Low TMB as predictor for additional benefit from neoadjuvant immune checkpoint inhibition in triple-negative breast cancer

581

Background: It is commonly anticipated that a high tumor mutational burden (TMB) is a predictor of response to immune checkpoint blockade (ICB). We previously showed that triple-negative breast cancer (TNBC) from the GeparNuevo study with high TMB displayed increased response both to neoadjuvant chemo-ICB with durvalumab but also to chemotherapy alone, with no significant interaction with treatment arm (Karn et al. Ann Oncol 2020). In contrast, we also observed that cases with very low TMB more often displayed a pCR after treatment with chemo-ICB than with chemotherapy alone. This may in fact suggest a benefit of ICB to those TNBC with rather low TMB. Methods: We have analyzed the distant disease-free survival (DDFS) of GeparNuevo patients according to TMB and treatment arm (neoadjuvant chemotherapy plus durvalumab or chemotherapy plus placebo). For TMB (mut/Mb) we applied the identical cutoff of the upper tertile as in our previous analysis. Results: The median follow-up of the time-to-event data was 43.7 months. Data of TMB was available in 149 of 174 patients. We found that within the high-TMB tumors (durvalumab: n=27; placebo: n=23), DDFS was similar between both arms of the trial (durvalumab vs. placebo: HR (hazard ratio) 0.95 [95%CI 0.19-4.69], p=0.95). Strikingly however, within the low-TMB group (durvalumab: n=47; placebo: n=52) we observed a significantly better DDFS in the durvalumab-chemotherapy combination arm, in contrast to the arm treated only with chemotherapy (durvalumab vs. placebo: HR 0.23 [95%CI 0.06-0.79], p=0.02; interaction test for TMB and treatment arm p=0.17). The observation was also robust to alternative TMB cutoffs. Similar results were obtained for invasive disease-free survival. Conclusions: Our results show, in contrast to other published data, that patients with early TNBC and low TMB/neoantigen counts may benefit from short-term neoadjuvant durvalumab treatment, while for those with high TMB, durvalumab plus chemotherapy does not improve efficacy over chemotherapy alone.

Biomarkers for response to immunotherapy in triple-negative breast cancer: Differences between survival and pCR biomarkers

583

Background: Immunotherapy is entering clinical practice as a promising new neoadjuvant therapeutic approach in triple-negative breast cancer, and it is important to identify biomarkers to focus this therapy on those patients that have the highest benefit. Interestingly, an improved survival outcome is observed in pCR and non-pCR patients, which raises the hypothesis that biomarkers might also be different for pCR prediction as well as prognosis. In this study, we investigated this hypothesis in the neoadjuvant GeparNuevo trial. Methods: A total of 174 patients were randomized to receive neoadjuvant chemotherapy with durvalumab vs. placebo. HTG EdgeSeq mRNA analysis was performed for a total of 2549 genes in 162 pretherapeutic core biopsies. In addition, tumor-infiltrating lymphocytes (stromal and intratumoural) as well as PD-L1 protein expression was evaluated by IHC. We systematically compared the distant disease-free survival (DDFS) of 5 predefined gene signatures (including the GeparSixto immune signature) as well as 12 single mRNA markers identified in previous projects between treatment arms using univariate Cox proportional-hazard regression analyses. In addition, exploratory biomarker analyses were performed. Results: The PSIP1 gene expression (per 1 unit hazard ratio [HR]: 0.58 95%CI 0.41-0.83; p=0.002), TAP1 (per 1 unit HR: 0.68 95%CI 0.48-0.95; p=0.025) as well as stromal TILs (sTILs) (per 10% HR: 0.73 95%CI 0.56-0.95; p=0.019) were significant for improved DDFS in the complete cohort. In the placebo arm PSIP1 (HR 0.50 95%CI 0.29-0.87; p=0.014) as well as sTILs (HR 0.73 95%CI 0.53-0.99; p=0.044) were significant for improved DDFS. In the durvalumab arm, the gene expression of PSIP1 (HR 0.54 95%CI 0.31-0.94; p=0.029), PD-L1/CD274 (per 1 unit HR: 0.41 95%CI 0.21-0.77; p=0.006), CD38 (per 1 unit: HR 0.52 95%CI 0.29-0.92; p=0.026) as well as the GeparSixto immune signature (per 1 unit HR: 0.51 95%CI 0.27-0.97; p=0.041) were significant for improved DDFS, with a positive test for interaction with treatment arm for PD-L1/CD274 (interaction p=0.020). Additional analyses, including multivariate Cox regressions for DDFS as well as systematic comparisons for biomarkers for DDFS and for pCR, will be presented. Conclusions: Our analysis suggests that biomarkers for immune response are linked to improved survival with neoadjuvant durvalumab therapy and that in this setting, survival biomarkers are not identical to pCR biomarkers. The results are a basis for a further dissection of the contribution of pCR to survival effects of immunotherapy.

