Analytical validation of HER2DX genomic test for early-stage HER2-positive breast cancer

BACKGROUND

HER2DX, a multianalyte genomic test, has been clinically validated to predict breast cancer recurrence risk (relapse risk score), the probability of achieving pathological complete response post-neoadjuvant therapy (pCR likelihood score), and individual ERBB2 messenger RNA (mRNA) expression levels in patients with early-stage human epidermal growth factor receptor 2 (HER2)-positive breast cancer. This study delves into the comprehensive analysis of HER2DX's analytical performance.

MATERIALS AND METHODS

Precision and reproducibility of HER2DX risk, pCR, and ERBB2 mRNA scores were assessed within and between laboratories using formalin-fixed paraffin-embedded (FFPE) tumor tissues and purified RNA. Robustness was appraised by analyzing the impact of tumor cell content and protocol variations including different instruments, reagent lots, and different RNA extraction kits. Variability was evaluated across intratumor biopsies and genomic platforms [RNA sequencing (RNAseq) versus nCounter], and according to protocol variations.

RESULTS

Precision analysis of 10 FFPE tumor samples yielded a maximal standard error of 0.94 across HER2DX scores (1-99 scale). High reproducibility of HER2DX scores across 29 FFPE tumors and 20 RNAs between laboratories was evident (correlation coefficients >0.98). The probability of identifying score differences >5 units was ≤5.2%. No significant variability emerged based on platform instruments, reagent lots, RNA extraction kits, or TagSet thaw/freeze cycles. Moreover, HER2DX displayed robustness at low tumor cell content (10%). Intratumor variability across 212 biopsies (106 tumors) was <4.0%. Concordance between HER2DX scores from 30 RNAs on RNAseq and nCounter platforms exceeded 90.0% (Cohen's κ coefficients >0.80).

CONCLUSIONS

The HER2DX assay is highly reproducible and robust for the quantification of recurrence risk, pCR likelihood, and ERBB2 mRNA expression in early-stage HER2-positive breast cancer.

Prognostic value of HER2DX in early-stage HER2-positive breast cancer: a comprehensive analysis of 757 patients in the Sweden Cancerome Analysis Network-Breast dataset (SCAN-B)

BACKGROUND

The HER2DX risk-score has undergone rigorous validation in prior investigations involving patients with early-stage human epidermal growth factor receptor 2 (HER2)-positive (HER2+) breast cancer. In this study, we present the outcomes of the HER2DX risk-score within the most recent release of the Sweden Cancerome Analysis Network-Breast (SCAN-B) HER2+ cohort. This updated examination benefits from a larger patient sample, an extended follow-up duration, and detailed treatment information.

MATERIALS AND METHODS

Clinical and RNAseq data from the SCAN-B dataset were retrieved from Gene Expression Omnibus (GSE81538). Among the 6600 patients, 819 had HER2+ breast cancer, with 757 individuals with research-based HER2DX risk-scores and corresponding survival outcomes. The HER2DX risk-score was evaluated (i) as a continuous variable and (ii) using predefined cut-offs. The primary endpoint for this study was overall survival (OS). The Kaplan-Meier method and Cox models were used to estimate OS and a multistate model with four states was fitted to better characterize patients' follow-up.

RESULTS

The median follow-up time was 7.5 years (n = 757). The most common systemic therapy was chemotherapy with trastuzumab (82.0%) and most tumors were classified as T1-T2 (97.1%). The HER2DX risk-score as a continuous variable was significantly associated with OS after adjustment for clinical variables and treatment regimen [hazard ratios (HR) per 10-unit increment = 1.31, 95% confidence interval (CI) 1.13-1.51, P < 0.001] as well as within predefined risk groups (high versus low; HR = 2.57, 95% CI 1.36-4.85, P < 0.001). Patients classified as HER2DX high-risk also had higher risk of (i) breast cancer recurrence and (ii) death without previous recurrence. Within the subgroup of HER2+ T1N0 tumors (n = 297), those classified as high-risk demonstrated inferior OS compared to low-risk tumors (7-year OS 77.8% versus 96.8%, P < 0.001). The HER2DX mRNA ERBB2 score was associated with clinical HER2 status (area under the receiver operating characteristic curve = 0.91).

CONCLUSIONS

In patients with early-stage HER2+ breast cancer, HER2DX risk-score provides prognostic information beyond clinicopathological variables, including treatment regimen with or without trastuzumab.

Association of HER2DX with pathological complete response and survival outcomes in HER2-positive breast cancer

BACKGROUND

The HER2DX genomic test predicts pathological complete response (pCR) and survival outcome in early-stage HER2-positive (HER2+) breast cancer. Here, we evaluated the association of HER2DX scores with (i) pCR according to hormone receptor status and various treatment regimens, and (ii) survival outcome according to pCR status.

MATERIALS AND METHODS

Seven neoadjuvant cohorts with HER2DX and clinical individual patient data were evaluated (DAPHNe, GOM-HGUGM-2018-05, CALGB-40601, ISPY-2, BiOnHER, NEOHER and PAMELA). All patients were treated with neoadjuvant trastuzumab (n = 765) in combination with pertuzumab (n = 328), lapatinib (n = 187) or without a second anti-HER2 drug (n = 250). Event-free survival (EFS) and overall survival (OS) outcomes were available in a combined series of 268 patients (i.e. NEOHER and PAMELA) with a pCR (n = 118) and without a pCR (n = 150). Cox models were adjusted to evaluate whether HER2DX can identify patients with low or high risk beyond pCR status.

RESULTS

HER2DX pCR score was significantly associated with pCR in all patients [odds ratio (OR) per 10-unit increase = 1.59, 95% confidence interval 1.43-1.77; area under the ROC curve = 0.75], with or without dual HER2 blockade. A statistically significant increase in pCR rate due to dual HER2 blockade over trastuzumab-only was observed in HER2DX pCR-high tumors treated with chemotherapy (OR = 2.36 (1.09-5.42). A statistically significant increase in pCR rate due to multi-agent chemotherapy over a single taxane was observed in HER2DX pCR-medium tumors treated with dual HER2 blockade (OR = 3.11, 1.54-6.49). The pCR rates in HER2DX pCR-low tumors were ≤30.0% regardless of treatment administered. After adjusting by pCR status, patients identified as HER2DX low-risk had better EFS (P < 0.001) and OS (P = 0.006) compared with patients with HER2DX high-risk.

CONCLUSIONS

HER2DX pCR score and risk score might help identify ideal candidates to receive neoadjuvant dual HER2 blockade in combination with a single taxane in early-stage HER2+ breast cancer.

Patritumab Deruxtecan in Untreated Hormone Receptor-Positive/HER2-Negative Early Breast Cancer: Final Results from Part A of the Window-of-Opportunity SOLTI TOT-HER3 Pre-Operative Study

BACKGROUND

Patritumab deruxtecan (HER3-DXd) is a HER3-directed antibody-drug conjugate composed of a fully human anti-HER3 monoclonal antibody (patritumab) covalently linked to a topoisomerase I inhibitor payload via a stable, tumor-selective, tetrapeptide-based cleavable linker. TOT-HER3 is a window-of-opportunity study designed to assess the biological activity, measured by CelTIL score [ = -.8 × tumor cellularity (in %) + 1.3 × TILs (in %)], and clinical activity of HER3-DXd during short-term (21 days) pre-operative treatment in patients with primary operable HER2-negative early breast cancer.

PATIENTS AND METHODS

Patients with previously untreated hormone receptor (HR)-positive/HER2-negative tumors were allocated to one of four cohorts according to baseline ERBB3 mRNA expression. All patients received one dose of HER3-DXd 6.4 mg/kg. The primary objective was to evaluate change from baseline in CelTIL score.

RESULTS

Seventy-seven patients were evaluated for efficacy. A significant change in CelTIL score was observed, with a median increase from baseline of 3.5 (interquartile range, -3.8 to 12.7; P=.003). Among patients evaluable for clinical response (n=62), an overall response rate of 45% was observed (tumor measurement by caliper), with a trend toward an increase in CelTIL score among responders compared with non-responders (mean difference, +11.9 vs +1.9). Change in CelTIL score was independent of baseline ERBB3 mRNA and HER3 protein levels. Genomic changes occurred, including switching toward a less proliferative tumor phenotype based on PAM50 subtypes, suppression of cell proliferation genes, and induction of genes associated with immunity. Treatment-emergent adverse events were observed in 96% of patients (14% grade ≥3); most common were nausea, fatigue, alopecia, diarrhea, vomiting, abdominal pain, and neutrophil count decrease.

CONCLUSIONS

A single dose of HER3-DXd was associated with clinical response, increased immune infiltration, suppression of proliferation in HR-positive/HER2-negative early breast cancer, and a tolerable safety profile consistent with previously reported results. These findings support further study of HER3-DXd in early breast cancer.

