Background: In postmenopausal, estrogen receptor-positive, human epidermal growth factor receptor 2 (HER2)negative early breast cancer, the risk for distant recurrence can extend beyond 5 years of adjuvant endocrine therapy. This study aims to identify genomic driver alterations associated with late distant recurrence. Patients and methods: Next generation sequencing was used to characterize driver alterations in primary tumors from a subset of 764 postmenopausal estrogen receptor-positive/HER2-negative patients from the BIG 1-98 randomized trial. Late distant recurrence events were defined as >= 5 years from time of randomization). The association of driver alterations with distant recurrence-free interval in early and late time periods was assessed using Cox regression models. Multivariable analyses were carried out to adjust for clinicopathological factors. Weighted analysis methods were used in order to correct for over-sampling of distant recurrences. Results: A total of 538 of 764 (70%) samples were successfully sequenced including 88 (63%) early and 52 (37%) late distant recurrence events after a median follow up of 8.1 years. In univariable analysis for late distant recurrence, PIK3CA mutations (58.8%) were significantly associated with reduced risk [hazard ratio (HR) 0.40, 95% confidence interval (CI) 0.20-0.82, P = 0.012], whereas amplifications on chromosome 8p11 (10.9%) (HR 4.79, 95% CI 2.30-9.97, P < 0.001) and BRCA2 mutations (2.3%) (HR 5.39, 95% CI 1.51-19.29, P = 0.010) were significantly associated with an increased risk. In multivariable analysis, only amplifications on 8p11 (P = 0.002) and BRCA2 mutations (P = 0.013) remained significant predictors. Conclusions: In estrogen receptor-positive/HER2-negative postmenopausal early breast cancer, PIK3CA mutations were associated with reduced risk of late distant recurrence, whereas amplifications on 8p11 and BRCA2 mutations were associated with increased risk of late distant recurrence. The characterization of oncogenic driver alterations may aid in refining treatment choices in the late disease setting, and help identify potential drug targets for testing in future trials.

Abstract Background and Objectives HRpos, HER2neg early breast cancer is molecularly heterogeneous. Classifications that integrate genetic, genomic and clinical-pathologic data (nodal status, tumour size, tumour differentiation, Ki-67 status, PR expression) are lacking. In a previous study we have defined three different molecular groups of invasive ductal luminal breast cancer (IDLBC) using clinical prognostic variables, gene expression, copy number variation and mutation data from the METABRIC series [1, 2] and showed that IDLBC groups (IDLBC1, -2, -3) had distinct prognostic outcomes [3, 4]. The 3 groups differ by DNA damage load, predominant lesion type, characteristic lesion patterns, oncogenic driver mechanisms and time to first recurrence. Here we evaluated the prognostic and predictive value of these groups on patients treated on the BIG1-98 study. Methods We used our predefined subgroups (IDLBC1, -2, -3) to classify patients in the BIG1-98 set [5, 6]. A weighted Cox proportional hazards regression model was used to assess the association between IDLBC subtype, treatment on OS and DRFI endpoints in the BIG1-98 study. Weighted chi-squared tests were used for categorical variables and weighted t-tests for mutational burden. All statistical computations were carried out in R v.3.2.3. Results Five hundred thirty eight breast cancers (ductal 75%, lobular 17%, 8% other) were included in this study. The DNA damage burden of IDLBC1 or -2 type tumours was significantly lower compared to IDLBC3 (mean lesion numbers: 9, 11, 18.5). IDLBC1 type tumours (48%) harboured mainly oncogenic point mutations (predominantly PIK3CA 51%, MAP3K1 27%, NCOR1 22%, FANCD2 20%). IDLBC2 (34%) and IDLBC3 type tumours (18%) were both mutated (PIK3CA 53% vs 38%, GATA3 12% vs 30%, TP53 10% vs 37%, CDH1 20% vs 8%), 8p11-12 (IDLBC3 38% vs IDLBC2 21% or IDLBC1 5%, p value < 0.001) and/or 11q13-14 amplified (IDLBC3 46% vs IDLBC2 28% or IDLBC1 2%, p value < 0.001). Consistent with the higher DNA damage load patients with IDLBC3 type tumours had significantly shorter distant recurrence free intervals (DRFI) (HR (95% CI) 2.35 [1.99; 2.78] p-value= <2e-16) and significantly shorter OS compared to IDLBC1 or -2 (HR (95% CI) 2.55 [2.2; 2.96] p-value= <2e-16). Patients with IDLBC1 or -2 type tumours seemed to have the same survival regardless of hormonal treatment assignment. However, the IDLBC3 group derived greater magnitude of benefit from Letrozole compared to Tamoxifen (HR (95% CI): DRFI 0.85 [0.66; 1.09] vs. 0.53 [0.35; 0.79] interaction p-value=0.03, OS 0.98 [0.8; 1.21] vs 0.34 [0.25; 0.47], interaction p-value=2.5e-09). Conclusions We have validated our integrated clinical-pathologic and genomic classification method on an independent dataset of HRpos, HER2neg early stage breast cancer. Our classification also suggests differing magnitude of benefit for adjuvant Letrozole benefit. Prospective clinical studies will be required to validate and corroborate this novel classification approach. References 1. Pereira, Nat Commun (2016) 2. Curtis, Nat (2012) 3. Demanse, ENBDC Workshop (2018) 4. Demanse, manuscript in preparation (2018) 5. Loi, SABCS (2016) 6. Luen, JAMA Oncology (2018) Citation Format: Demanse D, Luen SJ, Pinet D, Regan M, Kammler R, Thuerlimann B, Viale G, Colleoni MA, Loi S, Hackl W. Analysis of DRFI and OS of HR-positive, HER2-negative breast cancer patients treated on the BIG1-98 study, classified according to a novel, integrated, clinical-pathologic and genomic classification method [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P4-14-02.