Prognostic and predictive implications of the intrinsic subtypes and gene expression signatures in early-stage HER2+ breast cancer: A pooled analysis of CALGB 40601, NeoALTTO, and NSABP B-41 trials

509

Background: Several biologic features are implicated in the differences in response and survival to dual (trastuzumab and lapatinib [HL]) vs. single (trastuzumab [H]) HER2-blockade across neoadjuvant trials in early-stage HER2+ breast cancer. We evaluated the association of intrinsic subtypes and gene expression signatures with pathologic complete response (pCR) and event-free survival (EFS) in a pooled analysis of three independent phase III neoadjuvant studies with similar designs: CALGB 40601 (Alliance), NeoALTTO, and NSABP B-41. Methods: Gene expression profiling by RNA sequencing was assessed on 761 pre-treatment samples (264 from CALGB 40601, 249 from NeoALTTO, 248 from NSABP B-41). Intrinsic subtypes and 759 gene expression signatures were calculated. We studied the association of pCR and the benefit of dual (HL) vs. single (H) HER2-blockade by tumor intrinsic subtype in the pooled set. The ability of multiple gene expression signatures to predict pCR and EFS across the three studies was also tested by logistic and Cox regression analyses. Results: pCR status was associated with EFS only in HER2-Enriched (HR 0.45, 95% CI 0.29-0.71, p-value < 0.001) and Basal-like (HR 0.19, 95% CI 0.04-0.86, p-value 0.031) intrinsic subtypes, but not in Luminal and/or ER+ tumors. The EFS benefit of dual vs. single HER2-blockade was limited to HER2-Enriched tumors (HR 0.47, 95% CI 0.27-0.81, p-value 0.007). When evaluating the three clinical trials separately, we found 89/759 (11.7%) gene expression signatures in common for the prediction of pCR across the three clinical trials, including HER2-amplicon and immune activation signatures. Luminal-related signatures were associated with lower pCR rates but better EFS outcomes, especially in patients with residual disease. Stratified Cox regression models by study showed a significant and strong association of NK, B and plasma cells, as well as Ig-related signatures with a better EFS outcome, while vascular, proliferation, and metastasis signatures were associated with poor EFS. Conclusions: In early-stage HER2+ breast cancer, the relationship between pCR and EFS differs by tumor intrinsic subtype, and the benefit of dual vs. single HER2-blockade seems to be limited to HER2-Enriched subtype tumors. Immune signatures were associated with higher pCR rates and better EFS, luminal signatures were associated with lower pCR rates but good EFS outcomes, and vascular/proliferation/metastasis signatures were associated with poor EFS across the three clinical trials. Clinical trial identification: CALGB 40601: NCT00770809. (CALGB is part of the Alliance for Clinical Trials in Oncology). NeoALTTO: NCT00553358 NSABP B-41: NCT00486668

Adaptive immune signature in HER2-positive breast cancer in NCCTG (Alliance) N9831 and NeoALTTO trials

