Citations

Abstract Abstract #4018 Background: Preoperative breast magnetic resonance imaging (MRI) has been shown to identify occult breast disease in women with newly diagnosed breast cancer. We sought to determine the accuracy of MRI in predicting the actual pathologic tumor size in women with newly diagnosed breast cancer.
 Methods: Preoperative breast MRI was reviewed in 460 women with newly diagnosed breast cancer. The size in terms of maximum tumor dimension, of 476 incident breast cancers on MRI were recorded and correlated with the actual pathologic size upon surgical resection. The correlation of size from MRI to pathology was determined. The size discrepancy was analyzed in relation to tumor size and histology.
 Results: MRI demonstrated 476 breast cancers (460 in the ipsilateral breast and 16 with concurrent contralateral breast cancer). Median age was 62 with a range of 25 to 91. The T stages of the population included T0 in 19% of the women, 60% had T1 lesions, 18% had T2 lesions and 3% were T3 or T4. The histology consisted of DCIS alone in 20%, infiltrating ductal in 74% and infiltrating lobular in 6%. Pre-operative MRI was unable to define an enhancing lesion usually due to hematoma following core needle biopsy in 9.8% (47/476). MRI described the size of the incident lesion in 90% (429/476). The mean difference in size between the preoperative MRI and the actual pathologic size for the entire group was 2.4 mm. MRI underestimated tumor size in only 3% of cases. The mean difference according to T stage and histology between MRI and actual pathologic size is listed in Table 1. The percentage of difference between MRI size and pathologic size for the entire group is presented in Table 2.
 
 
 
 Discussion: The mean difference in tumor size estimated by preoperative MRI in women with newly diagnosed breast cancer was within 5 mm of the actual pathologic size in 68% of patients. MRI was accurate in predicting T stage with no significant difference in relation to histology or T stage. This analysis further clarifies the role of preoperative MRI in surgical and treatment planning. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 4018.

Abstract Abstract #4014 Background
 Preoperative magnetic resonance imaging (MRI) of the breasts in women with breast cancer (BC) has improved the detection of occult ipsilateral (I/L) and contralateral (C/L) malignancies. Breast MRI findings in patients with estrogen receptor (ER), progesterone receptor (PR), and HER2 negative (triple negative (TN)) disease have not been previously reported. Women with TN BC are at increased risk of distant metastases and a reduced survival vs. BC patients as a whole. One potential explanation may be the presence of occult disease. The role of bilateral breast MRI in these patients remains to be determined.
 Hypothesis
 Patients with TN BC are more likely to have occult malignancy identified by preoperative MRI vs. those with non-TN BC.
 Methods
 All newly diagnosed BC patients from 2003-2007 were reviewed. Information on age, race, histology, ER, PR, HER2, staging, treatment, MRI findings, subsequent investigations, and additional cancer diagnoses was recorded. The incidence of additional I/L or C/L malignany was calculated. The percentage of total patients initially felt to be candidates for partial breast irradiation who required a change in their cancer management was calculated.
 Results
 55 (11%) were diagnosed with TN BC, of whom 52 women completed preoperative MRI. 456 (89%) had non-TN BC; 421 completed preoperative MRI. The median age among TN and non-TN patients was 62 years (Table 1). A majority of women in both groups were Caucasian. African-American women were more likely to be diagnosed with TN vs. non-TN disease (11% vs. 6%).
 
 I/L BC was documented in 13/52 (25%) of TN patients vs. 25% of non-TN patients. No C/L cancers were detected in the TN group (6% in the non-TN group). Of the TN group, 15 patients required a change in their treatment (29%) based on findings from MRI vs. 32% of non-TN patients.
 Conclusions
 Patients with a new diagnosis of TN BC are not at increased risk of occult malignancy than those with non-TN BC. 25% of patients with TN BC had additional I/L disease not previously detected on clinical examination, mammogram, or ultrasound. Of those with non-TN BC, 25% were diagnosed with an additional I/L and 6% with C/L malignancy. The benefit of MRI was limited by false positives, with a number of patients in both groups undergoing a benign biopsy.
 These results suggest that patients with TN BC are not more likely than those with non-TN disease to have additional occult I/L or C/L cancer. The incidence of additional disease remains high and preoperative MRI may be beneficial. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 4014.

