Abstract PD9-05: Prognostic and tamoxifen-predictive effect of PAM50 and ROR score in premenopausal women included in the randomised SBII:2 trial

Abstract

Abstract PURPOSE The Luminal intrinsic subtypes and Luminal-like surrogate subtypes of breast cancer are based on gene expression and pathological markers, respectively. We aimed to study the prognostic value of PAM50 intrinsic subtypes and risk of recurrence (ROR) score in premenopausal women with early breast cancer. Prognostic differences of Luminal PAM50 vs. surrogate subtyping, and the predictive effect of Luminal PAM50 subtypes on tamoxifen benefit were further evaluated. METHODS PAM50 intrinsic subtypes and ROR score (n = 437) were determined by gene expression analyses (NanoString’s Breast Cancer 360 TM assay (BC360)) after RNA isolation of breast cancer tissues from patients included in the randomised SBII:2pre trial. Premenopausal women were randomised between two years of adjuvant tamoxifen vs. control. Surrogate subtyping was performed according to the St. Gallen 2013 classification. The endpoints breast cancer-free interval (BCFi) and overall survival (OS) (median follow-up: 28 and 33 years, respectively) were evaluated using the maximum follow-up and the two disjoint time intervals 0-10 years and >10 years. RESULTS After maximum follow-up, patients with Luminal B tumours by PAM50 subtyping had worse prognosis as compared to those with Luminal A tumours (Hazard ratio (HR)BCfi 1.60; 95% confidence interval (CI) 1.17-2.18; P = 0.004; Table 1). For this time interval, high vs. low ROR score was associated with higher incidence of breast cancer events (HRBCFi 1.70; 95% CI 1.01-2.85; P = 0.04), with a similar trend seen in node-negative patients. In total, 58% of tumours classified as Luminal B-like by surrogate subtyping were re-classified into Luminal A by PAM50 subtyping. At 10 years follow-up, patients whose tumours were classified as Luminal B-like/Luminal A, had a better prognosis as compared to those uniformly classified as Luminal B-like/Luminal B (HRBCFi 0.52; 95% CI 0.33-0.83; P = 0.006) but the effect weakened with longer follow-up (Table 1). According to 10 years data, the incidence of breast cancer events was reduced by two thirds by tamoxifen in patients with PAM50 Luminal A tumours, while no effect was seen in those with Luminal B tumours (HRBCFi 0.41 and HRBCFi 1.19, respectively, (Pinteraction = 0.02); Table 2). CONCLUSIONS PAM50 and ROR score provided prognostic information also in premenopausal women. More than 50% of patients with Luminal B-like tumours defined by surrogate subtyping were re-classified as Luminal A according to PAM50 subtyping. These patients had an improved prognosis at 10 years of follow-up, but the effect attenuated with time. At 10 years, PAM50 Luminal A, but not Luminal B, intrinsic subtype was associated with tamoxifen benefit. Table 1.Prognostic effect of PAM50 intrinsic subtypes and St. Gallen 2013 surrogate subtyping regarding BCFi and OS for different time intervals (uni- and multivariable analyses)UnivariableMultivariableaBCFiOSBCFiOSHR (95% CI); P valuePAM50 intrinsic subtype0–10 years(n = 437, n = 218 events)(n = 437, n = 176 events)(n = 411, n = 203 events)(n = 411, n = 166 events)LumA (Ref.)1.001.001.001.00LumB1.88 (1.31–2.71); 0.0012.33 (1.52–3.58); <0.0011.78 (1.18–2.68); 0.0062.08 (1.28–3.38); 0.003HER2-E2.55 (1.77–3.69); <0.0014.08 (2.70–6.18); <0.0011.99 (1.23–3.24); 0.0052.94 (1.71–5.03); <0.001Basal-like2.09 (1.45–2.99); <0.0013.36 (2.23–5.06); <0.0011.95 (1.19–3.22); 0.0083.00 (1.73–5.22); <0.001>10 yearsb(n = 210, n = 68 events)(n = 261, n = 124 events)(n = 199, n = 66 events)(n = 245, n = 119 events)LumA (Ref.)1.001.001.001.00LumB1.05 (0.55–2.00); 0.890.85 (0.52–1.39), 0.501.92 (0.92–4.01); 0.081.16 (0.66–2.04); 0.62HER2-E0.51 (0.20–1.29); 0.150.70 (0.38–1.29); 0.251.27 (0.33–4.90): 0.731.77 (0.76–4.14); 0.19Basal-like0.95 (0.50–1.82); 0.880.69 (0.40–1.17); 0.172.58 (0.94–7.08); 0.071.90 (0.86–4.18); 0.11Maximum