Trastuzumab acts in part through the adaptive immune system. Previous studies showed that enrichment of immune-related gene expression was associated with improved outcomes in HER2-positive (HER2+) breast cancer. However, the role of the immune system in response to lapatinib is not fully understood. Gene expression analysis was performed in 1,268 samples from the North Central Cancer Treatment Group (NCCTG) N9831 and 244 samples from the NeoALTTO trial. In N9831, enrichment of CD45 and immune-subset signatures were significantly associated with improved outcomes. We identified a novel 17-gene adaptive immune signature (AIS), which was found to be significantly associated with improved RFS among patients who received adjuvant trastuzumab (HR 0.66, 95% CI 0.49-0.90, Cox regression model p = 0.01) but not in patients who received chemotherapy alone (HR 0.96, 95% CI 0.67-1.40, Cox regression model p = 0.97). This result was validated in NeoALTTO. Overall, AIS-low patients had a significantly lower pathologic complete response (pCR) rate compared with AIS-high patients (χ2 p < 0.0001). Among patients who received trastuzumab alone, pCR was observed in 41.7% of AIS-high patients compared with 9.8% in AIS-low patients (OR of 6.61, 95% CI 2.09-25.59, logistic regression model p = 0.003). More importantly, AIS-low patients had a higher pCR rate with an addition of lapatinib (51.1% vs. 9.8%, OR 9.65, 95% CI 3.24-36.09, logistic regression model p < 0.001). AIS-low patients had poor outcomes, despite receiving adjuvant trastuzumab. However, these patients appear to benefit from an addition of lapatinib. Further studies are needed to validate the significance of this signature to identify patients who are more likely to benefit from dual anti-HER2 therapy. ClinicalTrials.gov Identifiers: NCT00005970 (NCCTG N9831) and NCT00553358 (NeoALTTO).

Ki67 as Proliferative Marker in Patients with Early Breast Cancer and Its Association with Clinicopathological Factors

INTRODUCTION

Ki67 as a proliferative marker has prognostic and therapeutic relevance in early breast cancer (EBC). However, standard cutoffs for distinguishing low and high Ki67 do not exist.

MATERIAL AND METHODS

Data from all patients treated at the University Hospital Ulm for EBC between January 2013 and December 2015 with documented results for internal Ki67 assessment of the primary (n = 917) tumor were retrospectively analyzed evaluating the associations between Ki67 and other clinicopathological factors.

RESULTS

595 (64.9%) patients had a Ki67 <20% and 322 (35.1%) a Ki67 ≥20%. The median Ki67 was 10% (range 1-90%). Median Ki67 values according to the hormone receptor (HR)/ human epidermal growth factor receptor 2 (HER2) subtypes were 10% for HR-positive/HER2 negative (HR+/HER2-) disease (n = 717), 20% for HR+/HER2+ (n = 76), 30% for HR-/HER2+ (n = 45), and 60% for HR-/HER2- (n = 75). 75.2% or 89.3% of all patients with HER2-positive or triple-negative disease had a Ki67 ≥20%, respectively. Using a multivariable logistic regression with Ki67 (<20% vs. ≥20%) as binary dependent variable, younger age, positive nodal status, higher grading, histological nonspecific type carcinoma, negative HR status, and positive HER2 status were shown to be significantly associated with a higher proliferative index (Ki67 ≥20%).

CONCLUSION

This analysis described Ki67 in different subtypes in EBC and its association with clinicopathological factors. According to more aggressive tumor biology, the respective subgroups also showed higher median Ki67 levels. However, definition of low and high proliferation index itself is difficult. It is essential to interpret Ki67 indices carefully with regard to the own institutional values and other clinicopathological factors.

