Reynolds, Christopher A.

Abstract Background: Metaplastic breast carcinoma (MBC) accounts for <5% of all breast malignancies. This entity represents a heterogenous group of tumors in which the adenocarcinomatous element is admixed with one or more neoplastic mesenchymal (spindle, squamous, chondroid, or osseous) elements. Patients typically present with large triple negative tumors and have a poor prognosis. The likelihood of lymph node (LN) involvement has been reported to be low in MBC patients. Due to the paucity of data, we undertook this study to explore the role of axillary ultrasound (US) and sentinel LN biopsy (SLNB) in the management of MBC. Methods: With IRB approval, we retrospectively identified patients diagnosed with MBC from 2001–2011 from the surgical pathology database. Histopathology and imaging were reviewed. Demographic, treatment and outcome data were obtained by clinical chart review. Data were analyzed using JMP 9.0 software. Results: We identified 41 women with MBC. Median age was 60 years (range 33–90). Histologic subtypes were spindle cell (46%), mixed adenocarcinoma and mesenchymal elements (20%), squamous cell (17%) chondroid/osseous (10%) and adenosquamous (7%). Tumor stage was T1 (24%), T2 (44%), T3 (12%) and T4 (20%). 26 patients (63%) were treated by mastectomy and 15 (37%) by wide excision. Of the 38 patients who underwent LN surgery (6 low-grade MBC, 32 intermediate/high-grade MBC), 10 (26%) were LN positive. All low-grade MBC patients were LN negative while 10 of 32 intermediate/high-grade MBC patients (31%) had LN metastasis. 22 patients had a preoperative axillary US. 14 patients had a negative axillary US and none had LN metastasis. 4 of 8 patients with suspicious axillary US findings had a preoperative axillary LN fine needle aspiration biopsy (FNAB) and 3 were positive for cancer. These patients proceeded directly to axillary LN dissection (ALND). 24 patients had a SLNB of whom one was SLN positive and underwent completion ALND. LN metastasis was associated with larger tumor size (p = 0.003), higher tumor grade (p = 0.04), angioinvasion (p = 0.07) and abnormal axillary US (p = 0.003). Surviving patients were followed for a mean (median) of 41 (30) months during which 13 (32%) recurred at a median of 6 months (IQR 3–17 months) and 11 (27%) subsequently died of disease. One SLN negative patient developed an axillary recurrence at 8 months, was successfully treated by ALND and is disease-free at 38 months. There were no axillary LN relapses after ALND. Conclusions: For clinically node-negative MBC patients, our contemporary data series suggests the incidence of occult LN metastasis is sufficiently high to warrant LN staging especially for patients with intermediate and high-grade tumors. This can be accomplished in a minimally invasive fashion with reasonable accuracy (∼97%) with axillary US, FNAB of sonographically suspicious LNs, and SLNB for patients with negative axillary US or FNAB. Axillary LND should be performed for patients with clinically and/or FNAB-positive LN for enhanced disease control. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-01-22.

Abstract Background: p16 is a nuclear protein encoded by the p16INK4a gene that regulates the G1-S cell cycle checkpoint. Its overexpression could reflect normal proliferative arrest, but also the evolution of carcinogenic processes that have overcome the p16 block, with progression from hyperplasic lesions to invasive breast cancer. Our aim with this study was to correlate p16 expression, both individually and in combination with expression of COX-2 and Ki67, with subsequent risk of breast cancer among women with atypical hyperplasia. Methods: p16 expression was assessed by immunohistochemical assays in archival paraffin-embedded, formalin-fixed sections from 233 women with atypia whose biopsy specimens were obtained via surgical excision at the Mayo Clinic from January 1, 1967, through December 31, 1991. p16 expression was scored as percent of cells positive and intensity of staining. Risk factor information and follow-up for breast cancer events were obtained via study questionnaire and the medical records. Standardized incidence ratios (SIRs) were used to compare observed numbers of breast cancer events to population-based expected counts, both at 10 years and over the entire course of follow-up. Results: Forty-seven patients (20%) developed breast cancer with a mean follow-up of 14.3 (SD=6.96) years. The overall risk of developing breast cancer was not modified by increasing p16 overexpression (low percent of staining with RR=3.89, 95% CI=2.72 to 5.56; high percent of staining with RR=3.91, 95% CI=2.43 to 6.28; similar results for intensity of staining). However, overexpression of both p16 and COX2 conveyed a stronger risk of breast cancer (SIR=4.01, 95% CI 2.28 to 7.07) as compared to dual low expression (SIR 1.78, 95% CI 0.67-4.74). Moreover, overexpression of both Ki67 and p16 was related to a higher 10-year breast cancer risk (SIR 6.25, 95% CI 2.35-16.6) as compared to low expression of both markers (SIR 0.87, 95% CI 0.22-3.49). Conclusions: although single expression of p16 does not stratify breast cancer risk, its overexpression in concert with biomarkers that reflect other carcinogenic processes (e.g. Ki67 and COX2) could help to define risk of tumor development. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 910.