Abstract Background and Objectives Endocrine therapy is one of the most effective anti-cancer treatments for HR-positive breast cancers. However, in a significant proportion of patients the disease will relapse irrespective of the treatment modality (SERM, SERD, AI). Despite the progress made with multigene signatures for molecular classification of luminal breast cancers, predicting sensitivity towards endocrine modalities remains challenging. With the help of invasive ductal luminal breast cancers (IDLBC) from the METABRIC study we have developed an integrated clinical-pathologic and genomic classification method, identifying three distinct groups (IDLBC1, -2 and -3) [1, 2]. The 3 groups differ by DNA damage load, predominant lesion type, characteristic lesion patterns and oncogenic driver mechanisms [1, 3]. Here we present characteristic gene expression profiles of IDLBCs and discuss their biologic and clinical implications. Methods Gene expression profiles based on cDNA microarray data of 1.104 IDLBCs on the METABRIC study were examined (EGAS00000000083) [3]. To identify differentially expressed genes across IDLBC1-3 groups (52%, 23%, 25%), we compiled a gene list (N=2034) from 34 gene signatures reported to be predictive of ER, PR, GATA3, FOXA1 and/or AR signalling. A multiclass significance analysis of microarrays with 100 iterations and a false discovery rate of 0.05 was used to extract differentially expressed genes. Results Of 2034 steroid receptor signalling associated genes 36 exhibited differential expression in the IDLBC1/2 and -3 groups. The IDLBC1/2 specific set included 2 ER-associated genes (WLS, SPARCL1), 1 FOXA1-associated gene (NTN4) and 2 PR-associated genes (ZBTB16, CRY2). The IDLBC3 overexpressed set included 31 genes, amongst them regulators of mitotic fidelity, as well as effectors of centromere and kinetochore functioning. Eight of these genes have been reported as ER-associated (POLR2H, ECE2, SCL7A5, KIF11, FAM83D, GPSM2, MASTL, POLE2). Additional IDLBC3 hallmarks included high proliferation activity, defect apoptosis, epithelial-mesenchymal-transformation (EMT), BRCA-ness, multiple-drug resistance (MDR) and early relapse. The estimated relapse-free-survival rate at 5 years in the IDLBC3 group was 77% [95%CI 72; 82], significantly lower than in IDLBC1/2 tumours (91% [95%CI 89; 93]) and comparable to triple negative IDLBCs on the METABRIC study (70% [95%CI: 65; 76]). Conclusions We have shown that our integrated classification method has the potential to segregate early relapsing luminal breast cancers (IDLBC3) from tumours with favourable prognosis (IDLBC1/2). The data presented here and in the abstract by Demanse et al [4] provide also an intuitive hypothesis why IDLBC3 tumours are sensitive towards Letrozole but not Tamoxifen. Prospective clinical studies will be required to validate this hypothesis and to demonstrate the value of the IDLBC classification approach. References 1. Demanse, manuscript in preparation (2018) 2. Pereira, Nat Commun (2016) 3. Curtis, Nat (2012) 4. Demanse, SABCS, abstract submitted (2018) 5. Loi SABCS, Oral Presentation (2016) 6. Luen, JAMA Oncology (2018) Citation Format: Hackl W, Pinet D, Demanse D. A novel integrated, clinical-pathologic and genomic classification method segregates early from late relapsing invasive ductal luminal breast cancers (IDLBC) of the METABRIC study [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-08-06.