follow-up timec(n = 437, n = 286 events)(n = 437, n = 300 events)(n = 411, n = 269 events)(n = 411, n = 285 events)LumA (Ref.)1.001.001.001.00LumB1.60 (1.17–2.18); 0.0041.41 (1.03–1.93); 0.031.66 (1.17–2.37), 0.0051.48 (1.04–2.10); 0.03HER2-E1.84 (1.32–2.56); <0.0011.99 (1.45–2.73); <0.0011.81 (1.16–2.81); 0.0092.19 (1.43–3.36); <0.001Basal-like1.69 (1.24–2-31), 0.0011.68 (1.24–2.28); 0.0011.92 (1.24–2.99); 0.0042.24 (1.45–3.45); <0.001St. Gallen 2013 surrogate subtype/PAM50 intrinsic subtype0–10 years(n = 207, n = 87 events)(n = 207, n = 60 events)(n = 205, n = 87 events)(n = 205, n = 60 events)LumB-like/LumB (Ref.)1.001.001.001.00LumB-like/LumA0.52 (0.33–0.83); 0.0060.37 (0.21–0.66); 0.0010.50 (0.29–0.84), 0.0090.38 (0.20–0.74); 0.004LumA-like/LumB0.49 (0.12–2.05), 0.330.76 (0.18–3.20); 0.710.77 (0.18–3.32); 0.721.12 (0.25–5.07); 0.88LumA-like/LumA0.39 (0.21–0.73), 0.0030.32 (0.15–0.68); 0.0030.45 (0.23–0.91); 0.030.44 (0.19–1.02); 0.05>10 yearsb(n = 117, n = 42 events)(n = 147, n = 70 events)(n = 115, n = 41 events)(n = 145, n = 70 events)LumB-like/LumB (Ref.)1.001.001.001.00LumB-like/LumA1.30 (0.62–2.69); 0.491.11 (0.62–1.98); 0.720.64 (0.26–1.60); 0.340.59 (0.30–1.15); 0.12LumA-like/LumB----LumA-like/LumA0.45 (0.16–1.23); 0.120.81 (0.41–1.59); 0.540.26 (0.08–0.80); 0.020.39 (0.18–0.84); 0.02Maximum follow-up timec(n = 207, n = 129 events)(n = 207, n = 130 events)(n = 205, n = 128 events)(n = 205, n = 130 events)LumB-like/LumB (Ref.)1.001.001.001.00LumB-like/LumA0.70 (0.47–1.02); 0.060.65 (0.44–0.96); 0.030.58 (0.37–0.90); 0.020.50 (0.32–0.79); 0.003LumA-like/LumB0.34 (0.08–1.41); 0.140.37 (0.09–1.52); 0.170.53 (0.12–2.24); 0.380.48 (0.11–2.05): 0.32LumA-like/LumA0.39 (0.23–0.67); 0.0010.51 (0.32–0.83); 0.0070.39 (0.24–0.70); 0.0020.43 (0.25–0.74); 0.003Abbreviations: BCFi, breast cancer free-interval; CI, confidence interval; HR, hazard ratio; Lum, Luminal; NHG, Nottingham histological grade; OS, overall survival; TAM, tamoxifenaAll analyses were stratified by study region and adjusted for age (continuous), tumour size (>20 mm vs. ≤20), NHG (1 vs. 2 vs. 3), nodal status (N0 vs. N1 vs. N2) and treatment arm in addition to surrogate/molecular subtype b10 years–maximum follow-up time, starting at year 10c32 and 36 years regarding BCFi and OS, respectively Table 2.Predictive value (univariable analyses) of Lum PAM50 intrinsic subtypes for TAM response with respect to BCFi and OS (ER-positive/HER2-negative cohort)BCFiOSHR (95% CI); P value0-10 years(n = 217, n = 92 events)(n = 217, n = 64 events)TAM vs. control in Lum A0.41 (0.23-0.74); 0.0030.61 (0.30-1.26); 0.18TAM vs. control in Lum B1.19 (0.63-2.27); 0.591.76 (0.85-3.63); 0.13Interaction Lum PAM50 subtype x TAM (HR ratio)0.34 (0.14-0.83); 0.020.35 (0.13-0.97); 0.04>10 yearsa(n = 121, n = 43 events)(n = 153, n = 74 events)TAM vs. control in Lum A0.69 (0.35-1.37); 0.290.74 (0.44-1.25); 0.26TAM vs. control in Lum B0.17 (0.04-0.80); 0.030.25 (0.08-0.77); 0.02Interaction Lum PAM50 subtype x TAM (HR ratio)4.05 (0.74-22.1); 0.112.95 (0.85-10.2); 0.09Maximum follow-up timeb(n = 217, n = 135 events)(n = 217, n = 138 events)TAM vs. control in Lum A0.52 (0.34-0.81); 0.0040.71 (0.46-1.08); 0.11TAM vs. control in Lum B0.80 (0.45-1.41); 0.440.87 (0.49-1.54); 0.63Interaction Lum PAM50 subtype x TAM (HR ratio)0.65 (0.32-1.34); 0.240.82 (0.40-1.65); 0.57Abbreviations: BCFi, breast cancer-free interval; CI, confidence interval; ER, oestrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; Lum, Luminal; TAM, tamoxifenAll analyses were stratified by study regiona10 years-maximum follow-up time, starting at year 10b32 and 36 years regarding BCFi and OS, respectively Citation Format: Christine Lundgren, Pär-Ola Bendahl, Maria Ekholm, Mårten Fernö, Carina Forsare, Ute Krüger, Bo Nordenskjöld, Olle Stål, Lisa Rydén. Prognostic and tamoxifen-predictive effect of PAM50 and ROR score in premenopausal women included in the randomised SBII:2 trial [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD9-05.