Abstract Background: Uncontrolled proliferation is a defining feature of the malignant phenotype. Ki67 is a marker for proliferating cells and is overexpressed in many breast cancers. Atypical hyperplasia is a premalignant lesion of the breast (relative risk ∼ 4.0). Here we asked if Ki67 expression could stratify risk in women with atypia.Methods: Ki67 expression was assessed immunohistochemically by digital image analysis in archival sections from 192 women with atypia diagnosed at the Mayo Clinic 1/1/67-12/31/91. Risk factor and follow-up data were obtained via study questionnaire and medical records. Observed breast cancer events were compared to population expected rates (Iowa SEER) using standarized incidence ratios (SIRs). We examined both short-term (within 10 years) and long-term (after 10 years) risk of breast cancer (BC) following atypia biopsy.Results: The median value for percent positive cells for Ki67 was 1.0%; the 75th percentile value was 2.3%. Based on the empirical distribution of staining values in our cohort, we selected a cutoff of 2% cells positive to separate high from low staining. There were no differences in Ki67 levels by age at biopsy, type of atypia (ADH, ALH, or both), number of foci of atypia, or family history. 32 women developed BC over a median of 14.6 years. Among them, those with ≥2% Ki67 expression had a shorter time to breast cancer (median 5.5 years, IQR=3.2-7.2) than those with <2% cells positive (median=13.8 years, IQR=11.6-20.3). There was a positive association between Ki67 overexpression (≥2% of cells positive) and risk of BC in the first 10 years of follow-up (SIR=4.42 [95% CI 2.21-8.84]). This excess risk resulted in a 10-year cumulative incidence of 14.1%. In contrast, in the women with low Ki67 expression, we found no increased risk of BC in the first 10 years, with SIR 1.01 (95% CI 0.38-2.70), which was significantly lower than the women with high expression (p=0.01). The 10-year cumulative incidence of BC was 3% for the low Ki67group, in line with population averages. After 10 years, risk increased significantly in the low Ki67 group [SIR 5.69 (3.63-8.92)] vs. no increased risk in the high Ki67 group [SIR 0.78 (0.11-5.55)]. We formally examined this apparent time-dependent difference in incidence patterns by Ki67 staining levels via Poisson regression analysis, and found a statistically significant interaction between pre- vs. post-10 years incidence and high vs. low staining levels (p<0.001).Conclusions: Ki67 appears to be a time-varying biomarker of risk of breast cancer in women with atypical hyperplasia. Women whose atypias show higher proliferation rates have an increased risk for an earlier BC (within 10 years). Those with lower proliferation rates tend to develop BCs later, after 10 years. Ki67 levels may help to define the time period of greatest risk for women with atypia. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 909.