Prognostic value of intrinsic subtypes in hormone-receptor-positive metastatic breast cancer: systematic review and meta-analysis

BACKGROUND

In hormone receptor-positive (HoR+) breast cancer (BC), gene expression analysis identifies luminal A (LumA), luminal B (LumB), human epidermal growth factor receptor 2 (HER2)-enriched (HER2-E), basal-like (BL) intrinsic subtypes and a normal-like group. This classification has an established prognostic value in early-stage HoR+ BC. Here, we carried out a trial-level meta-analysis to determine the prognostic ability of subtypes in metastatic BC (MBC).

MATERIALS AND METHODS

We systematically reviewed all the available prospective phase II/III trials in HoR+ MBC where subtype was assessed. The primary endpoint was progression-free survival (PFS)/time to progression (TTP) of the LumA subtype compared to non-LumA. Secondary endpoints were PFS/TTP of each individual subtype, according to treatment, menopausal and HER2 status and overall survival (OS). The random-effect model was applied, and heterogeneity assessed through Cochran's Q and I². Threshold for significance was set at P < 0.05. The study was registered in PROSPERO (ID: CRD42021255769).

RESULTS

Seven studies were included (2536 patients). Non-LumA represented 55.2% and was associated with worse PFS/TTP than LumA [hazard ratio (HR) 1.77, P < 0.001, I² = 61%], independently of clinical HER2 status [P_{subgroup difference} (P_sub) = 0.16], systemic treatment (P_sub = 0.96) and menopausal status (P_sub = 0.12). Non-LumA tumors also showed worse OS (HR 2.00, P < 0.001, I² = 65%), with significantly different outcomes for LumB (PFS/TTP HR 1.46; OS HR 1.41), HER2-E (PFS/TTP HR 2.39; OS HR 2.08) and BL (PFS/TTP HR 2.67; OS HR 3.26), separately (PFS/TTP P_sub = 0.01; OS P_sub = 0.005). Sensitivity analyses supported the main result. No publication bias was observed.

CONCLUSIONS

In HoR+ MBC, non-LumA disease is associated with poorer PFS/TTP and OS than LumA, independently of HER2, treatment and menopausal status. Future trials in HoR+ MBC should consider this clinically relevant biological classification.

Identification of cell surface targets for CAR-T cell therapies and antibody-drug conjugates in breast cancer

BACKGROUND

Two promising therapeutic strategies in oncology are chimeric antigen receptor-T cell (CAR-T) therapies and antibody-drug conjugates (ADCs). To be effective and safe, these immunotherapies require surface antigens to be sufficiently expressed in tumors and less or not expressed in normal tissues. To identify new targets for ADCs and CAR-T specifically targeting breast cancer (BC) molecular and pathology-based subtypes, we propose a novel in silico strategy based on multiple publicly available datasets and provide a comprehensive explanation of the workflow for a further implementation.

METHODS

We carried out differential gene expression analyses on The Cancer Genome Atlas BC RNA-sequencing data to identify BC subtype-specific upregulated genes. To fully explain the proposed target-discovering methodology, as proof of concept, we selected the 200 most upregulated genes for each subtype and undertook a comprehensive analysis of their protein expression in BC and normal tissues through several publicly available databases to identify the potentially safest and viable targets.

RESULTS

We identified 36 potentially suitable and subtype-specific tumor surface antigens (TSAs), including fibroblast growth factor receptor-4 (FGFR4), carcinoembryonic antigen-related cell adhesion molecule 6 (CEACAM6), GDNF family receptor alpha 1 (GFRA1), integrin beta-6 (ITGB6) and ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1). We also identified 63 potential TSA pairs that might be appropriate for co-targeting strategies. Finally, we validated subtype specificity in a cohort of our patients, multiple BC cell lines and the METABRIC database.

CONCLUSIONS

Overall, our in silico analysis provides a framework to identify novel and specific TSAs for the development of new CAR-T and antibody-based therapies in BC.

Immune microenvironment characterisation and dynamics during anti-HER2-based neoadjuvant treatment in HER2-positive breast cancer

Despite their recognised role in HER2-positive (HER2+) breast cancer (BC), the composition, localisation and functional orientation of immune cells within tumour microenvironment, as well as its dynamics during anti-HER2 treatment, is largely unknown. We here investigate changes in tumour-immune contexture, as assessed by stromal tumour-infiltrating lymphocytes (sTILs) and by multiplexed spatial cellular phenotyping, during treatment with lapatinib-trastuzumab in HER2+ BC patients (PAMELA trial). Moreover, we evaluate the relationship of tumour-immune contexture with hormone receptor status, intrinsic subtype and immune-related gene expression. sTIL levels increase after 2 weeks of HER2 blockade in HR-negative disease and HER2-enriched subtype. This is linked to a concomitant increase in cell density of all four immune subpopulations (CD3+, CD4+, CD8+, Foxp3+). Moreover, immune contexture analysis showed that immune cells spatially interacting with tumour cells have the strongest association with response to anti-HER2 treatment. Subsequently, sTILs consistently decrease at the surgery in patients achieving pathologic complete response, whereas most residual tumours at surgery remain inflamed, possibly reflecting a progressive loss of function of T cells. Understanding the features of the resulting tumour immunosuppressive microenvironment has crucial implications for the design of new strategies to de-escalate or escalate systemic therapy in early-stage HER2+ BC.

De-escalated therapy for HR+/HER2+ breast cancer patients with Ki67 response after 2-week letrozole: results of the PerELISA neoadjuvant study

BACKGROUND

In human epidermal growth factor receptor 2 (HER2+) breast cancers, neoadjuvant trials of chemotherapy plus anti-HER2 treatment consistently showed lower pathologic complete response (pCR) rates in hormone receptor (HR) positive versus negative tumors. The PerELISA study was aimed to evaluate the efficacy of a de-escalated, chemotherapy-free neoadjuvant regimen in HR+/HER2+ breast cancer patients selected on the basis of Ki67 inhibition after 2-week letrozole.

PATIENTS AND METHODS

PerELISA is a phase II, multicentric study for postmenopausal patients with HR+/HER2+ operable breast cancer. Patients received 2-week letrozole, and then underwent re-biopsy for Ki67 evaluation. Patients classified as molecular responders (Ki67 relative reduction >20% from baseline) continued letrozole and started trastuzumab-pertuzumab for five cycles. Patients classified as molecular non-responders started weekly paclitaxel for 13 weeks combined with trastuzumab-pertuzumab. Primary aim was breast and axillary pCR. According to a two-stage Simon's design, to reject the null hypothesis, at least 8/43 pCR had to be documented.

RESULTS

Sixty-four patients were enrolled, 44 were classified as molecular responders. All these patients completed the assigned treatment with letrozole-trastuzumab-pertuzumab and underwent surgery. A pCR was observed in 9/44 cases (20.5%, 95% confidence interval 11.1% to 34.5%). Among molecular non-responders, 16/17 completed treatment and underwent surgery, with pCR observed in 81.3% of the cases. PAM50 intrinsic subtype was significantly associated with Ki67 response and pCR. Among molecular responders, the pCR rate was significantly higher in HER2-enriched than in other subtypes (45.5% versus 13.8%, P = 0.042).

CONCLUSIONS

The primary end point of the study was met, by reaching the pre-specified pCRs. In patients selected using Ki67 reduction after short-term letrozole exposure, a meaningful pCR rate can be achieved without chemotherapy. PAM50 intrinsic subtyping further refines our ability to identify a subset of patients for whom chemotherapy might be spared.

EUDRACT NUMBER

2013-002662-40.

CLINICALTRIALS.GOV IDENTIFIER

NCT02411344.

Association between PD1 mRNA and response to anti-PD1 monotherapy across multiple cancer types

Background

We hypothesized that the abundance of PD1 mRNA in tumor samples might explain the differences in overall response rates (ORR) observed following anti-PD1 monotherapy across cancer types.

Patients and methods

RNASeqv2 data from 10 078 tumor samples representing 34 different cancer types was analyzed from TCGA. Eighteen immune-related gene signatures and 547 immune-related genes, including PD1, were explored. Correlations between each gene/signature and ORRs reported in the literature following anti-PD1 monotherapy were calculated. To translate the in silico findings to the clinical setting, we analyzed the expression of PD1 mRNA using the nCounter platform in 773 formalin-fixed paraffin embedded (FFPE) tumor samples across 17 cancer types. To test the direct relationship between PD1 mRNA, PDL1 immunohistochemistry (IHC), stromal tumor-infiltrating lymphocytes (sTILs) and ORR, we evaluated an independent FFPE-based dataset of 117 patients with advanced disease treated with anti-PD1 monotherapy.

Results

In pan-cancer TCGA, PD1 mRNA expression was found strongly correlated (r > 0.80) with CD8 T-cell genes and signatures and the proportion of PD1 mRNA-high tumors (80th percentile) within a given cancer type was variable (0%-84%). Strikingly, the PD1-high proportions across cancer types were found strongly correlated (r = 0.91) with the ORR following anti-PD1 monotherapy reported in the literature. Lower correlations were found with other immune-related genes/signatures, including PDL1. Using the same population-based cutoff (80th percentile), similar proportions of PD1-high disease in a given cancer type were identified in our in-house 773 tumor dataset as compared with TCGA. Finally, the pre-established PD1 mRNA FFPE-based cutoff was found significantly associated with anti-PD1 response in 117 patients with advanced disease (PD1-high 51.5%, PD1-intermediate 26.6% and PD1-low 15.0%; odds ratio between PD1-high and PD1-intermediate/low = 8.31; P < 0.001). In this same dataset, PDL1 tumor expression by IHC or percentage of sTILs was not found associated with response.