Abstract Background Next generation sequencing (NGS) has revealed that ER+/HER2- BCs have diverse somatic copy number and mutation profiles but thus far the clinical relevance of such findings is unknown. We characterized the molecular alterations in post-menopausal primary BC patients treated in the BIG 1-98 adjuvant letrozole and tamoxifen study and evaluated their associations with prognosis. Materials and Methods NGS was used to genotype DNA from archival primary tumor blocks for 286 cancer-related genes. From 2706 available eligible samples (confirmed ER+, excluding HER2-positivity or neoadjuvant treatment, adequate DNA quality/quantity), a case-cohort design selected 764 samples: all distant relapses and a stratified sampling of non-relapses after 8yr median follow-up. Mutation prevalence and associations with clinicopathological factors (CP) as well as distant recurrence-free interval (DRFS) were analyzed using weighted tests and Cox regression models, with sampling weights to represent the ER+/HER2-negative trial population. Multivariable analyses were adjusted for tumor size, nodal status, grade and age. Results NGS data was available from 538/764 samples (70%), there were 140 (26%) distant relapses. Median sequencing depth was 483x. There was a mean of 11 mutations (1-46) per sample, with 25% having 13 or more mutations and no tumors without a mutation. Overall 28 genes were altered at a frequency of >10%. The most commonly mutated genes were PIK3CA (49.3%), NCOR1 (27.2%), MAP3K1 (23.8%) TP53 (16.6%), CCND1 (17.8%) and GATA3 (17.1%). Alterations that were significantly associated with both Luminal B (Ki67 >14%) and grade 3 included TP53 mutations (p<0.001), FGFR1 (p=0.001) and MYC amplifications (p=0.004). Gene alterations that were significantly associated with shorter DRFS included TP53 (HR:2.16), ARID1A (HR:2.43), CHEK2 (HR:2.54), BRCA2 (HR:1.93), PTEN (HR:2.03), CCND1 (HR:1.82) and FGFR1 (HR:1.78). PIK3CA was significantly associated with lower risk of distant relapse (HR:0.64; 0.43-0.97). Increasing number of total mutations was significantly associated with shorter DRFS (HR:1.04; 95% CI 1.01-1.07; p=0.006). In the multivariable model adjusted for CP factors, ARID1A, BRCA2, CCND1, CHEK2 and PTEN remained independent for shorter DRFS. Greater than 90% of PIK3CA mutations co-existed with another alteration, most common being NCOR1 (29%), MAP3K1 (24%), CDH1 (16%), GATA3 (16%), TP53 (14%) and CCND1 (13%). Patients with a PIK3CA mutation had greater benefit with letrozole over tamoxifen monotherapy (HR 0.32; 0.13-0.8) than those without (HR:0.70; 0.33-1.48) (Pint=0.06). This effect was strongest in the subgroup of PIK3CA mutant patients who were CCND1 and TP53 wild-type (HR:0.24, 0.12-0.48; Pint=0.02) with only 1% relapsing at 5 years. Conclusion: For the first time, we report the prognostic relevance of oncogenic mutations in ER+/HER2- postmenopausal early-stage BCs from a clinical trial. Tumors with PIK3CA mutations derived greater benefit from letrozole over tamoxifen monotherapy, especially if wild-type for CCND1 and TP53. These findings could significantly improve prognostic risk classification and guide future clinical trials of targeted therapies in ER+/HER2- BCs. Citation Format: Loi S, Asher R, Lee CK, Luen S, Savas P, Kammler R, Dell'Orto P, Blasi OM, Demanse D, JeBailley L, Dolan S, Hackl W, Thuerlimann B, Viale G, Regan M, Colleoni MA. Clinical implications of somatic mutations in post-menopausal early-stage estrogen receptor (ER)-positive HER2-negative breast cancer (BC): Results from the BIG 1-98 study [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr S1-10.

BACKGROUND This phase I dose-escalation study investigated the maximum tolerated dose (MTD), safety, pharmacokinetics, pharmacodynamics (PDs), and preliminary antitumor activity of BGT226, a potent, oral dual phosphatidylinositol-3-kinase (PI3K)/mammalian target of rapamycin inhibitor. PATIENTS AND METHODS Fifty-seven patients with advanced solid tumors received BGT226 2.5-125 mg/day three times weekly (TIW). Dose escalation was guided by an adaptive Bayesian logistic regression model with overdose control. Assessments included response per RECIST, [18F]-fluorodeoxyglucose uptake, and phosphorylated-S6 in skin and paired tumor samples. RESULTS Three patients (125 mg cohort) had dose-limiting toxic effects (grade 3 nausea/vomiting, diarrhea). BGT226-related adverse events included nausea (68%), diarrhea (61%), vomiting (49%), and fatigue (19%). BGT226 demonstrated rapid absorption, variable systemic exposure, and a median half-life of 6-9 h. Seventeen patients (30%) had stable disease (SD) as best response. Nine patients had SD for ≥16 weeks. Thirty patients (53%) achieved stable metabolic disease as assessed by [18F]-fluorodeoxyglucose-positron emission tomography; however, no correlation between metabolic response and tumor shrinkage according to computed tomography was observed. PD changes suggested PI3K pathway inhibition but were inconsistent. CONCLUSIONS The MTD of BGT226 was 125 mg/day TIW, and the clinically recommended dose was 100 mg/day TIW. Limited preliminary antitumor activity and inconsistent target inhibition were observed, potentially due to low systemic exposure.