Abstract Background: Risk prediction models are important clinical tools for counseling and decision making for women with an elevated risk of breast cancer, such as women with atypical lobular or ductal hyperplasia (atypia). The Gail model has been shown to underestimate the risk of breast cancer in women with atypia. We sought to evaluate the performance of the Tyrer-Cuzick model in a well-defined cohort of women with atypia.Methods: The Mayo Benign Breast Disease (BBD) Cohort includes 9,376 women aged 18 to 85 who had an open breast biopsy with benign pathological findings between 1967 and 1991. The subset of women with atypia was identified by our study pathologists. Risk factor data required by the Tyrer-Cuzick model were collated for each woman, and the model was used to predict each woman's risk of developing invasive breast cancer within ten years. The predicted numbers of breast cancers were compared to the observed numbers to assess overall model calibration, and the concordance between breast cancer outcomes and predicted probabilities was used to measure the accuracy of the model on the individual level. Subset analysis of women with strong family history was performed. Strong family history was defined as: at least one first-degree relative with breast cancer before the age of 50 years or two or more relatives with breast cancer, at least one being a first degree relative.Results: Three hundred and thirty one women with atypia were identified. Over a mean follow up of 13.7 years, 62 (19%) women have developed invasive breast cancer. Thirty one women developed invasive cancer in the first ten years after biopsy. The Tyrer-Cuzick model predicted that 64.6 invasive breast cancers would occur in the first ten years after biopsy. Thus, the predicted to observed ratio was 2.08 [95% CI: 1.47 – 2.96], p < 0.001).The concordance statistic was 0.51, revealing that the Tyrer-Cuzick model did not accurately distinguish, on an individual level, between women who were diagnosed with invasive breast cancer within ten years of biopsy and those who were not.Performance was somewhat better in the subset of 68 women with a strong family history of breast cancer. The model predicted 12.3 invasive breast cancers, and 7 were observed, for a predicted to observed ratio of 1.76 [95% CI: 0.84 – 3.70]. The concordance statistic was 0.57.Conclusion: The Tyrer-Cuzick model significantly overestimated the risk of breast cancer in the first ten years after biopsy for this group of women with atypia. Individual risk estimates showed poor concordance. Performance was slightly improved in women with a strong family history of breast cancer. Physicians should be cautious when using the Tyrer-Cuzick model to predict ten-year risk in patients with atypia. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 6063.

Abstract Background: The decision to use adjuvant chemotherapy in patients with early stage breast cancer is based in part on the estimation of risk of tumor recurrence by physicians, which traditionally relies heavily on tumor size, nodal status and a set of biologic tumor characteristics such as hormone receptor and HER2 expression. The Oncotype DX® assay is a 21-gene expression profile aiming to improve risk stratification, recurrence prediction and optimize selection of patients for adjuvant chemotherapy.Methods: We selected 154 consecutive patients with early stage estrogen receptor (ER) positive breast cancer and available Oncotype Dx® recurrence score (RS) for the study. Clinicopathologic data, including patient age, menopausal status, tumor size, histologic type, grade, mitotic activity, presence of lymphatic invasion (LVI), nodal status, hormone receptor and HER2 status on all patients were provided to four surgical oncologists, four medical oncologists and three pathologists, specializing in breast cancer diagnosis and management. Participants were asked to estimate the risk of recurrence of tumors based on available clinicopathologic data and to provide the three most important tumor features their risk estimates were based on. Risk estimates of participants were compared with RS results.Results: Based on the Oncotype Dx® results, 95 (61.7%), 45 (29.2%) and 14 (9.1%) tumors were of low (RS <18), intermediate (RS 18-30) and high (RS ≥31) risk, respectively. RS values showed a highly significant correlation with tumor grade, mitotic activity, LVI, hormone receptor and HER2 status, while no correlation with patient age, menopausal status, tumor size and histologic type was found. Participants' risk estimates agreed with those of the Oncotype Dx® assay in 54.2 ± 2.3 % (mean ± SEM, range 41.6 - 63.0%) of cases, while the risk of recurrence was over- and underestimated compared to RS results in 31.8 ± 3.1% (16.2 - 43.5%) and 14.1 ± 1.4% (7.1 - 22.7%), respectively. The rates of overestimation were significantly higher than those of underestimation (p = 0.0003). Although medical oncologists tended to overestimate the risk more frequently (38.1 ± 2.0%) compared to surgeons (28.7 ± 5.9%) and pathologists (27.5 ± 7.8%), the difference did not reach statistical significance. Estimation of the agreement of participants' risk assessment with RS results showed a mean kappa value of 0.2955 (range 0.1506 - 0.4123). No statistically significant difference in overall concurrence with RS results was found between surgeons, medical oncologists and pathologists. Participants ranked tumor stage/nodal status, hormone receptor status and tumor size to be the most important features when estimating recurrence risk.Conclusions: Based on traditional clinicopathologic features alone, surgeons, medical oncologists and pathologists tend to overestimate the risk of tumor recurrence as compared to Oncotype Dx® assay results. The RS may provide additional information regarding the intrinsic biological features of ER positive breast cancers and help tailoring treatment recommendations. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4061.