Conclusions

Our study provides a clinically applicable assay that links PD1 mRNA abundance, activated CD8 T-cells and anti-PD1 efficacy.

A predictive model of pathologic response based on tumor cellularity and tumor-infiltrating lymphocytes (CelTIL) in HER2-positive breast cancer treated with chemo-free dual HER2 blockade

Background

The presence of stromal tumor-infiltrating lymphocytes (TILs) is associated with increased pathologic complete response (pCR) and improved outcomes in HER2-positive early-breast cancer (BC) treated with anti-HER2-based chemotherapy. In the absence of chemotherapy, the association of TILs with pCR following anti-HER2 therapy-only is largely unknown.

Patients and methods

The PAMELA neoadjuvant trial treated 151 women with HER2-positive BC with lapatinib and trastuzumab [and hormonal therapy if hormone receptor (HR)-positive] for 18 weeks. Percentage of TILs and tumor cellularity were determined at baseline (N = 148) and at day 15 (D15) of treatment (N = 134). Associations of TILs and tumor cellularity with pCR in the breast were evaluated. A combined score based on tumor cellularity and TILs (CelTIL) measured at D15 was derived in PAMELA, and validated in D15 samples from 65 patients with HER2-positive disease recruited in the LPT109096 neoadjuvant trial, where anti-HER2 therapy-only was administer for 2 weeks, then standard chemotherapy was added for 24 weeks.

Results

In PAMELA, baseline and D15 TILs were significantly associated with pCR in univariate analysis. In multivariable analysis, D15 TILs, but not baseline TILs, were significantly associated with pCR. At D15, TILs and tumor cellularity were found independently associated with pCR. A combined score (CelTIL) taking into account both variables was derived. CelTIL at D15 as a continuous variable was significantly associated with pCR, and patients with CelTIL-low and CelTIL-high scores had a pCR rate of 0% and 33%, respectively. In LPT109096, CelTIL at D15 was found associated with pCR both as a continuous variable and as group categories using a pre-defined cut-off (75.0% versus 33.3%).

Conclusions

On-treatment TILs, but not baseline TILs, are independently associated with response following anti-HER2 therapy-only. A combined score of TILs and tumor cellularity measured at D15 provides independent predictive information upon completion of neoadjuvant anti-HER2-based therapy.

Clinical trial number

NCT01973660.

Abstract P6-17-07: Gene signatures and subtype changes during HER2 dual blockade in PAM50 HER2-enriched HER2-positive breast cancer

Background: HER2-positive (HER2+) breast cancer (BC) is composed of 4 molecular subtypes: Luminal A and B, HER2-enriched (HER2-E) and Basal-like. Among them, the HER2-E is highly sensitive to anti-HER2 treatment. However, ˜60% of HER2-E tumors do not achieve a pathological complete response (pCR) following neoadjuvant dual HER2 blockade without chemotherapy. Here, we aimed to better understand the molecular changes of the HER2-E subtype during anti-HER2 treatment.

Methods: Gene expression was evaluated in 101 patients with HER2-E tumors from the PAMELA neoadjuvant phase II trial (Lancet Oncol 2017). Briefly, women with HER2+ BC were treated with lapatinib and trastuzumab (and hormonal therapy if hormone receptor [HR]-positive) for 18 weeks. The median time between the last dose of treatment and surgery was 35 days (range=213; interquartile range=16). Expression of the PAM50 genes and 6 PAM50 signatures (Luminal A, Luminal B, HER2-E, Basal-like, normal-like and the PAM50 proliferation score) were determined using the nCounter platform at baseline (n=101), after 2 weeks of treatment (n=96) and in residual tumors (non-pCR) at surgery (n=57). Same analyses were done in 2 HER2+/HER2-E cell line models (BT474 [HR+] and SKBR3 [HR-]) following in vitro treatment with trastuzumab in combination with lapatinib. Biological changes between 2 time-points were determined by paired t-tests with a false discovery rate (FDR) &lt;5%.

Results: After 2 weeks of treatment, 85.7% and 94.6% of the 56 genes/signatures were found differentially expressed (FDR&lt;5%) in HER2-E/HR+ (n=35) and HER2-E/HR- (n=61) tumors, respectively. The two gene lists were highly correlated (correlation coefficient=0.93). Overall, a significant relative increase in Luminal A and normal-like signature scores, and a relative decrease in proliferation, HER2-E and Luminal B signature scores, were observed between baseline and week 2. Interestingly, a PAM50 subtype switch to Luminal A was observed in 31.6% and 4.8% of HER2-E/HR+ and HER2-E/HR- tumors. In BT474 and SKBR3, all genes/signatures were also found differentially expressed (FDR&lt;5%) following 72h of dual HER2 blockade. The in vitro findings recapitulated the in vivo findings in 80-86% of the genes/signatures. Similar to tumors, a switch to a Luminal A subtype following dual HER2 blockade was observed in BT474 but not in SKBR3. Finally, 92.9% of the 56 genes/signatures were found differentially expressed (FDR&lt;5%) in residual tumors at surgery compared to week 2. Contrary to the findings in the first 2 weeks of treatment, a general rebound effect in gene expression was observed between week 2 and surgery. Similarly, a rebound effect was observed in 60% of the genes/signatures in BT474 after removing anti-HER2 therapy for 72h, leading to a subtype switch from Luminal A back to HER2-E.

Conclusions: Dual HER2 blockade in the HER2-E subtype induces large biological changes that lead to a more low-proliferative Luminal A phenotype both in tumors and in vitro models, especially in HER2-E/HR+ disease. These phenotypic changes are reversible upon stopping anti-HER2 treatment. This finding supports the use of maintenance anti-HER2 treatment +/- endocrine therapy (if HR+) in advanced HER2+ BC.

Citation Format: Brasó-Maristany F, Griguolo G, Llombart-Cussac A, Pascual T, Paré L, Bermejo B, Oliveira M, Morales S, Martinez N, Vidal M, Pernas S, Lopez R, Muñoz M, Galvan P, Garau I, Manso L, Alarcón J, Martínez E, Villagrasa P, Cortés J, Prat A. Gene signatures and subtype changes during HER2 dual blockade in PAM50 HER2-enriched HER2-positive breast cancer [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 P6-17-07.

Abstract GS5-06: No survival benefit of chemotherapy escalation in patients with pCR and “high-immune” triple-negative early breast cancer in the neoadjuvant WSG-ADAPT-TN trial

Background:Immune markers such as tumor infiltrating lymphocytes (TILs), CD8, PDL1, PD1 and other protein or mRNA-based genomic markers have been identified as prognostic / predictive in TNBC regarding survival / chemotherapy (CTx) efficacy.

In the adjuvant WSG-PlanB trial, patients with high TILs and/or CD8 by mRNA had excellent outcome, irrespective of anthracycline use; in the neoadjuvant ADAPT-TN trial, high PDL1, PD1 and CD8 and/or TILs were predictive for pCR. Still, optimal markers for potential treatment de-escalation have yet to be determined. Here, we analyse for the first time impact of immune mRNA-based markers and TIL's as prognostic and predictive survival markers.

Methods: TNBC patients (ER/PR&lt;1%, HER2-,) were randomized to neoadjuvant 4x nab-paclitaxel 125 mg/m2/gemcitabine 1000 mg/m2 d1/8 q3w (gem arm) or 4x nab-paclitaxel 125 mg/m2/carboplatin AUC2 day 1/8 3-weekly (q3w) (carbo arm). Primary endpoint of WSG-ADAPT-TN was pCR (ypT0/is/ypN0); secondary endpoints included translational analyses, e.g., TILs or expression of 119 genes by nCounter platform. Standard adjuvant chemotherapy (4xEC) was optional (not randomized) in patients achieving pCR after 12 weeks. According to protocol, 1st safety survival analysis was performed after 3y median follow-up.

Results: Present translational analysis included 306 of 336 TNBC patients (36 months median FU). pCR was associated with significantly better survival (3y EFS: 92% vs. 71%, p&lt;.001), but despite substantially higher pCR in the carbo arm (46% vs. 29%), no significant EFS advantage was seen (p=.6) (gem: 78%; carbo: 80%; 3y-EFS).

Bivariate Spearman correlations among CD8, PD1, and PDL1 were strongly positive; their correlations with TILs were moderately positive.

Preliminary Cox analysis of EFS was performed with clinical variables (cN, cT, menopausal status); neoadjuvant study arm; pCR; TILs; proliferation markers (baseline Ki67 by IHC, scores derived from PAM50); baseline immune markers; risk scores; and individual gene expression scores previously identified as prognostic for pCR in one or both neoadjuvant arms. Independent prognostic factors included pCR, cN, Ki67, PD1, and CD8; these were entered into (prognostic) interaction analysis. The resulting model contained cN, high Ki67 and low TILs as (unfavorable) main effects and the interaction of (higher) PD1*pCR (favorable).