Durvalumab improves long-term outcome in TNBC: results from the phase II randomized GeparNUEVO study investigating neodjuvant durvalumab in addition to an anthracycline/taxane based neoadjuvant chemotherapy in early triple-negative breast cancer (TNBC)

506

Background: The GeparNuevo trial investigated the addition of durvalumab, an anti-PD-L1 checkpoint inhibitor (CPI), to standard neoadjuvant chemotherapy (NACT) in patients with early TNBC. Durvalumab increased the pathological complete response (pCR) rate particularly in patients treated with durvalumab alone before start of chemotherapy (Loibl et al. Ann Oncol 2019). Methods: GeparNuevo randomized patients with cT1b-cT4a-d tumors and centrally confirmed TNBC to durvalumab (D) 1.5 g i.v. or placebo every 4 weeks. D/placebo monotherapy (0.75 g i.v.) was given for the first 2 weeks (window phase), followed by D/placebo plus nab-paclitaxel 125 mg/m² weekly for 12 weeks, followed by D/placebo plus epirubicin/cyclophosphamide (EC) q2 weeks for 4 cycles. Randomization was stratified by stromal tumor infiltrating lymphocytes (sTILs) (low (≤10%), intermediate (11-59%), high (≥60%)). The primary objective was pCR (ypT0 ypN0). Secondary time-to-event endpoints included invasive disease-free survival (iDFS), distant disease-free survival (DDFS) and overall survival (OS). Results: A total of 174 patients were enrolled between June 2016 and September 2017. The pCR rate with durvalumab was 53.4% versus placebo 44.2% (OR 1.45, 95% CI 0.80–2.63, unadjusted Wald p = 0.224). Durvalumab effect was seen only in the window cohort (pCR 61.0% versus 41.4%, OR 2.22, 95% CI 1.06–4.64, p = 0.035; interaction p = 0.048). After a median follow-up of 42.2 months, 34 events occurred in 174 patients. 3-year iDFS in pCR vs non pCR was 92.0% vs 71.9% (log-rank p = 0.002). 3-year iDFS was 84.9% with durvalumab vs 76.9% with placebo (HR 0.54, 95%CI 0.27-1.09, stratified log-rank p = 0.0559); 3-year DDFS 91.4% vs 79.5% (HR 0.37, 95%CI 0.15-0.87, p = 0.0148); 3-year OS 95.1% vs 83.1% (HR 0.26, 95%CI 0.09-0.79, p = 0.0076). No difference was seen in iDFS, DDFS and OS between the window and no window cohort. Conclusions: Durvalumab added to neoadjuvant chemotherapy in TNBC significantly improved long-term outcome despite a small pCR increase and no continuation after surgery. It needs to be questioned whether adjuvant therapy with CPI is needed at all. Clinical trial information: NCT02685059.

GAIN-2: Neo-/adjuvant phase III trial to compare intense dose-dense chemotherapy (CT) to tailored dose-dense CT in patients (pts) with high risk early breast cancer (EBC): Results on safety and interim invasive disease-free survival (iDFS)