Abstract Background: Current guidelines recommend administration of chemotherapy for patients with breast carcinomas >1 cm in size, with consideration for patient age, comorbidities and tumor grade. However, it is unknown which patients actually benefit from therapy and overtreatment of a significant proportion of patients is a major concern. We investigated the impact of the Oncotype Dx® Recurrence Score (RS) on chemotherapy recommendations in early stage estrogen receptor (ER) positive breast cancer patients.Methods: We selected 154 patients with early stage ER positive breast cancer and available RS for the study. Clinicopathologic data, including age, menopausal status, tumor size, type, grade, mitotic activity, presence of lymphatic invasion, nodal status, hormone receptor and HER2 status on all patients were provided to four surgical oncologists, four medical oncologists and three pathologists, specializing in breast cancer diagnosis and management. Assuming that all patients were in good general health and would receive endocrine therapy, participants were asked whether they would also advovate adjuvant chemotherapy based on clinicopathologic data with and without knowledge of the RS, and to provide the three salient clinicopathologic features on which their recommendations were based. Changes in recommendations of participants following inclusion of RS data were compared.Results: Based on RS results, 95 (61.7%), 45 (29.2%) and 14 (9.1%) tumors were of low (RS <18), intermediate (RS 18-30) and high (RS ≥31) risk, respectively. The results are summarized in Table 1. Assuming that the hypothesis previously put forward that patients with low to intermediate risk RS are not likely to benefit from chemotherapy, 82.3 ± 1.3% (75.5 - 89.0%) and 69.0 ± 6.9% (5.9 - 85.7%) of patients for whom chemotherapy was recommended by the participants would be "overtreated" without and with the use of RS results (p = 0.0322). No statistically significant difference was found among surgeons, medical oncologists and pathologists. Participants ranked patient age/menopausal status, hormone receptor status and tumor stage/nodal status to be the most important features when recommending chemotherapy.Conclusions: Although current recommendations for adjuvant chemotherapy for early stage ER positive breast cancer patients are largely in line with published guidelines, inclusion of RS alters recommendations in about 25% of cases. While medical oncologists recommended chemotherapy more frequently compared to surgeons and pathologists, they were more likely to change recommendations in light of RS results.Table 1. Summary of results SurgeonsMedical oncologistsPathologistsp* Mean ± SEMRangeMean ± SEMRangeMean ± SEMRange Chemo without RS (%)29.2 ± 1.824.0 - 31.859.0 ± 5.046.8 - 70.846.8 ± 3.741.6 - 53.90.0156Chemo with RS (%)27.0 ± 5.611.0 - 36.438.6 ± 9.517.5 - 63.644.4 ± 5.833.8 - 53.90.1794No change (%)75.3 ± 7.054.5 - 85.766.7 ± 6.851.3 - 83.885.9 ± 6.478.6 - 98.70.3682Add chemo (%)11.2 ± 4.13.2 - 22.76.5 ± 1.93.2 - 11.75.8 ± 3.20.6 - 11.70.6882Avoid chemo (%)13.5 ± 3.94.5 - 22.726.8 ± 7.94.5 - 41.68.2 ± 4.50.6 - 16.20.2186*Kruskal-Wallis test Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4058.