Among pCR patients, the groups with/without additional adjuvant CTX were similar with respect to explanatory factors. Baseline TILs, Ki67, cN, and PD1 were entered into exploratory predictive analysis; the model retained only the interaction [adjuvant CTx * (fractionally ranked) PD1]. In patients with pCR, those with low PD1 benefited from standard anthracycline-containing adjuvant CTx, whereas patients high PD1 did not with an 98% 3y-EFS.

Conclusions: Our exploratory results suggest independent prognostic impact of mRNA markers and TIL's in early TNBC. Patients with both pCR (after 12 weeks) and “high-immune” signature (defined here by PD1) had excellent 3y-EFS and may be candidates for treatment de-escalation (e.g. omission of anthracyclines), whereas “low-immune” pCR patients may benefit from standard adjuvant poly-chemotherapy.

Citation Format: Gluz O, Nitz U, Liedtke C, Prat A, Christgen M, Feuerhake F, Garke M, Grischke E-M, Forstbauer H, Braun M, Warm M, Hackmann J, Uleer C, Aktas B, Schumacher C, Kuemmel S, Pelz E, Gebauer D, Paré L, Kates R, Wuerstlein R, Kreipe HH, Harbeck N. No survival benefit of chemotherapy escalation in patients with pCR and “high-immune” triple-negative early breast cancer in the neoadjuvant WSG-ADAPT-TN trial [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 GS5-06.

Abstract P6-18-02: Primary and secondary results of the first nationwide molecular screening program in Spain for patients with advanced breast cancer (AGATA SOLTI-1301 study)

Background: Metastatic breast cancer is the second leading cause of death among women globally. A better understanding of tumor biology, and the availability of high-throughput technologies, have enabled the emergence of precision medicine bringing new expectations and giving rise to molecular screening programs (MSP). SOLTI, as a collaborative Spanish network, designed AGATA, the first multi-institutional MSP ever implemented in this country. Here, we report both the primary and some of the secondary results of the pilot study.

Methods: A total of 10 sites within SOLTI network in Spain participated. DNA-sequencing of 56 cancer related genes was performed using FFPE tumor samples (primary or metastatic). Each clinical case was reviewed by a multidisciplinary advisory board (MAB), which recommended, in a prospective manner, potential experimental treatments, mainly in the context of clinical trials. The primary objective was to determine the success rate of matching a DNA alteration to an experimental drug or drug class. Secondary objectives included a comprehensive molecular characterization of tumor samples by PAM50 subtyping and quantification of protein expression levels by MASS-SPEC (70 proteins panel).

Results: 305 patients (pts) were screened from September 2014 to July 2017 and 260 (85.3%) were finally evaluated by the MAB. Pts characteristics were: mean age 54 years (29-80), ER+/HER2- (n=192; 74%), HER2+ (n=30; 11.5%) and TNBC (n=38; 14.5%). 163 primary tumors and 97 metastatic samples were profiled. Regarding the primary objective, 116 pts (45%) presented at least one mutation (range 1-6) that could be matched to a drug or drug class. Of these, 13 pts (11.2%) received therapy matched to their molecular profile according to the MAB recommendation and their follow-up is still on-going. No mutation was detected in 97 (37%) pts (WT), and 47 patients (18.1%) presented a mutation but no match was possible. The most common mutations were PIK3CA (34%), TP53 (22%), AKT1 (5%), ESR1 (3%) and ERBB2 (3%). Intrinsic subtype distribution in 177 samples was as follows: 34% Luminal A (n=60); 21% Luminal B (n=36); 13% HER2E (n=22); 19% Basal-like (n=34) and 13% Normal-like (n=23). Compared to primary tumors (n=110), the proportion of HER2-enriched disease in metastatic tumors (n=63) was significantly higher (6% vs 20%; p=0.005). Protein expression analysis was performed in 146 samples (94 primary and 57 metastasis). In 19 cases (13%), the outlier expression of some targetable proteins (FGFR1 [n=4, 2.7%], IGF1R [n=4, 2.7%], EGFR [n=1, 0.7%], CEACAM5 [n=6, 4.1%], IDO1 [n=2, 1.37%], TROP2 [n=2, 1.37%]) were identified. Of note, HER2 overexpression (&gt;740 amol/μg) was observed in 4 HER2- cases. Finally, among WT tumors, 21% presented a potential drug-matched protein target.

Conclusions: Nationwide molecular screening in Spain is feasible. Nearly half of patients had tumors with mutation(s), mostly PIK3CA, that could be matched to a potential drug or drug class. PAM50 profile might be helpful to navigate towards a therapeutic decision making, although the MAB could not make any targeted-driven recommendation yet with this data. More clinical evidence is needed to use MASS-SPEC as a diagnostic tool.

Citation Format: Pernas S, Villagrasa P, Nuciforo PG, Vivancos A, Scaltriti M, Rodón J, Burgués O, Canes J, Dueñas M, Cecchi F, Vidal M, Lluch A, Perelló A, Llombart A, Dorca J, Montaño A, Oliveira M, Ribas G, Rapado I, Paré L, Prat A, Ciruelos E. Primary and secondary results of the first nationwide molecular screening program in Spain for patients with advanced breast cancer (AGATA SOLTI-1301 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 P6-18-02.

Abstract P6-17-08: Dynamics of tumor-infiltrating lymphocytes (TILs) during neoadjuvant dual HER2 blockade in HER2-positive (HER2+) breast cancer in the absence of chemotherapy

Background: TILs in HER2+ breast cancer (BC) predict 1) prognosis in early setting, 2) complete pathological response (pCR) following neoadjuvant antiHER2-based therapy and 3) response to trastuzumab and pembrolizumab in the metastatic setting. However, less is known regarding changes in TILs during antiHER2-based treatment.

Methods: Stromal TILs where evaluated centrally using H/E slides in tumor samples from the PAMELA (NCT01973660) neoadjuvant phase II trial. Briefly, 151 women with HER2+ BC were treated with lapatinib and trastuzumab, and hormonal therapy if HR positive, for 18 weeks. TIL levels were determined at baseline (n=148), after 2 weeks of treatment (n=134) and at surgery (n=137). Expression of 560 genes, including immune-related genes (e.g. CD8A, CD4, PD1 and PDL1) was measured at the same timepoints (baseline n=151, 2-weeks n=144, surgery n=144) using the nCounter platform. Intrinsic subtyping at baseline was determined using the PAM50 gene expression predictor. Changes in TILs between 2 time-points were determined by paired t-tests. Correlation of TILs with gene expression was assessed by quantitative SAM analysis using a False Discovery Rate &lt;1%. All statistical tests were two-sided and considered significant when p&lt;0.05. All statistical analyses were carried out using the R software.

Results: Compared to baseline, a significant increase in TILs was observed at week 2 in HR- (p&lt;0.001) and HER2-enriched (HER2-E) tumors (p=0.001), but not in HR+ (p=0.133) and non-HER2-E tumors (p=0.067). Within HR- and HER2-E tumors, increase in TILs at week 2 from baseline was observed regardless of pathological response at surgery (pCR and HR- [p=0.008]; RD and HR- [p=0.037]; pCR and HER2-E [p=0.010]; RD and HER2-E [p=0.056]). Compared to week 2, a significant decrease in TILs at surgery was observed in HR- (p=0.002) and HER2-E (p=0.003) tumors, but not in HR+ (p=0.616) and non-HER2-E tumors (p=0.578). Within HR- and HER2-E tumors, a significant decrease in TILs between week 2 and surgery was observed in tumors achieving pCR (p=0.004 and p=0.005), while, in tumors not achieving pCR, no significant tendency was observed (26.4% and 33.0% of tumors showed an increase and a decrease of TILs between week 2 and surgery). Nonetheless, the vast majority of residual tumors (non-pCR) at surgery had TILs above ≥5%: 34.3% 5-10%, 21.0% 10-20%, 15.2% 20-40% and 11.4% ≥40%. Finally, TILs scoring was found highly enriched (FDR&lt;1%) for immune-related genes tracking activated CD8 T-cells (i.e. CD8A, CD3G, LAG3 and PD1). Expression of these immune genes consistently correlated with TIL levels across the 3 time-points.

Conclusions: In early HER2+ BC, a general increase in TILs is observed following 2 weeks of dual HER2 blockade. This observation is mostly observed in HR- and HER2-E subtype, but regardless of pathological response at surgery. After 2 weeks of treatment, TILs consistently decrease in patients achieving a pCR, whereas two main patterns of TILs expression are observed in patients with residual disease at surgery. Nonetheless, most residual tumors at surgery are inflamed (i.e. TILs ≥5%) and might be good candidates for clinical trials evaluating adjuvant immune checkpoint inhibitors.

