Erlander, MG

550 Background: Ductal carcinoma in situ (DCIS) is a heterogeneous group of non-invasive cancers with varying propensity for recurrence. There are no validated markers that identify patients at risk for invasive recurrence following lumpectomy. HOXB13/IL17BR (H/I) is expressed in both DCIS and invasive cancer (Ma Cancer Cell. 2004) and associated with recurrence in stage I cancer. A 5-gene molecular grade index (MGI) distinguishes outcome in grade II disease (Ma Clin Cancer Res. 2008). We sought to determine whether H/I or MGI was associated with invasive recurrence in resected DCIS. Methods: We identified patients at Mayo Rochester who underwent lumpectomy ± radiation for DCIS between 1988 and 2001 and performed a nested case/control study. Cases were patients with ipsilateral invasive recurrence, matched to controls based on age, surgery date, length of follow-up, and adjuvant radiation. DCIS was macrodissected from paraffin sections and RNA extracted to obtain H/I and MGI RT-PCR profiles. The association of H/I and MGI with case-control status was assessed with conditional logistic regression; results reported are odds ratios (OR) with 95% confidence intervals (CI). The association of H/I and MGI as continuous variables with nuclear grade (low, intermediate, high) was estimated with Spearman's rank correlation. Results: 427 patients underwent lumpectomy for DCIS, and 33 were identified as possible cases. Histologic review excluded 8 (non-sufficient tissue or DCIS recurrence without invasion). 25 cases were matched to 48 controls (2 controls for 23 cases, 1 control for 2 cases). Both H/I (r = 0.43, p = 0.0001) and MGI (r = 0.41, p = 0.0004) were significantly correlated with nuclear grade. Positive H/I was associated with trend towards higher recurrence (OR 1.9; p = 0.22) that was most pronounced in cases/controls (15/28) treated with lumpectomy alone (OR 3.3; CI: 0.8–13.2, p = 0.09). Similarly, positive MGI was associated with a trend towards higher recurrence risk in patients treated with surgery alone (OR 2.0; CI: 0.6–6.4, p = 0.27). Conclusions: These data suggest that both HOXB13/IL17BR and MGI can identify a subset of patients with surgically resected DCIS who are at risk for invasive recurrence. Further studies in larger cohorts are needed to confirm these findings. [Table: see text]

Abstract Abstract #57 Background: ABCSG 8 is the only prospective trial examining the value of switching patients (pts) to an aromatase inhibitor after tamoxifen (Tam) in which randomization occurred prior to therapy. This trial demonstrated that 2 years (yrs) of Tam followed by 3 yrs of anastrozole (TAM to A) was superior to 5 yrs of Tam (Jakesz SABC 2005). Tam is metabolically activated by CYP2D6 to endoxifen and pts with impaired CYP2D6 metabolism have a higher risk of recurrence (Goetz JCO 2005). The HOXB13/IL17BR (H/I) index (Ma Cancer Cell 2004) is associated with Tam resistance while a 5-gene molecular grade index (MGI) distinguishes outcome in grade II tumors. H/I and MGI are complementary prognostic factors that define low (low MGI), intermediate (low H/I, high MGI) and high risk groups (high H/I and MGI) (Ma Clin Cancer Res 2008). We sought to validate these biomarkers in ABCSG Trial 8, and hypothesized that they may predict the benefit of sequencing TAM to A versus Tam alone.
 Methods: From 1996-2004, ABCSG 8 randomized 3,901 pts with ER positive breast cancer. The primary endpoint was event-free survival. Assuming 40% of pts had reduced CYP2D6 metabolism, we designed a nested case control study in which 110 Tam treated cases matched to 2 controls would yield 80% power to detect a relative risk of 2.3 for pts with reduced CYP2D6 metabolism. Additionally, we sought to perform an exploratory analysis of the benefit of switching from Tam to A in pts with impaired vs non-impaired CYP2D6 metabolism. DNA and RNA derived from paraffin embedded sections were used to genotype for CYP2D6 and to obtain H/I and MGI RT-PCR profiles.
 Findings: Thus far, CYP2D6 genotyping is complete in 345 pts corresponding to 53 cases/106 controls (Tam) and 62 cases/124 controls (Tam to A). DNA and RNA yields were excellent with a CYP2D6 genotype call rate of >95%. CYP2D6 allele frequencies were as expected: *3 (.01), *4, (.23), *6 (.01), *10 (.02), *17 (.00), and *41 (.10). 44% of pts had reduced CYP2D6 metabolism (at least one CYP2D6 null allele or intermediate metabolizer) and 8% were CYP2D6 poor metabolizers. For H/I and MGI, 340 pts were profiled with 64%, 20% and 16% in the low, intermediate, and high risk groups, respectively. Tissue sectioning/genotyping for additional cases/controls is ongoing.
 Discussion: This is the first retrospective analysis of a large Tam versus Tam to A trial to evaluate the association between 1) CYP2D6, HOXB13/IL17BR, and MGI with TAM treatment outcome and 2) the predictive benefit of these biomarkers in pts randomized to switch from Tam to AI versus Tam alone. Final results will be presented at the meeting. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 57.