CPRs and 26% NCRs; P = 0.002). Similarly, a number of studies that evaluated the proliferative capacity of primary tumors by Ki67 immunostaining and cell cycle analyses showed a correlation between high proliferative activity and an increased sensitivity to chemotherapy.13, 27–30 Although some studies have suggested that there is no correlation between proliferative activity and tumor response,25, 31 our study showed a significant difference in response to neoadjuvant chemotherapy between patients who had highly proliferative tumors compared the response of patients who had tumors with low proliferative activity, as demonstrated by either a high mitotic count (P = 0.004) or a high percentage of nuclear positivity for mitosin (P = 0.028). Mitosin is a recently identified nuclear phosphoprotein that is involved in cell division and is expressed in the late G1, S, G2, and M phases of the cell cycle, but not in G0 phase.32 Clark et al. demonstrated that the tumor proliferative fraction, as determined by immunohistochemical assessment of mitosin, can provide prognostic information in patients with lymph node negative breast carcinoma.33 Highly proliferative tumors show more frequent cytokinetic modifications during neoadjuvant chemotherapy, which generally is related significantly to tumor regression, and low-proliferation tumors frequently show unchanged cytokinetic parameters or fewer cytokinetic modifications.34 A recent study showed that a decrease in Ki67 on Day 10 or Day 21 of the first cycle of chemotherapy significantly predicted for subsequent clinical response.34 Therefore, high proliferative activity before neoadjuvant chemotherapy and a decrease of proliferative activity after neoadjuvant chemotherapy may be one of the important underlying mechanisms for tumor response to chemotherapy.
Her2-neu is a tyrosine kinase receptor located on the cell membrane. Activation of the receptor induces a phosphorylation cascade in cytoplasmic proteins, including raf and ras, which results in the activation of nuclear transcription proteins and cellular growth.35 Her2-neu gene overexpression is detected in 15–30% of breast tumors.36–39 Her2-neu gene amplification reportedly is associated with large tumor size, lymph node positivity, advanced stage, absence of steroid receptors, and a high rate of proliferation.40–45 A number of studies have evaluated the correlation between Her2-neu expression in tumors and response to doxorubicin-based chemotherapy. Initial studies suggested that Her2-neu-positive tumors were associated with an increased sensitivity to dose intensive, doxorubicin-based chemotherapy and an improved clinical outcome in patients with axillary lymph node positive breast tumors, whereas patients with Her2-neu negative tumors showed no difference in response.45 Although some of the subsequent studies reported similar results,15, 16, 29, 46 others yielded conflicting data or failed to demonstrate any value of Her2-neu overexpression in predicting response to doxorubicin-based chemotherapy.17, 34, 47–49 A number of factors probably contributed to these discordant results, including various chemotherapy protocols, different dosage intensities, patient selection criteria, and different methods of detection and interpretation of Her2-neu overexpression. In the current study, 18% of tumors showed Her2-neu overexpression, and there was a trend showing that the status of Her2-neu gene overexpression was associated with a higher percentage of CPR and NCR to FAC-based chemotherapy (P = 0.152). However, this result was limited by the relatively small number of Her2-neu positive tumors in our series, and larger prospective studies are needed to address the issue of whether Her2-neu overexpression is associated significantly with a good histopathologic response to FAC-based chemotherapy.
In summary, the current results suggest that nuclear grade and proliferative activity not only are indicators of tumor biologic behavior but also may serve as important predictors of tumor response to doxorubicin-based neoadjuvant chemotherapy. Prospective studies with a larger series of patients, including more patients with well-differentiated tumors (low nuclear grade), are needed to better assess the predictive value of tumor nuclear grade, mitotic activity, and biomarker expression profiles in predicting the pathologic response of patients with primary breast carcinoma to neoadjuvant chemotherapy.
Histology support was provided by The University of Texas M. D. Anderson Cancer Center Breast Tumor Bank, which is supported by the Nellie B. Connally Breast Cancer Research Fund.
- 14p53 and c-erbB-2 expression and response to preoperative chemotherapy in locally advanced breast carcinoma. Breast Dis. 1995; 8: 149–158., , , et al.
- 21American Joint Committee on Cancer. Manual for staging of cancer, 6th ed. Philadelphia: Lippincott, 1998.
- 44Expression of c-erbB-2 oncoprotein: a prognostic indicator in human breast cancer. Cancer Res. 1989; 15: 49(8): 2087–2090., , , et al.
- 45HER2/neu amplification and comedo type breast carcinoma. Lancet. 1989; 3: 1268–1269., , , et al.