Citation Format: Griguolo G, Holgado E, Cortés J, Fasani R, Pascual T, Paré L, Bermejo B, Oliveira M, Morales S, Martinez N, Vidal M, Pernas S, Lopez R, Muñoz M, Galvan P, Garau I, Manso L, Alarcón J, Martínez E, Villagrasa P, LLombart-Cussac A, Prat A, Nuciforo P. Dynamics of tumor-infiltrating lymphocytes (TILs) during neoadjuvant dual HER2 blockade in HER2-positive (HER2+) breast cancer in the absence of chemotherapy [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 P6-17-08.

Abstract P5-11-04: Anti-proliferative effect of oral metronomic vinorelbine (mVNB) in PAM50 Luminal/HER2-negative early breast cancer (SOLTI-1501 VENTANA): A randomized, three-arm, window-of-opportunity study

BACKGROUND: The anti-proliferative effect of mVNB alone or in combination with endocrine therapy in patients with hormone receptor-positive/HER2- breast cancer (BC) has been scarcely addressed.

METHODS: Postmenopausal women with untreated stage I-III BC were randomized (1:1:1) to receive 3 weeks of letrozole (LTZ) 2.5mg/day, oral mVNB 50mg 3 days/week or the combination. The 1ary objective was to evaluate, within PAM50 Luminal A/B disease, if the anti-proliferative effect of mVNB+LTZ was superior to monotherapy. An anti-proliferative effect was defined as the mean relative decrease of the PAM50 11-gene Proliferation Score in each arm. 2ary objectives included safety and the comparison of the anti-proliferative effect between arms. An unplanned analysis of stromal tumor infiltrating lymphocytes (sTILs) was performed. PAM50 analyses were performed using the nCounter®-based Breast Cancer 360TM panel. Changes in the expression of 790 genes/signatures tracking multiple biological processes from tumor cells and the microenvironment were evaluated within each arm using paired (surgery vs. baseline) univariate analyses. P-values were corrected for multiple comparisons using false discovery rate (FDR).

RESULTS: A total of 61 patients were randomized and 54 paired samples (89%) were analyzed. Main patient characteristics were mean age 67, mean tumor size 1.7 cm, stage I (55.7%) and grade 1-2 (90%). Grade 3 toxicities occurred in 3.3% of cases. Baseline samples were Luminal A (72.3%) or B (27.7%). The anti-proliferative effect of mVNB+LTZ (-73.2%) was superior to both monotherapy arms combined (-49.9%; p=0.001) and mVNB (-19.1%; p&lt;0.001). The anti-proliferative effect of mVNB+LTZ (-73.2%) was higher compared to LTZ (-65.7%) but did not reach statistical significance (p=0.328). Across the mVNB+LTZ, LTZ and VNB arms, 413 (52.3%), 403 (51.0%) and 21 (2.6%) genes/signatures were found differentially expressed (FDR&lt;5%) between baseline and surgery samples. Compared to mVNB+LTZ baseline samples, surgical samples showed higher expression of AP-1 transcription factor subunits FOS and JUN, inflammatory chemokines (e.g. CCL4 and IL6), stromal-related genes (e.g. CAV1 and stroma signature) and immune infiltration (e.g. CD8 T-cell signature) and lower expression of proliferation-related genes (e.g. MKI67 and UBE2C), estrogen receptor-signaling and Risk of Recurrence. Of the 413 genes found differentially expressed in surgical samples compared to baseline samples in the mVNB+LTZ arm, 108 (26.2%) were not found in the LTZ arm. Among them, high expression of LAG3, CD24, CD84 and CCR5. Under the microscope, sTILs (≥10% at week 3) were observed in 6.6% (mVNB), 15% (LTZ) and 26% (mVNB+LTZ) of the cases. In tumors with ≤10% TILs at baseline, an increase in TILs was observed following LTZ (p=0.049) and mVNB+LTZ (p=0.012).

CONCLUSIONS: mVNB is well-tolerated and presents antiproliferative activity alone and in combination with LTZ. The increase of activated CD8 T-cells or TILs observed with LTZ+mVNB opens the possibility of studying combinations with immunotherapy. Further investigation comparing these biological results with other metronomic schedules or combinations is warranted.

Citation Format: Prat A, Adamo B, Pascual T, Perez Fidalgo JA, Blanch S, Martínez N, Gomez Pardo P, Lopez Gonzalez A, Murillo Jaso L, Amillano K, Vidal M, Paré L, Canes J, Galvan P, Gonzalez Farre B, Ortega Cebrián V, Gonzalez X, Bellet Ezquerra M, Villagrasa P, Ciruelos E. Anti-proliferative effect of oral metronomic vinorelbine (mVNB) in PAM50 Luminal/HER2-negative early breast cancer (SOLTI-1501 VENTANA): A randomized, three-arm, window-of-opportunity 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 P5-11-04.

Abstract P2-09-14: A predictive model of pathological response following dual HER2 blockade-only based on tumor cellularity and tumor-infiltrating lymphocytes (CelTIL) in HER2-positive breast cancer

Background: Increased number of tumor-infiltrating lymphocytes (TILs) at baseline is associated with pathological complete response (pCR) and improved outcomes in HER2-positive early breast cancer treated with anti-HER2-based chemotherapy. In the absence of chemotherapy, the association of TILs with pCR following anti-HER2 therapy-only is currently unknown.

Methods: The PAMELA (NCT01973660)neoadjuvant trial treated 151 women with HER2-positive breast cancer with lapatinib and trastuzumab (and hormonal therapy if hormone receptor-positive) for 18 weeks. Percentage of TILs and tumor cellularity were determined at baseline (n=148) and after 2 weeks of treatment (n=134). Associations of TILs and tumor cellularity with pCR in the breast were evaluated using univariate and multivariable logistic regression models. The regression coefficients were used to derive a score based on TILs and tumor cellularity measured at week 2 (CelTIL) was derived in PAMELA,and tested in week 2 samples from 65 patients with HER2+ disease recruited in the LPT109096 (NCT00524303) phase 2 neoadjuvant trial, where anti-HER2 therapy-only (trastuzumab, lapatinib or the combination) was administered for 2 weeks, followed by the addition of standard multi-agent chemotherapy for 24 weeks.

Results: In PAMELA, at baseline, TILs were significantly associated with pCR in univariate analysis but not in multivariable analysis (adjusted odds ratio [OR]=1.01, 0.98-1.03; p-value=0.620). A statistically significant increase in TILs was observed at week 2 compared to baseline (mean difference +6.9%; p-value&lt;0.001). At week 2, TILs were significantly associated with pCR in univariate and multivariable analyses (adjusted OR=1.04, 1.01-1.06; p-value=0.009). TILs and tumor cellularity at week 2 were independently associated with pCR and a combined score (from 0 to 100) taking into account both variables was derived. CelTIL as a continuous variable was significantly associated with pCR, and patients with CelTIL-low and CelTIL-high scores (cutoff &lt; 33.59) had a pCR rate of 0% and 33%, respectively. Independent validation of CelTIL in week 2 samples from 65 patients with HER2+ disease recruited in the LPT109096 phase 2 neoadjuvant trial will be presented at the conference.

Conclusions: A combined score of TILs and tumor cellularity at week 2 following anti-HER2 therapy-only is associated with pCR upon completion of neoadjuvant anti-HER2-based therapy.

Citation Format: Nuciforo P, Holgado E, Llombart A, Fasani R, Paré L, Pascual T, Oliveira M, Martínez N, Bermejo B, Vidal M, Pernas S, López R, Muñoz M, Garau I, Manso L, Alarcón J, Martínez E, Villagrasa P, Cortés J, Prat A. A predictive model of pathological response following dual HER2 blockade-only based on tumor cellularity and tumor-infiltrating lymphocytes (CelTIL) in HER2-positive breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-09-14.

Abstract P4-04-05: Primary endocrine therapy (PETx) induces PAM50 intrinsic subtype migration with prognostic implications

Backgorund: ER activation is a major ruler of cell biology in ER+ve breast cancer. Hence, ER dependent gene expression at diagnosis, may unveil most of the oncogenic mechanisms responsible of potential tumour relapse and metastasis. Thus, we hypothesized that oestrogen deprivation through PETx may unveil underlying tumour biology with deeper prognostic implications. To explore this, we studied changes in PAM50 intrinsic subtyping and Risk of Recurrence score throughout PETx and their correlation with known prognostic factors.

Methods: Clinical-pathological data were evaluated in a series of patients with stage I-III ER-positive/HER2-negative breast cancer treated in 6 centers in Spain with PETx during more than two months with available baseline and surgical samples. The expressions of 50 genes were measured in baseline samples and surgical specimens using the nCounter platform. Intrinsic subtypes and Risk of Recurrence score (ROR) were determined by the research-based PAM50 predictor. Response by ultrasonography (US) and magnetic resonance (RMI) between diagnosis and before surgery and PEPI score in surgical samples were used as the endpoints. Association between two variables was evaluated using χ2 test or Pearson correlation. All statistical tests were two-sided and considered significant when P≤0.05.