Abstract Abstract #3103 Introduction: Lymphedema is well-known in the arm but can also occur in the breast. Diagnosis is currently non-standardized and based on clinical impression. We sought to determine if skin thickness could serve as a useful measure in the diagnosis of breast lymphedema (BLE).
 Methods: Patients undergoing unilateral non-mastectomy procedures were enrolled on this prospective clinical study preoperatively and evaluated for signs and symptoms of lymphedema in the operated breast in the postoperative period. Patients with established BLE were also studied. BLE diagnosis was independent of ultrasound findings and was based on a graded physical exam targeting clinical signs of edema and erythema. Skin thickness was measured with ultrasound at the 6 o'clock position of both breasts. Differences in skin thickness were assessed by the ratio of skin thickness in the operated and contralateral breasts, with comparison of means by two-sample t-test. Receiver operating characteristic curves were constructed to estimate the area under the curve (AUC) and to determine optimal cutpoints.
 Results: Ninety-seven women were studied; 85 were enrolled preoperatively and 12 postoperatively after a diagnosis of BLE. Median length of follow-up overall was 8 months. Of the 97 women, 46 had BLE at one or more follow-up visit with median time to first diagnosis of 3 months. Mean measured skin thickness was 2.3 ± 0.5mm in unaffected breasts and 3.2 ± 1.0 mm in operated breasts. Women with BLE had significantly greater skin thickness in the operated breast compared to the contralateral breast, with a mean skin thickness ratio of 1.83 ± 0.57 in patients with BLE compared to 1.18 ± 0.40 among those without BLE (p<0.0001). Skin thickness ratio provided excellent discrimination for BLE, with an area under the curve (AUC) of 0.86. At a ratio of 1.32 or greater (32% greater skin thickness in operated versus contralateral breast), sensitivity and specificity for diagnosing BLE were 81% and 82% respectively. A single measure of skin thickness in the operated breast also correlated well with diagnosis of BLE, with skin thickness of 3.2 mm or greater demonstrating sensitivity of 78% and specificity of 82%, AUC 0.82.
 Conclusions: Breast lymphedema results in skin edema and increased skin thickness that can be quantified with ultrasound. Measuring breast skin thickness- either as a single reading or as a ratio to the unaffected breast- may be useful in the diagnosis of breast lymphedema. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 3103.

Abstract Abstract #3057 Introduction: Patients (pts) with ER/PR/HER2 negative advanced breast cancer (triple negative, TN, basal-like) have limited treatment options and a poor prognosis. Ixabepilone (ixa), the first in a new class of antineoplastic agents, showed clinical benefit in combination with capecitabine (C) in 2 large clinical trials in metastatic breast cancer (MBC) pts either resistant to (study 046; JCO, 2007) or pretreated with anthracycline (A) and taxanes (T) (study 048). In 048, ixa + C, compared to C alone, demonstrated significant increases in PFS (HR 0.79 [0.69-0.90]) and ORR (43 vs. 29%). A trend towards increased OS was seen in both 048 (HR 0.90 [0.78-1.03]) and 046 (HR 0.90 [0.77-1.05]), which did not reach statistical significance. Here we present a pooled analysis of efficacy endpoints (ORR, PFS and OS) in pts with TN disease from these 2 phase III studies.
 Methods: 1973 pts with MBC previously treated with A and T were randomized in 2 phase III trials (046 and 048) to receive either ixa (40 mg/m2 IV over 3h Q3w) + C (1000 mg/m2 PO BID x14d Q3w) or C alone (1250 mg/m2 PO BID x14d Q3w). Due to the similarity of the study populations, individual pt data from both studies was pooled to better evaluate treatment effect within pre-planned patient subgroups.
 Results: 443 pts had TN disease in the two studies combined. ORR and PFS were superior in pts receiving combination therapy compared to those on C alone with a trend towards improved survival that did not reach statistical significance.
 
 Conclusions: Ixa + C is the first combination to show statistically significant PFS benefit in pts with advanced TN breast cancer from a pooled analysis of 2 phase III trials. Clinical benefit was consistently seen pts with TN disease in the individual studies as well. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 3057.