516

Background: GAIN-2 (NCT01690702) compared efficacy and safety of intense, dose-dense epirubicin, nab-paclitaxel, and cyclophosphamide (iddEnPC) vs dose-dense, dose-tailored epirubicin/ cyclophosphamide followed by dose-dense, dose-tailored docetaxel (dtEC-dtD) as adjuvant or neoadjuvant CT for node-positive or high risk node-negative EBC. Here, we report safety results and interim analysis (IA) of the primary endpoint iDFS. Methods: Pts (luminal A ≥N2; luminal B N+; HER2+ and TNBC) were randomized between iddEnPC (E 150 mg/m2, nP 330 mg/m2, C 2000 mg/m2, all q2w x 3) or dtEC-dtD (dtEC q2w x 4 followed after 1 week rest by dtD q2w x 4). Primary objective was to compare iDFS. 797 events are needed to detect a hazard ratio of 0.819 with a 2-sided log-rank-test with 80% power and α=0.05. The IA of iDFS was planned after 50% of the events have occurred. Safety and compliance were secondary objectives. Results: Between 10/2012 and 09/2018, 2887 pts were randomized and 2857 started treatment (iddEnPC 1429; dtEC-dtD 1428). Median age was 51 (range 18-75) years. Overall, 18.1% were luminal A, 31.5% luminal B/HER2-, 18.8% hormone-receptor (HR)+/HER2+, 8.5% HR-/HER2+ and 23.2% TNBC. Overall, 88.1% of pts completed all treatment in both arms. 66.8% with iddEnPC vs 58.8% with dtEC-dtD delayed CT dose (p<0.001). Grade 3-4 non-hematological adverse events (AEs) were more frequent with iddEnPC (iddEnPC 50.8% vs dtEC-dtD 45.1%, p=0.002). Grade 3-4 leukopenia, neutropenia, febrile neutropenia, arthralgia, and peripheral sensory neuropathy were significantly higher with iddEnPC. There were 1464 serious AEs (iddEnPC 870 vs dtEC-dtD 594) and 26 (9 vs 17) predefined AEs of special interest (anaphylaxis, any AE affecting cranial nerves, macula edema). Two deaths occurred during dtEC-dtD. After a median follow-up of 45.8 months, there was no difference in iDFS between arms (log-rank p=0.9102, hazard ratio iddEnPC vs dtEC-dtD 1.01, 95% CI 0.83-1.23). Conclusions: No new safety concerns were observed. Use of both iddEnPC and dtEC-dtD appears feasible in the (neo)adjuvant treatment of high risk EBC. Clinical trial information: NCT01690702 .

Association of T- and B-cell receptor repertoires with molecular subtypes and outcome in HER2+ breast cancer: An analysis of the NeoALTTO clinical trial

511

Background: Clinicopathological and molecular features, including estrogen receptor (ER) status and PAM50 subtypes, have shown an association with immunogenicity and tumor-infiltrating lymphocyte (TIL) levels in breast cancer (BC). To investigate the complexity of the immune response in HER2+ BC, we explored the association of T- and B-cell receptor (TCR and BCR) repertoires with clinicopathological characteristics, PAM50 subtypes and outcome in the NeoALTTO phase III trial. Methods: RNA sequencing (RNAseq) data from baseline tumor biopsies were available for 254 out of the 455 patients enrolled. TCR and BCR repertoires were extracted from RNAseq data using the MiXCR software. Repertoire and diversity measures (read counts, number of clones, evenness, Gini index, Shannon entropy, length of the complementarity-determining region 3 [CDR3], top and second top clone proportions) were estimated. PAM50 subtypes were computed from RNAseq data. Univariate and multivariate (adjusted for clinicopathological characteristics, TIL levels dichotomized using the median value of 12.5% and treatment arm) Cox proportional hazard models were used for survival analysis, while logistic regressions were used for pathological complete response (pCR), defined as ypT0/is. All results reported had a false discovery rate (FDR) <0.05. Results: Higher BCR read counts, number of clones and Gini index were significantly associated with ER-negative as well as grade 3 tumors. Among the PAM50 subtypes, HER2-enriched (HER2-E) showed significantly higher BCR read counts, number of clones and Gini index along with lower evenness compared to luminal A and B, as well as higher length of CDR3 than luminal A. Of note, basal-like showed similar BCR diversity measures to HER2-E. No significant differences were noted for TCR diversity measures. In multivariate analyses, neither TCR nor BCR features were associated with pCR, while BCR evenness (HR 1.5; 95%CI 1.1-2.1) and Gini index (HR 0.66; 95%CI 0.5-0.88) were associated with event-free survival. Conclusions: BCR repertoire measures suggest a clonal expansion in HER2-E and basal-like PAM50 subtypes. Furthermore, the implementation of BCR-derived biomarkers can help to identify patients with an improved clinical outcome after neoadjuvant anti-HER2 treatment. Our findings highlight the heterogeneity of the immune response within HER2+ BC and provide support for biomarker-driven treatment strategies including immunotherapy in this BC subtype. Further validation is required. Clinical trial information: NCT00553358 .