Results: Gene expression profile was feasible in 58 pre/post sample pairs with a median of 7.8 months (range 2.5-40.6) of PETx with AIs (98.3%) or tamoxifen (1.7%). At baseline, 68.9%(n=40) were classified as Luminal A, followed by Luminal B (24.1%; n=14), HER2 enriched (HER2-E) (5.2%; n=3) and Normal-like (1.8%; n=1). Radiologic response did not change significantly according to intrinsic subtype either by MRI or US (P&gt;0.05). Instead, PEPI score varied according to intrinsic subtype (P=0.024).Thirteen (32%) of LumA, while neither of LumB or HER2-E tumours showed a PEPI score Group 1. PETx resulted in changes in the intrinsic subtype in 29 (50%) of tumours

Table 1 Her2-E postLumA postLumB postNormal postHer2 pre2 (66.7%)1 (33.33%)00LumA pre1 (2.5%)21 (52.5%)3 (7.5%)15 (37.5%)LumB pre06 (42.86%)5 (35.71%)3 (21.43%)Normal pre0001 (100%)

. Of note, 2 of 3 (66.7%) HER2 tumors, and 5 of 14 (35.7%) LumB tumors did not change their profile. Forty-eight (83%) tumours showed a decrease in ROR score after PETx (P&lt;0.001). ROR was over 30 in 29 tumours (49.1%) at baseline and 16 (27%) at surgery. Change in ROR was moderately correlated with percentage of change in Ki-67 [correlation: 0.417 (P&lt;0.01)]. Correlation of surgical tumour subtype with IHC markers of cell-viability/apoptosis will be presented.

Conclusion: Oestrogen deprivation of luminal tumours through PETx results in profound changes in tumour biology including a migration in intrinsic subtype in 50% of tumours.

Correlation of the largely decreased ROR with changes in Ki67 reveals the potential prognostic additional information generated by profiling tumours after PETx. ER-blockade may unveil underlying tumour oncogenic capabilities for relapse, survival and metastasis.

Hence, the post-PETx gene expression profile, molecular subtype and ROR may bear incremental prognostic and predictive information generating a novel scenario for optimal clinical decision making.

Citation Format: Larburu L, Paré L, Rezola R, Carrera M, Buch E, Gimenez J, Quiroga V, Fernandez M, Aragón S, Pascual T, Prat A, Urruticoechea A. Primary endocrine therapy (PETx) induces PAM50 intrinsic subtype migration with prognostic implications [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P4-04-05.

Abstract P2-09-05: Independent validation of the PAM50-based chemo-endocrine score (CES) in hormonal receptor positive (HR+)/HER2+ breast cancer (BC) treated with neoadjuvant (NA) anti-HER2-based therapy

Background. HER2+/HR+ BC is heterogeneous and subgroups with different treatment sensitivities need to be identified. We previously reported a PAM50-based CES in HR+/HER2-negative BC (Prat et al. CCR 2016). Here, we evaluated the association of CES with pathologic complete response (pCR) following anti-HER2-based therapy in HR+/HER2+ BC across 6 NA studies.

Methods. Intrinsic subtype and clinico-pathological data were obtained from 6 NA clinical studies (CHERLOB, OptiHERHEART[OHH], PAMELA, LPT109096, ICO and CALGB 40601 [CALGB is part of the Alliance for Clinical Trials in Oncology]). All patients (pts) received chemotherapy (CT) and trastuzumab, except for pts in the PAMELA study who did not receive CT. A second anti-HER2 agent (lapatinib or pertuzumab) was included in the NA of all pts in OHH and PAMELA and in one of the treatment arms in CHERLOB, LPT and C40601. CES was evaluated as a continuous variable, and categorically (CES-E[endocrine-sensitive], CES-U[uncertain] and CES-C[chemo-sensitive]) using the previously reported cut-offs. In all studies, except CHERLOB and C40601, ERBB2 mRNA levels were measured using the nCounter platform. pCR in the breast was the endpoint. We first performed statistical analyses in each dataset individually, and then in a dataset with combined patient-level data. Univariate and multivariable logistic regressions analyses were used.

Results. A total of 345 pts were included in the analysis. In the combined cohort, CES-E, CES-U and CES-C were identified in 27.5%, 23.7% and 48.6% of the pts, respectively. Table 1 summarizes the distribution and pCR rates by CES groups. In the combined cohort, CES-C was associated with higher pCR rates compared to CES-U and CES-E independent of tumor size, nodal status, anti-HER2 treatment (single vs. dual HER2 blockade), intrinsic subtype (HER2-enriched [HER2E] versus not) and study (Table 2). In the PAMELA trial (no CT), CES-C was also found associated with higher pCR rates compared to CES-E (31.6% vs. 0%). Concordant with this finding, CES-C tumors showed higher ERBB2 mRNA levels than non-CES-C (P&lt;0.01).

Conclusion. CES shows clinical validity for predicting CT/HER2-targeting sensitivity in HER2+/HR+ BC beyond intrinsic subtype and clinicopathologic characteristics. HER2+/HR+/CES-E tumors, which represent ˜20% of all HER2+ tumors, show low sensitivity to anti-HER2 regimens (with and without CT); other treatment strategies might be needed for this group.

Support: U10CA180882;U10CA180821;U24CA196171;P50-CA58823;BCRF;Komen.

Table 1. Distribution and pCR rates of CES-E, CES-U and CES-C groups across studies.n(%pCR)LPTCHERLOBOHHICOPAMELAC40601TOTALCES-E00/9(0)1/5(20)0/7(0)0/14(0)16/60(27)17/95 (18)CES-U0/1(0)4/15(27)5/7(71)4/13(30)2/25(8)10/21(47)25/82 (30)CES-C20/29(69)12/32(37)22/28(78)15/29(52)12/38(32)3/6(50)84/162(52)P0.330.080.040.03&lt;0.01&lt;0.01&lt;0.01

Table 2. Association of CES, intrinsic subtype, clinicopathologic variables and study with pCR in a multivariable model (including type of study). OR95% CIPCES   CES-E1  CES-U2.91.2-7.0&lt;0.01CES-C6.22.3-16.4&lt;0.01T3-4 vs T1-20.40.2-0.90.02N+ vs N00.90.4-1.70.68HER2E vs not2.21.1-4.30.02AntiHER2: 2vs12.90.7-3.20.20

Citation Format: Pascual T, Tanioka M, Dieci MV, Pernas S, Gavila J, Guarnieri V, Cortes J, Llombart A, Conte P, Oliveira M, Paré L, Galván P, Perou CM, Prat A, Carey L. Independent validation of the PAM50-based chemo-endocrine score (CES) in hormonal receptor positive (HR+)/HER2+ breast cancer (BC) treated with neoadjuvant (NA) anti-HER2-based therapy [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-09-05.

Abstract P6-05-02: Intrinsic subtype and gene expression changes between primary and metastatic breast cancer

Background: A better understanding of the biological changes occurring during metastatic progression of breast cancer is needed to identify new biomarkers, targets and novel treatment strategies. Here, we compared the intrinsic subtype and the expression of a gene panel across a large dataset of paired primary and metastatic tissues.

Methods: Expression profiling of 105 breast cancer-related genes was performed on 254 (127 pairs) formalin-fixed paraffin-embedded tumor tissues using the nCounter platform. Tumor samples were obtained from 3 independent sources (ConvertHER trial [BCRT 2014] and two in-house datasets). Tumors were classified into each intrinsic subtype using the research-based PAM50 classifier (Parker et al. J Clin Oncol 2009). Chi-square tests were performed to determine the differences in the distribution of variables. Paired two-class Significance of Microarrays (SAM) was performed to determine the genes differentially expressed between paired primary and metastatic tissues. In vitro stable transfection of FGFR4-GFP was performed on Luminal B MCF7 cell line. RNA was purified on control vs. transfected cell lines. 7-AAD cell viability was performed following estrogen deprivation for 6 days.

Results: Subtype distribution in primary vs. metastatic disease was 39.0% vs. 26.8% for Luminal A (p=0.012), 26.0% vs. 35.0% for Luminal B (p=0.322), 11.4% vs. 20.3% for HER2-enriched (p=0.115) and 10.6% vs. 13.0% for Basal-like tumors (p=0.843). The rate of subtype conversion was 7.7% in Basal-like, 23.1% in HER2-enriched, 30.0% in Luminal B and 54.3% in Luminal A disease. The majority of subtype conversions in Luminal A disease were to Luminal B (72.0%) and HER2-enriched (24.0%). Overall, 13.2% of primary Luminal A/B tumors progressed to a HER2-E subtype despite 70% of them being clinically HER2-negative. In a paired analysis using all samples, 10- and 12- genes were found up- and down- regulated in metastatic tissues (False Discovery Rate [FDR] &lt;5%). The up-regulated gene list in metastatic disease was composed of FGFR4 (top gene) and proliferation genes (CDC6, CCNB1, CEP55). The down-regulated gene list in metastatic disease was enriched for luminal-related genes (ESR1, PGR, NAT1 and MAPT). A similar paired analysis within Luminal A, Luminal B, HER2-enriched and Basal-like disease revealed 22, 8, 7 and 0 differentially expressed genes (FDR&lt;5%), respectively. Finally, MCF7 cell line transfected with FGFR4 showed a relative increase in the HER2-enriched profile compared with transfected control. In vitro, MCF7-FGFR4 cells showed estrogen independent growth compared to transfected controls.