Abstract Abstract #2149 Background: Inhibitors of the insulin-like growth factor 1 receptor (IGF-1R) are currently undergoing clinical testing. Preclinical investigations have identified IGF-1 signaling as a key mechanism for breast cancer growth and resistance to clinically useful therapies, including tamoxifen and trastuzumab. Thus, agents targeting IGF-1R have promise in the treatment of breast cancer. Determining mechanisms that can confer resistance to these agents may aid their clinical development.
 Methods: To understand factors may be important in predicting sensitivity to targeting the IGF-1 signaling pathway, we developed a cell line (MCF-7R4) that is resistant to BMS-554417, a small molecule, dual-kinase inhibitor of IGF-1R and insulin receptor (InsR). Compared with the parental MCF-7 cells, MCF-7R4 cells are 40- to 50-fold resistant to BMS-554417 and cross-resistant to the similar compound BMS-536924. The expression profiles of MCF-7R4 and that of MCF-7 were compared using Affymetrix GeneChip Human Genome U133 Plus 2.0 Arrays. Intracellular concentrations of BMS-536924 were examined by reverse phase high performance liquid chromatography. BMS-536924 cellular accumulation in vitro was visualized by fluorescence microscopy using a DAPI filter set. MCF-7 cells stably transfected with either the empty mammalian expression vector pcDNA (MCF-EV) or full length BCRP (MCF-BCRP) were examined for sensitivity to BMS 536924 by MTS assays.
 Results: Compared to MCF-7 cells, BCRP expression was increased 9-fold in MCF-7R4, which was highly statistically significant by t-test (p= 7.13E-09). Little change was observed in other ABC transporter proteins, including ABCB1. No change was observed in IGF-1R or InsR expression. BCRP overexpression in MCF-7R4 cells was confirmed by western blotting. MCF-7R4 cells had significantly lower intracellular accumulation of BMS-536924 compared to MCF-7 cells. Confirming these results, MCF-BCRP cells were significantly less sensitive to the cytoxic effects of BMS-536924 cells than MCF-EV cells.
 Conclusions: BCRP expression was stimulated by prolonged exposure of MCF-7 cells to BMS-554417. Upregulation of BCRP is one of the most significant changes observed in MCF-7R4 cells in comparison to parental cells. BCRP expression decreased cellular exposure to BMS-536924 and was sufficient to confer resistance. These data suggest that BSM-536924 is a substrate for BCRP-mediated efflux. Expression of BCRP may be important in de novo and acquired resistance to benzimidazole –based inhibitors of IGF-1R/InsR. Supported in part by the Mayo Clinic Breast SPORE (CA116201-03), NIH K12 (CA090628-05) and the Mayo Clinic Cancer Center (CA15083). Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2149.

Abstract Abstract #2131 Introduction: TAS-108 is an oral steroidal anti-estrogen agent that selectively inhibits ERα and is mainly metabolized by CYP3A4. The purpose of this study is to investigate the efficacy and safety of TAS-108 administrated orally in three dose levels in patients with locally advanced, locally recurrent inoperable or metastatic breast carcinoma (BC) in four countries (USA, Russia, Mexico and Chile).
 Methods: Postmenopausal women with confirmed ER and/or PgR positive BC who had previously responded to one or two standard endocrine therapies, with or without one prior chemotherapy were randomly assigned to three doses of TAS-108, 40 mg, 80 and 120 mg daily. Using a modified Panageas' optimal two-stage trinomial design, the enrollment of 60 evaluable patients (first-stage: 19) was required to each individual dose group. Tumor response was assessed every 8 weeks according to RECIST criteria. Adverse events (AEs) were graded by CTC-AE v3.0.
 Results: A total of 146 patients with mean age of 63 years old were enrolled with 61 patients in the 40 mg group and 66 in the 80 mg group. The 120 mg group was terminated at the end of stage 1 with 19 patients enrolled due to lack of efficacy. The mean duration of study treatment was 172 days for the 40 mg group and 160 days for the 80 mg group. Partial response (PR) was documented in 6 (10%) patients in the 40 mg group and 4 (6.7%) patients in the 80 mg group. The rate of disease stabilization (CR+PR+SD) reported by the investigators was 43% (95%CI, 31%,56%) in the 40 mg group and 45% (95%CI, 32%,58%) in the 80 mg group. Adjudicated clinical benefit (CR+PR+SD more than 24 wks) was observed in 22% of patients in the 40 mg group and 20% of patients in the 80 mg group. Clinical benefit was achieved in 25% of patients with 1 prior hormonal therapy and 15% of patients with more than one line of prior hormonal therapy. Median time to progression (TTP): 15 weeks for the 40 mg group and 15.9 weeks for the 80 mg group. Median duration of clinical benefit was 32 weeks in the 40 mg group and 64 weeks in the 80 mg group. In the 40/80/120 mg groups, the commonly reported treatment-related AEs included nausea (15%/11%/16%), fatigue (10%/11%/16%), headache (10%/6%/16%), hot flushes (10%/5%/32%), diarrhea (2%/6%/5%), constipation (3%/3%/5%), and arthralgia (2%/5%/11%). No endometrial cancer and treatment-related deaths occurred during the study.
 Conclusions: TAS-108 has demonstrated anti-tumor activity in this population and was generally well tolerated. The 40 mg dose was chosen as the recommended dose for future clinical evaluation in patients with advanced breast cancer. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2131.