Immunophenotype and proliferation to predict for response to neoadjuvant chemotherapy in TNBC: Results from BrighTNess phase III study

510

Background: In TNBC, the interplay between immunophenotype, tumor proliferation (prolif) and achievement of pathologic complete response (pCR) with neoadjuvant chemotherapy (NAC) remains unknown. Methods: RNA seq was performed on pre-tx research biopsies of stage II/III TNBC enrolled in BrighTNess. NAC regimens included paclitaxel alone or with carboplatin (Cb) or Cb plus veliparib, followed by AC. Computational analysis included subtyping (i.e. PAM50, Pietenpol), prolif (PAM50) and GeparSixto immune signature (GSIS). Cb-containing arms were combined due to similar pCR. Results: High quality RNA seq data was obtained from 482 of 634 pts. PAM50 classified 80.1% of tumors as basal-like. TNBC subtypes were mostly BL1 or BL2 (23.3%), IM (22.4%) or M/MSL (31.7%); 6% were LAR. pCR was higher for basal vs non-basal tumors (52.3% vs 35.4%, p = 0.003). IM had the highest pCR rate (64.2%, 95% CI 59.9%,68.5%). Basal-like was not a significant predictor for Cb benefit (p-interaction = 0.8). Prolif (OR = 0.30 p < 0.001) and GSIS (OR 0.68 p < 0.001) were significantly correlated with pCR but did not correlate with each other (Pearson’s r2 = 0.027). In multivariate analysis, prolif (HR = 0.36 95% CI, 0.21-0.61 p = 0.0002) and GSIS (HR = 0.62 95% CI, 0.49-0.79 p < 0.0001) increased the ability to predict pCR beyond standard clinico-pathologic variables (likelihood ratio = 14.9, p = 0.0001115). Among all pts, those above the median for both prolif. and GSIS had the highest pCR (67%; 84/125) while those below the median for both had the lowest pCR rate (34%; 42/125). Tumors with higher inferred CD8+ T-cell infiltration demonstrated greater benefit from Cb using either TIMER (HR = 0.83 [0.73-0.95]) or CIBERSORT (HR = 0.83 [0.76-0.91]). Tumors with higher inferred total macrophages, particularly immune suppressive M2 macrophages had a higher pCR rate on the non-Cb arm (AC-T) using CIBERSORT (HR = 1.27 [1.07,1.50]). Conclusions: Immunophenotype and proliferation are independent predictors of pCR to standard NAC regimens in TNBC. PAM50 is not a significant predictor of Cb benefit. Exploratory findings suggest that tumor infiltrating immunophenotype (i.e. CD8 T cells and macrophages) may predict response to specific NAC regimens in TNBC. Clinical trial information: NCT02032277.

GeparOLA: A randomized phase II trial to assess the efficacy of paclitaxel and olaparib in comparison to paclitaxel/carboplatin followed by epirubicin/cyclophosphamide as neoadjuvant chemotherapy in patients (pts) with HER2-negative early breast cancer (BC) and homologous recombination deficiency (HRD)