Conclusions: Metastatic tissues are relatively more proliferative and less luminal compared to primary tumors. This is especially relevant in primary Luminal A disease. In contrast, metastatic tissues from Basal-like primary disease remain largely unchanged. In luminal disease, a significant increase in the HER2-enriched profile is observed in metastatic disease despite most tumors being clinically HER2-negative. A potential driver of the HER2-enriched profile and estrogen independence in clinically HER2-negative metastatic tissues might be FGFR4.

Citation Format: Prat A, Martínez de Dueñas E, Galván P, Garcia S, Burgués O, Paré L, Antolín S, Martinello R, Blancas I, Adamo B, Guerrero Á, Muñoz M, Nuciforo P, Vidal M, Pérez RM, Chacón JI, Caballero R, Gascón P, Carrasco E, Rojo F, Perou CM, Cortés J, Adamo V, Albanell J, Lluch A. Intrinsic subtype and gene expression changes between primary and metastatic breast cancer. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P6-05-02.

Abstract P2-08-16: Prognostic and predictive abilities of intrinsic subtype in hormone receptor-positive metastatic breast cancer from the EGF30008 phase III clinical trial

Background

Combination of letrozole and lapatinib improved progression-free survival (PFS) compared with letrozole and placebo in patients with hormone receptor-positive (HR+)/HER2+ metastatic breast cancer (MBC), but not HR+/HER2-negative (HER2-) disease (JCO 2009). However, HR+ disease is clinically and biologically heterogeneous with all intrinsic molecular subtypes (Luminal A, Luminal B, HER2-enriched [HER2E] and Basal-like) identified. Here, we tested retrospectively the prognostic and predictive ability of intrinsic subtype in tumor samples of the EGF30008 trial.

Methods

Expression profiling from FFPE tumor tissues was performed on the nCounter platform. Tumors were classified into each intrinsic subtype using the research-based PAM50 classifier (JCO 2009). Cox proportional hazard models for PFS and overall survival (OS) were used to generate point estimates of hazard ratios (HR) and corresponding 95% confidence intervals (CIs). Changes in likelihood ratio χ2 values were used to measure and compare the relative amount of information of each variable. Variables evaluated were: age, prior endocrine therapy, presence of visceral disease, number of metastatic sites, performance status, clinical HER2 status, and treatment. To determine whether the intrinsic subtypes were predictive of lapatinib benefit, we tested the interaction term of subtype by treatment arm in a Cox model that also included the main effects. Kaplan-Meier plots were used to depict the proportion of patients free from progression as a function of time.

Results

Tumor samples from 821 patients (63.8%) were profiled (85.7% primary and 14.3% metastatic tumor samples). Clinical-pathological features of this patient subset were well balanced compared with the original set. Within the entire cohort, all subtypes were identified: Luminal A (46.5%); Luminal B (29.7%); HER2E (7.4%); Basal-like (3.4%) and normal-like (12.9%). Within HER2+ disease, 28.6% of samples were HER2E. Intrinsic subtype was found the strongest prognostic factor independently associated with PFS and OS in all patients, and in patientswith HER2-negative or HER2+ disease (P&lt;0.0001). Median PFS and OS for each subtype within clinically HER2-negative disease were: Luminal A (16.85 and 45.0 months), Luminal B (10.97 and 37.0 months), HER2E (4.67 and 16.0 months) and Basal-like (4.14 and 23.0 months). Within clinically HER2-negative disease (n=644), 16 patients (2.5%) had HER2E disease. Patients with HER2-/HER2E disease benefited from lapatinib (6.5 vs 2.6 months; PFS HR =0.24, 95% CI: 0.07-0.86; P=0.019; HER2E vs not treatment interaction P=0.016). Finally, intrinsic subtype was not predictive of benefit from lapatinib within HER2+ disease.

Conclusions

HR-positive disease is biologically heterogeneous and intrinsic subtypes are strongly prognostic in a first-line MBC setting. HR+/HER2- disease with a HER2E profile may benefit from lapatinib. The clinical value of intrinsic subtyping in HR+ MBC warrants further investigation, but patients with Luminal A/HER2-negative MBC disease might be good candidates for letrozole monotherapy in the first-line setting regardless of visceral disease and number of metastases.

Citation Format: Prat A, Cheang MCU, Galván P, Nuciforo P, Paré L, Adamo B, Viladot M, Press MF, Gagnon R, Ellis C, Johnston S. Prognostic and predictive abilities of intrinsic subtype in hormone receptor-positive metastatic breast cancer from the EGF30008 phase III clinical trial. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P2-08-16.

Pharmacogenomics in colorectal cancer: a genome-wide association study to predict toxicity after 5-fluorouracil or FOLFOX administration

The development of genotyping technologies has allowed for wider screening for inherited causes of variable outcomes following drug administration. We have performed a genome-wide association study (GWAS) on 221 colorectal cancer (CRC) patients that had been treated with 5-fluorouracil (5-FU), either alone or in combination with oxaliplatin (FOLFOX). A validation set of 791 patients was also studied. Seven SNPs (rs16857540, rs2465403, rs10876844, rs10784749, rs17626122, rs7325568 and rs4243761) showed evidence of association (pooled P-values 0.020, 9.426E-03, 0.010, 0.017, 0.042, 2.302E-04, 2.803E-03) with adverse drug reactions (ADRs). This is the first study to explore the genetic basis of inter-individual variation in toxicity responses to the administration of 5-FU or FOLFOX in CRC patients on a genome-wide scale.

How core competencies are taught during clinical supervision: participatory action research in family medicine

OBJECTIVES

The development of professional competence is the main goal of residency training. Clinical supervision is the most commonly used teaching and learning method for the development of core competencies (CCs). The literature provides little information on how to encourage the learning of CCs through supervision. We undertook an exploratory study to describe if and how CCs were addressed during supervision in a family medicine residency programme.

METHODS

We selected a participatory action research design to engage participants in exploring their precepting practices. Eleven volunteer faculty staff and six residents from a large family medicine residency programme took part in a 9-month process which included three focus group encounters alternating with data gathering during supervision. We used mostly qualitative methods for data collection and analysis, with thematic content analysis, triangulation of sources and of researchers, and member checking.

RESULTS

Participants realised that they addressed all CCs listed as programme outcomes during clinical supervision, albeit implicitly and intuitively, and often unconsciously and superficially. We identified a series of factors that influenced the discussion of CCs: (i) CCs must be both known and valued; (ii) discussion of CCs occurs in a constant adaptation to numerous contextual factors, such as residents' characteristics; (iii) the teaching and learning of CCs is influenced by six challenges in the preceptor-resident interaction, such as residents' active engagement, and (iv) coherence with other curricular elements contributes to learning about CCs. Differences between residents' and preceptors' perspectives are discussed.

CONCLUSIONS

This is the first descriptive study focusing on the teaching of CCs during clinical supervision, as experienced in a family medicine residency programme. Content and process issues were equally influential on the discussion of CCs. Our findings led to a representation of factors determining the teaching and learning of CCs in supervision, and suggest directions for research, for faculty development, and for interventions with learners.

Training in interprofessional collaboration: pedagogic innovation in family medicine units

PROBLEM ADDRESSED

A number of agencies that accredit university health sciences programs recently added standards for the acquisition of knowledge and skills with respect to interprofessional collaboration. Within primary care settings there are no practical training programs that allow students from different disciplines to develop competencies in this area.

OBJECTIVE OF THE PROGRAM

The training program was developed within family medicine units affiliated with Université Laval in Quebec for family medicine residents and trainees from various disciplines to develop competencies in patient-centred, interprofessional collaborative practice in primary care.

PROGRAM DESCRIPTION

Based on adult learning theories, the program was divided into 3 phases--preparing family medicine unit professionals, training preceptors, and training the residents and trainees. The program's pedagogic strategies allowed participants to learn with, from, and about one another while preparing them to engage in contemporary primary care practices. A combination of quantitative and qualitative methods was used to evaluate the implementation process and the immediate results of the training program.

CONCLUSION

The training program had a positive effect on both the clinical settings and the students. Preparation of clinical settings is an important issue that must be considered when planning practical interprofessional training.

Interprofessional education at Laval University: Building an integrated curriculum for patient-centred practice

The Laval University Faculty of Medicine, Faculty of Nursing, and School of Social Work, in partnership with the Vieille-Capitale Health and Social Services Centre in Quebec City, obtained funding from Health Canada to design, implement, and evaluate an integrated interprofessional education (IPE) program for family medicine, nursing, and social work students and for professionals from those professions working in primary care. The program was developed around four components and produced the following outcomes: a 45-hour undergraduate curriculum; IPE practical training for professionals, supervisors, residents and trainees in primary care teaching settings; a continuing education model for professionals based on a coaching approach, and; information and communication technology resources. After briefly describing the implementation process, educational content, and evaluation highlights of each component, the integrated program is discussed with respect to the inter-area complementarities and coherence with the conceptual dimensions that have guided the development of the program: IPE, collaboration, and collaborative patient-centered practice. The positive evaluation outcomes, the sustainability of the educational activities, and the enthusiasm of the different partners led to the creation of the Collaborative Network on Interprofessional Practices at Laval University and its affiliated health and social services clinical network.