Abstract Abstract #2109 Background: Despite the impressive results of the recently released trastuzumab adjuvant therapy trials, 15% of patients with HER2 overexpressing or amplified breast cancer developed tumor relapse at 4 years. Lapatinib is a small molecule reversible TKI that inhibits both ErbB1 and ErbB2. The current study was developed to assess the cardiac safety and feasibility of adding lapatinib to paclitaxel and trastuzumab following as part of adjuvant therapy.
 Methods: A single-arm phase II study of doxorubicin (A, 60 mg/m2 day 1) and cyclophosphamide (C, 600 mg/m2 day 1) [q2w or q3w for 4 cycles]; followed by paclitaxel (P, 80 mg/m2 days 1, 8, 15), trastuzumab (T, 4 mg/kg loading dose then 2 mg/kg days 1, 8, 15), and lapatinib (L, 1000 mg days 1-21) [12 weeks]; followed by T (6 mg/kg day 1) and L (1000 mg days 1-21) [40 weeks] was conducted. The primary endpoint was the incidence of congestive heart failure. The current unplanned safety analysis was undertaken due to the observance of a high rate of G3/4 diarrhea.
 Results: From April 2007 to June 2008, 98 pts were enrolled and initiated study treatment. Median age was 51 (range 32-72). Among 83 pts with adverse event (AE) data available, 50 (60%) pts have experienced a G3/4 non-hematologic AE. During post-AC treatment, among 53 pts with AE data available, 31 (58.5%) patients have experienced a G3/4 non-hematologic AE with 24 (45%) patients reporting G3/4 diarrhea. Median cycle of onset of G3/4 diarrhea was cycle 5 (first cycle of PTL) with 16 (64%) cases first reported during cycle 5 and 5 (20%) cases first reported during cycle 6. Among 57, 46, 38, and 32 pts receiving treatment with PTL during cycles 5-8, 65%, 57%, 61%, and 72% of patients received the full L dose, respectively. 31 patients have ended active treatment with 10 due to patient refusal and 8 due to adverse events.
 Conclusions: Preliminary data suggest that L given concurrently with P and T at a dose of 1000 mg per day induces an unacceptable rate of moderate to severe diarrhea. Careful monitoring of diarrhea as well as L dose reduction and initiation of loperamide at first occurrence of diarrhea are recommended. The dose of L when given concurrently with P and T has been amended to 750 mg per day in the current study and safety data for the 1000 mg and 750 mg per day cohorts will be presented. Implications for the ongoing ALTTO study will also be presented. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2109.