Abstract Background: Endoxifen, the active metabolite of tamoxifen (Tam), is currently being developed as a drug for the treatment of estrogen receptor (ER) + breast cancer (BC). HER2 expression in ER+ BC is associated with Tam resistance, and in vivo Tam administration to mice bearing ER+/HER2+ xenografts stimulates BC growth (Shou, JNCI 2004). In humans, endoxifen is the most important Tam metabolite responsible for inhibiting estrogen induced BC growth (Wu, Cancer Research 2009). CYP2D6 metabolism affects the concentrations (conc) of endoxifen (Stearns, JNCI, 2003) and associated with worse disease free survival (DFS) in Tam treated BC (Goetz, JCO, 2005). We sought to determine the activity of Tam and its metabolites in ER+/HER2+ BC cell lines and to evaluate the role of CYP2D6 metabolism in Tam-treated patients (pts) with ER+/HER2+ BC. Additionally, we sought to determine endoxifen conc in mice administered oral Tam.Methods: MCF7 (parental and HER2-expressing) and BT474 (ER+/HER2+) cells were used to compare the activity of Tam, 4HT, and endoxifen on estrogen- stimulated growth. Oral tam PK were characterized in mice treated with standard dose of Tam (4 mg/kg; 100 μg). Clinical data were obtained via a retrospective analysis of Tam-treated pts with ER+/ HER2+ BC randomized to receive 5 years of Tam (NCCTG 89-30-52). CYP2D6 metabolism (extensive or decreased) was based on CYP2D6 genotype (*3, 4, 6, 10, 17, 41) and co-administration of a CYP2D6 inhibitor (yes/no). HER2 was determined by immunohistochemistry (IHC) or FISH (tumors 0, 1, or 2+ by IHC). The association between CYP2D6 and DFS was assessed using the log-rank test and proportional hazards modeling.Results: Compared to Tam, endoxifen potently inhibited the growth of estrogen- stimulated BT474 cells. In MCF7 cells, expression of HER2 shifted the conc of endoxifen required for 50% inhibition of growth (IC50) from 54 nM (parental) to 131 nM (HER2 expressing). Using the range of conc of Tam and its metabolites observed in humans (Tam, 300-500 nM; 4HT, 5-10 nM; and endoxifen, 10-200 nM), only endoxifen potently inhibited estrogen- stimulated growth of MCF7HER2+ cells and only at conc achievable in CYP2D6 extensive metabolizers (>50nM). In mice, conc of 4HT and endoxifen were below 15 nM following an oral dose of 4 mg/kg. In NCCTG 89-30-52, both CYP2D6 phenotype and HER2 status was determined in 201/256 randomized pts. HER2 was expressed in 23/215 (11%) but not associated with DFS overall (p=0.62). In the HER2+ subset, pts with decreased CYP2D6 metabolism (n=10) had significantly shorter DFS compared to extensive metabolizers (n=9) (HR 9.5, p=0.03; 95% CI 1.16-76.9).Conclusions: Our in vitro and clinical data provide a simple pharmacological model for understanding HER2 resistance in Tam-treated breast cancer. Mice, which lack the CYP2D6 enzyme, may not be an appropriate model for understanding tam resistance given low conc of both 4HT and endoxifen. Given that the in vitro conc of endoxifen needed to inhibit the growth of ER+/HER-2+ BC are achievable in only a subset of humans (CYP2D6 extensive metabolizers), the primary administration of endoxifen could overcome de novo Tam resistance in ER+/HER2+ BC. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 2006.

Abstract Abstract #1005 Background: Several models for the prediction of non-sentinel lymph node metastasis (+NSLN) in sentinel lymph node positive (+SLN) breast cancer patients have been published. Multiple variables have been shown to be important in prediction of +NSLN. To this date, models that utilize 7-8 variables have been significantly more discriminatory than models using only 2-3 variables. In this study, we evaluate the performance of the newly created Stanford online calculator (SOC) to predict the likelihood of +NSLN. The SOC uses 3 variables: primary tumor size, sentinel lymph node metastasis size, and presence/absence of lymphovascular invasion.
 Methods: From 1997-2004, 465 breast cancer patients with clinically negative axillae were found to have a +SLN and underwent completion axillary lymph node dissection (ALND) at the Mayo Clinic. Using the SOC the probability of having +NSLN was determined for each patient. Complete variable data was available for 464 patients which make up the cohort reported here. A receiver-operating characteristic (ROC) curve was created and the area-under-the-curve (AUC) was calculated. Mean probabilities of patients with and without +NSLN were compared. Additionally, patients with a low Stanford probability were examined to determine the model's accuracy in discriminating patients who could potentially be spared ALND. These results were compared to the results of the Mayo Clinic and Memorial Sloan Kettering Cancer Center (MSKCC) nomograms which have been previously applied to this same cohort of patients.
 Results: The AUC of the Stanford model was 0.72 (95% CI 0.67-0.77). The mean Stanford probabilities for patients with and without NSLN disease were 0.75 (range 0.06 to 1.0) and 0.50 (range 0.05 to 1.0), respectively (p < 0.0001). Examining patients with a Stanford probability <= 10%, we found that 47 patients met this criterion; 6 had +NSLN and 41 did not, for a false negative (FN) prediction rate of 13%. There was only 1 patient with Stanford probability <= 5%, and this patient did not have +NSLN. The AUC of the MSKCC and Mayo nomograms in this same patient population was 0.74 and 0.77, respectively. The AUCs did not differ significantly (p = 0.13) among the 3 models.
 
 Conclusion: Despite using only three variables, the Stanford nomogram appears to perform as well as, but not better than the MSKCC and Mayo nomograms in our patient cohort. Further validation in other patient populations is needed prior to widespread utilization of this nomogram. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 1005.