Impact of CYP2D6 polymorphisms in tamoxifen adjuvant breast cancer treatment

The aim of this study is to evaluate the impact of CYP2D6 genotyping in predicting disease-free survival and toxicity in breast cancer patients treated with adjuvant tamoxifen. DNA from 91 patients was genotyped using the AmpliChip CYP450 GeneChip, Roche that facilitates the classification of individuals by testing 27 alleles. When patients were grouped into group 1 (*4/*4, *4/*41, *1/*5 and *2/*5) and group 2 (the remaining genotypes), a significant difference in disease-free survival (DFS) was observed between groups (P = 0.016). The mean DFS in group 1 was 95 months in contrast with 119 months in group 2. No significant relationship was found between the CYP2D6 genotype classification and severe, mild or no toxicity (P = 0.2). Nevertheless, severe, and mild toxicity was more frequent among poor metabolizer patients than in patients with a normal metabolizer pattern (18.8 and 43.8% vs. 10.7 and 36%, respectively). In breast cancer, patients treated with adjuvant tamoxifen, non-functional and severely impaired CYP2D6 variants are associated with a worse DFS and with a higher frequency of severe and mild toxicities. Larger studies of the CYP2D6 genotype-clinical outcomes association are needed to complement initial results.

Transcription factor-binding sites in the thymidylate synthase gene: predictors of outcome in patients with metastatic colorectal cancer treated with 5-fluorouracil and oxaliplatin?

The identification of clinical and genetic parameters to predict the outcome in advanced colorectal cancer is a key issue in the management of this disease. We ascertained whether the clinical determinants of survival defined in a large cohort of patients treated with 5-fluorouracil (5-FU) (European Organization for the Research and Treatment of Cancer, EORTC model) also apply to 109 colorectal cancer patients receiving a therapy including oxaliplatin/5-FU as their first-line treatment. Our results confirm the considerable discriminatory power of the clinical model proposed in patients treated with a combined chemotherapy regimen. With the aim of identifying additional genetic prognostic parameters, we determined whether the polymorphisms in the promoter region of the thymidylate synthase (TS) gene that modifies the number of operative binding sites of a transcription factor (USF) could predict the clinical outcome of our patients and complement the EORTC clinical model. Our results indicate that this new genetic parameter (the number of USF-binding sites) could be considered when evaluating the role of TS genotype in the efficacy of the 5-FU-based regimens. Further, confirmatory studies aimed at evaluating the effect of the number of binding sites of transcription factors for selecting 5-FU-treated patients are warranted.

Primary health care professionals' views on barriers and facilitators to the implementation of the Ottawa Decision Support Framework in practice

OBJECTIVE

To describe primary health care professionals' views on barriers and facilitators for implementing the Ottawa Decision Support Framework (ODSF) in their practice.

METHODS

Thirteen focus groups with 118 primary health care professionals were performed. A taxonomy of barriers and facilitators to implementing clinical practice guidelines was used to content-analyse the following sources: reports from each workshop, field notes from the principal investigator and written materials collected from the participants.

RESULTS

Applicability of the ODSF to the practice population, process outcome expectation, asking patients about their preferred role in decision making, perception that the ODSF was modifiable, time issues, familiarity with the ODSF and its practicability were the most frequently identified both as barriers as well as facilitators. Forgetting about the ODSF, interpretation of evidence, challenge to autonomy and total lack of agreement with using the ODSF in general were identified only as barriers. Asking about values, health professional's outcome expectation, compatibility with the patient-centered approach or the evidence-based approach, ease of understanding and implementation, and ease of communicating the ODSF were identified only as facilitators.

CONCLUSION

These results provide insight on the type of interventions that could be developed in order to implement the ODSF in academic primary care practice.

PRACTICE IMPLICATIONS

Interventions to implement the ODSF in primary care practice will need to address a broad range of factors at the levels of the health professionals, the patients and the health care system.

Supporting patients facing difficult health care decisions: use of the Ottawa Decision Support Framework

OBJECTIVE

To investigate family physicians' views on factors that make health care decisions difficult for patients, interventions family physicians use to support patients making decisions, and interventions proposed by the Ottawa Decision Support Framework (ODSF).

DESIGN

Thirteen group discussions.

SETTING

Five family practice units.

PARTICIPANTS

One hundred twenty family physicians.

INTERVENTIONS

The multifaceted implementation intervention consisted of feedback from participants, a reminder at point of care, and an interactive workshop. During the workshop, family physicians were asked about their views on 2 videos both showing the concluding phase of a simulated clinical encounter with a woman facing a decision about hormone therapy. One video showed usual care; the other showed use of the ODSF process and related tools. Content was analyzed using observations by non-participants, field notes, material collected from participants during workshops, evaluation forms completed at the end of workshops, and comments written on exit questionnaires from the implementation trial.

MAIN OUTCOME MEASURES

Family physicians' views on the types of difficult decisions their patients face, the factors that make decisions difficult for patients, the interventions family physicians use to support patients' decisions, and the interventions proposed by the ODSF.

RESULTS

The 2 most frequently cited factors making decisions difficult for patients were experiencing uncertainty and fears about adverse outcomes. Before being introduced to the ODSF, participants had used mostly information-related strategies to provide decision support. After learning about the ODSF, participants overwhelmingly identified assessing patients' values as a priority. At the end of the workshop, the 5 changes in practice participants most frequently intended to make were, in order of importance, to assess patients' values, to ask about patients' preferred role in decision making, to screen for decisional conflict, to assess support or undue pressure on patients, and to increase patients' involvement in decision making.

CONCLUSION

The ODSF process and related tools have the potential to broaden family physicians' views on supporting patients facing difficult decisions.

Learning about sex outside the gutter: attitudes toward a computer sex-expert system

In an ongoing research program investigating the utility of Sexpert, an expert computer system designed to counsel couples about their sexual relationships, we tested whether sexually active couples' attitudes concerning computerized sexual counseling could be affected by interacting with Sexpert. Eighty-one young heterosexual couples drawn from the university and general community were assigned to one of three sources of feedback about couple sexual functioning (Sexpert, a video, a self-help book) or a control condition. Attitudes toward all counseling sources used in the study were assessed through semantic differential and similarity-dissimilarity questionnaires. Subjects in the Sexpert condition showed significant improvements in their attitudes toward computerized sexual counseling and evaluated Sexpert significantly better and more similar to a human therapist as a result of exposure. Subjects in the other three conditions showed no significant changes in attitudes. This study provides strong evidence for the acceptability of a computerized sex-expert system.

The relationship of ventricular drainage to aneurysmal rebleeding

Given the widespread use of continuous external ventricular drainage in patients with aneurysmal subarachnoid hemorrhage (SAH), this investigation was undertaken to define the relationship of ventricular drainage to aneurysmal rebleeding. A historical cohort study of 128 patients with confirmed aneurysmal SAH was performed using a multivariate stepwise logistic regression analysis to examine the relationship between aneurysmal rerupture and ventricular drainage, while controlling for important clinical and radiological independent variables. The variables for ventricular drainage selected in the regression analysis were clinical grade, aneurysm size, and presence of hydrocephalus. The rate of rerupture was significantly higher in cases with ventricular drainage (odds ratio 5.31:1, p less than 0.05), poor clinical grade (odds ratio 4.90:1, p less than 0.02), and large aneurysm size (odds ratio 11.25:1, p less than 0.01). The significant effect of ventricular drainage was limited to patients with hydrocephalus. The increased risk of aneurysmal rebleeding in patients undergoing ventricular drainage may result from both: 1) a rise in aneurysmal transmural pressure, since intracranial pressure is lowered by ventricular drainage; and 2) an association between ventricular drainage and a more severely disrupted aneurysm which is more prone to rebleed as part of its natural history. This study found an increased risk of aneurysmal rebleeding among patients undergoing ventricular drainage, particularly in the presence of hydrocephalus.

[Not Available]

Various authors have suggested the possibility of a link between biological age and sexual activity. Zelnick (1981) described the relationship between sexual activity and age at menarche. To determine the relationship between biological age and age at first sexual intercourse, we presented a questionnaire to 450 students attending a comprehensive secondary school in Montreal. Age at menarche and age at first conscious ejaculation were used to measure biological age. A partial correlation was made between biological age and age at first sexual intercourse by controlling for chronological age. The results were r = 0.24 (p <0.029) for female students and r = 0.21 (p <0.08) for male students. The value of the correlation was weak but significant where the female students were concerned. For the male students, the value of the correlation could not be determined due to the insufficient number of respondents. Biological age would appear to be an indicator of the age at which sexual activity first occurs. This indicator could be used by clinicians directly, and to develop a more general model for determining the age at which sexual activity first occurs.