Abstract Abstract #19 Background: Tamoxifen (TAM) is a standard endocrine therapy for the treatment of women with estrogen receptor (ER) positive breast cancer. TAM is activated by the cytochrome P450 2D6 enzyme system into two potent and active metabolites, 4-hydroxytamoxifen (4HT) and 4-hydroxy-N-desmethyl-tamoxifen (endoxifen). While human concentrations of 4HT are negligible and vary little in plasma (5-10 nM), endoxifen concentrations vary widely (10-180 nm), and women with genetically impaired CYP2D6 metabolism have significantly reduced endoxifen levels and a higher risk of breast cancer recurrence. Despite these observations, endoxifen's contribution to tamoxifen's overall drug effectiveness is uncertain.
 Methods: Using cells endogenously expressing ERa (MCF7, T47D) and cells stably transfected with ERa (Hs578T and U2OS), we examined the relative effects of TAM and its primary metabolites on ERa protein levels by western blotting, ERa transcriptional activity by luciferase reporter assays and real-time RT-PCR, and on ER positive breast cancer cell growth through the use of proliferation assays.
 Results: We have discovered that endoxifen induces ERa protein turnover through proteasomal degradation similar to that of ICI in a concentration and time-dependent manner. These findings are in stark contrast to TAM, N-desmethyl-tamoxifen (NDT) and 4HT, which stabilize the ER. Optimal degradation occurs only at endoxifen concentrations observed in human CYP2D6 extensive metabolizers (> 40 nM) and persists even in the presence of TAM (300 nM), 4HT (7 nM), and NDT (700 nM) at concentrations observed in patients receiving tamoxifen therapy. In contrast, reducing endoxifen concentrations to those observed in a CYP2D6 poor metabolizer (20 nM), without altering TAM, 4HT, and NDT, results in ER stabilization. High endoxifen concentrations (100-1000 nM) completely block estrogen (E2)-induced ER transcriptional activity even in the presence of TAM, 4HT, and NDT, while low endoxifen concentrations (20-40 nM) do not. Further, low concentrations of endoxifen (20 nM) do not significantly alter E2-induced cell proliferation; however, high concentrations of endoxifen (100-1000 nM) completely block this process. Discussion: Our data demonstrate that endoxifen is a potent anti-estrogen that targets ERa for proteasomal degradation, blocks ERa transcriptional activity and inhibits E2-induced breast cancer cell proliferation. Importantly, these effects of endoxifen are observed at concentrations found in CYP2D6 extensive metabolizers and are maintained even in the presence of TAM, 4HT and NDT. These studies suggest that endoxifen may be the primary metabolite responsible for tamoxifen's effectiveness in the treatment of ER positive breast cancer and provide the impetus to evaluate whether TAM-related side-effects, such as thrombo-embolism and endometrial hyperplasia/carcinoma, are inversely associated with a patient's ability to metabolically activate tamoxifen. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 19.

Abstract Abstract #12 Background: Meta-analyses (EBCTCG, Lancet 2005; 365:2687) have established in estrogen receptor (ER) positive breast cancer (BC) that 5 yrs of adjuvant tamoxifen (Tam) substantially reduces recurrence and mortality. The aromatase inhibitors (AIs) anastrozole, exemestane and letrozole have been studied in prospective trials in postmenopausal patients (pts) in comparison with Tam either as initial monotherapy (Cohort 1) or after 2 to 3 years of Tam in a switching strategy (Cohort 2); both to a total of 5 years (yrs) of endocrine therapy.
 Methods: Outcomes data submitted to EBCTCG were used in separate meta-analyses of Cohort 1 (trials: ATAC, BIG 1-98/IBCSG 18-98) and of Cohort 2 (trials: ABCSG 8, ARNO 95, IES/BIG 2-97, ITA). The primary analysis involved pts with tumors reported to be ER positive with endpoints of BC recurrence (all, local only, contralateral BC only, distant), death with and without recurrence, any death. Subgroup analyses included progesterone receptor (PgR) status, age, tumor grade and nodal status. All p-values were 2-sided.
 Results: Highlights include: Cohort 1: 9,856 pts with 50,000 woman-years of follow-up. At 5 yrs, AI therapy was associated with an absolute 2.7% (standard error [SE] 0.7) decrease in BC recurrence (10.7% vs 13.4%, relative decrease 20% [SE 5], p=0.00004). There appeared to be greater proportional decreases in isolated local recurrence (30% [SE 10], p=0.003) and in contralateral disease (38% [SE 12], p=0.003) than in distant recurrence (12% [SE 6], p= 0.04). AIs yielded an absolute 1.0% (SE 0.5) decrease in BC mortality (5.5% vs 6.5%, relative decrease 7% [SE 7], p=0.28). Cohort 2: 9,015 pts with 33,000 woman-years of follow-up. At 6 yrs from treatment divergence (i.e., 8-9 yrs from diagnosis), AI therapy was associated with an absolute 3.5% (SE 1.1) decrease in BC recurrence (12.6% vs 16.1%, relative decrease 29% [SE 6], p<0.00001). There appeared to be greater proportional reductions in isolated local recurrence (40% [SE 13], p=0.002) and in contralateral disease (35% [SE 16], p=0.03) than in distant recurrence (24% [SE 7], p= 0.001). AIs yielded an absolute 1.6% (SE 0.8) decrease in BC mortality (6.3% vs 8.0%, relative decrease 22% [SE 9], p=0.02). Subset analyses with respect to PgR status, age, tumor grade and nodal status revealed no apparent heterogeneity between the proportional reductions in recurrence and no indication of an increase or decrease in non-breast deaths with AIs in either cohort 1 or 2.
 Conclusions: In 2 separate meta-analyses considering both the initial monotherapy setting and the switching setting after 2 to 3 yrs of Tam, AIs produced significantly lower BC recurrence rates compared with Tam. Additional follow-up data will provide important information on long-term survival. Meta-analyses can be helpful even in the era of relatively large clinical trials. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 12.