506

Background: The efficacy and toxicity of olaparib in early BC pts with homologous DNA repair deficiency (here defined as HRD score high tumors +/- germline (g) or tumor (t) BRCA mutation) is not well described. GeparOLA investigates olaparib in HER2 negative early BC with HRD. Methods: GeparOLA (NCT02789332) randomized 102 pts to either paclitaxel 80 mg/m² weekly (Pw) plus olaparib 100 mg twice daily for 12 weeks (PwO n = 65) or Pw plus carboplatin (Cb) AUC2 weekly for 12 weeks (PwCb n = 37), both followed by EC. Randomization was stratified by hormone receptor-status (HR+ vs HR-) and age ( < 40 vs ≥40 years). Pts with untreated primary cT2 - cT4a-d or cT1c with either cN+ or pNSLN+ or triple negative or Ki-67 > 20% were included, with either g/t BRCA mutation and/or high HRD score. The primary endpoint is pathological complete response (pCR; ypT0/is ypN0). A one group χ2-test was planned to exclude the pCR rate of ≤55% in PwO→EC arm. Secondary endpoints are other pCR definitions, breast conservation rate, clinical and imaging response, tolerability and safety. Results: A total of 107 pts were randomized between 9/2016 and 7/2018; 106 started treatment. Median age was 47.0 years [range 25.0-71.0]; 36.2% of pts had cT1, 61.0% cT2, 2.9% cT3, and 31.8% cN-positive tumors; G3: 86.8%; Ki-67 > 20%: 89.6%; TNBC 72.6%; confirmed g /tBRCA 1/2 mutation: 60.4%. pCR rate with PwO was 55.1% (90%CI 44.5%-65.3%) vs PwCb 48.6% (90%CI 34.3%-63.2%). Analysis for the stratified subgroups showed higher pCR rates with PwO in the cohorts of pts < 40 years and HR+ pts Clinical trial information: NCT02789332. Conclusions: GeparOla could not exclude a pCR rate of ≤55% in the PwO arm. Subgroup analysis is hypothesis generating and need further confirmation.[Table: see text]

Exome analysis of oncogenic pathways and tumor mutational burden (TMB) in triple-negative breast cancer (TNBC): Results of the translational biomarker program of the neoadjuvant double-blind placebo controlled GeparNuevo trial

509

Background: GeparNuevo (G9) showed a numerical increase in pCR rate of 53% vs 44%; p = 0.287 compared to placebo in TNBC with the addition of the anti-PD-L1 antibody durvalumab (D) to a neoadjuvant anthracycline-taxane containing chemotherapy (Loibl S et al. ASCO 2018). Somatic mutations in malignant cells manifest over the evolutionary history of a tumor. Reports in selected tumor types suggest that TMB may predict clinical outcomes on immune-checkpoint inhibitors (ICI). The clinical relevance of TMB in breast cancer has not been studied widely. Here, we investigated the hypothesis that TMB predicts response to ICI. Methods: Whole exome sequencing was conducted on patient-matched fresh-frozen core biopsies and blood samples with Illumina (n = 149/174). SNVs and indels were called with Mutect and pureCN was used for copy number calls. Mutational signatures were identified as described by Alexandrov et al. (Cell Rep. 3, 2013). Data from G9 were compared to The Cancer Genome Atlas (TCGA) TNBC cohort. Results: A similar genomic landscape was observed between G9 and TCGA with primary mutations in TP53 (69%), c-MYC (26%), BRCA1 (13%), BRCA2 (6%), PIK3CA (11%) and PTEN (11%). Median TMB was 1.52 mut/MB. TMB in G9 was slightly lower than TCGA TNBC. TMB correlated with older age, higher mutation rates in BRCA2, ARID1A, and TP53, and higher burden in variant signatures such as DDR, HRD, GFRs, APOBEC and Alexandrov’s signatures 3 and 6. Continuous TMB predicted pCR in univariate (OR = 1.62, 95%-CI: 1.20 - 2.20, p = 0.0018) and multivariate (OR = 2.06, 95%-CI: 1.33 - 3.20, p = 0.0012) logistic regression models, but did not predict a D effect. After dichotomisation of TMB at the top tertile, 50 patients had high TMB and 29 of these (58%) achieved a pCR, while 99 had low TMB and only 38 of these (38%) had a pCR (p = 0.0242). Conclusions: Results show that TMB may predict pCR in primary TNBC. The trial was financially supported by Astra Zeneca and Celgene. Clinical trial information: NCT02685059.