Abstract Abstract #1134 Background: The selective estrogen receptor modulators (SERMs) tamoxifen and raloxifene reduce the risk of ER+ (but not ER-) invasive breast cancers in healthy women at high risk for developing breast cancer. Aromatase inhibitors (AIs) given as adjuvant therapy to treatment-naïve or post-tamoxifen patients significantly reduce the risk of in-breast recurrences (IBRs) and contralateral breast cancers (CBCs) and are currently in clinical trials for breast cancer prevention (NCIC CTG MAP.3 and IBIS-II). It is hypothesized that SERMS inhibit promotion of ER+ breast cancer whereas AIs may reduce both ER+ and ER- breast cancer by inhibiting both tumor initiation and promotion. Little is known about the characteristics of IBRs and CBCs that arise on AI therapy. We present the ER/PR expression and clinicopathologic features of IBRs and CBCs that occurred on MA.17.
 Methods: We examined ER/PR status of IBRs and CBCs that arose on letrozole vs. placebo among women enrolled in MA.17, a placebo-controlled (PLAC) trial of letrozole (LET) following 5 years of tamoxifen in postmenopausal women with early stage breast cancer.
 Results: Seventy-one patients (pts) developed an IBR and 87 developed a CBC on trial. Consistent with results previously reported, fewer IBRs (LET 20 vs PLAC 51) and CBCs (LET 35 vs PLAC 52) were observed in the LET group. ER and PR status is currently available on 35 women with an IBR and 39 with a CBC. The majority of IBRs were ER+ in both the LET and PLAC groups (10/11 [91%] vs 18/24 [75%], respectively; p=NS) but numbers of both ER+ and – IBRs were less in LET group, suggesting that letrozole may decrease both ER+ and ER- IBRs. CBCs that arose on PLAC were more likely to be ER+ than on LET (16/22 [73%] vs 6/19 pts [32%], respectively; p=0.01), suggesting that letrozole predominantly prevents ER+ CBCs. Discordance in ER expression between primary breast cancer and IBRs among women randomized to LET vs. PLAC was observed in 1/11 [9%] and 6/24 [26%] women respectively (p=NS) and between primary breast cancer and CBCs in 12/18 pts [67%] vs. 6/21 [29%] women respectively (p=0.01). Other clinicopathologic characteristics such as grade, tumor size, PR, HER-2/neu, and nodal status of IBRs and CBCs will be presented at the meeting.
 Conclusion: Extended adjuvant endocrine therapy with letrozole results in fewer IBRs and CBCs compared with placebo as previously reported. Our data suggests that letrozole may decrease both ER+ and ER- IBRs. Letrozole appears to prevent ER+ CBCs but has little or no apparent effect on the development of ER- CBCs. These results need confirmation in the primary prevention trials of AIs.
 
 Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 1134.