Influence of the new FIGO classification for cervical cancer on patients’ survival: Retrospective analysis of 265 histologically confirmed cases with FIGO stages Ia to IIb

e17006

Background: End of 2018 a new FIGO Classification for cervical cancer was published, mainly revising stage Ib and introducing a new stage IIIc, which includes irrespectively of tumor size and local spread all patients with lymph node metastasis. Methods: We retrospectively analyzed all cases of cervical cancer stage I to IIb who underwent surgery as primary treatment at our institution from 2000 until 2016 and therefore had a histological confirmation of tumor stage. We reclassified all histologies according to the new FIGO classification and calculated outcome parameters according to the new stage. Results: Out of 265 patients, 146 (55%) patients were reclassified into a higher FIGO stage. Most changes appeared within stage Ib and from any stage to stage IIIc1. Kaplan Meier curves for new stages showed a significant difference for stages I vs. II vs. III (log rank test, p < 0.001). Overall, patients that were upstaged had a significant worse PFS (p = 0.012) and OS (p = 0.008) than patients whose stage did not change. Similar observations were made within sub-stages, when node-positive Ib or IIb tumors were upstaged to IIIc tumors. Conclusions: The new FIGO classification for cervical cancer reflects the strong impact of lymph node metastases on survival and is a clear improvement compared to the old FIGO classification with regard to risk stratification.

Correlation of the tumor mutational burden with the composition of the immune cell subpopulations in peripheral blood of triple-negative breast cancer patients undergoing neoadjuvant therapy with durvalumab: Results from the prospectively randomized GeparNuevo trial

588

Background: The GeparNuevo trial is a randomized, double-blind, multi-center phase II trial of neoadjuvant therapy in patients with early-stage triple negative breast cancer (TNBC) investigating the role of durvalumab, an anti-PD-L1 antibody, which blocks PD-L1 binding to PD1 and CD80, in addition to standard chemotherapy with nab-Paclitaxel (nab-Pac) followed by Epirubicin plus Cyclophosphamid (EC; Loibl S et al. ASCO 2018). Since the tumor mutational burden (TMB) has been suggested to be associated with a better outcome of patients undergoing immunotherapy and an increased T cell response, we determined whether there exists a link between TMB and immune cell composition, frequency and function in patients of the GeparNuevo trial. Methods: In order to determine possible predictive and / or prognostic biomarkers, tumor biopsies taken at recruitment from 149 patients out of the 174 enrolled patients underwent deep sequencing in order to determine the TMB. In addition, for 120 patients blood samples were taken at recruitment and during different time points of treatment (after durvalumab pre-treatment, after Nab-Pac and at surgery after EC) and evaluated using multicolor flow cytometry by monitoring the absolute cell counts of T cells, B cells and NK cells as well as the frequency, composition and functionality of different immune cell populations. Results: The TMB of the GeparNuevo cohort was in line with published data with a mean of 1.8 mutations/MB (range 0.02 – 7.65), respectively. Preliminary evaluation demonstrated a significant correlation of TMB with blood parameters, in particular with subsets of CD8+ T cells. Interestingly, the data suggest a negative correlation of TMB with the frequency of effector cells while a positive correlation exists with the effector memory cells at recruitment. In depth analyses of a correlation with treatment arm and clinical responses are currently performed. Conclusions: Using this approach we hope to identify biomarkers, which will allow a better selection of TNBC patients undergoing specific immunotherapies. Clinical trial information: NCT02685059.

A randomized phase II neoadjuvant study (GeparNuevo) to investigate the addition of durvalumab, a PD-L1 antibody, to a taxane-anthracycline containing chemotherapy in triple negative breast cancer (TNBC)

3062

Background: Adding an anti-PD-L1 checkpoint inhibitor durvalumab to standard chemotherapy (CT) may increase pathological complete response (pCR) in patients (pts) with TNBC. Methods: GeparNuevo randomizes pts to durvalumab (D) 1.5 g i.v. or placebo (pl) every 4 weeks (wks). D/pl monotherapy (0.75 g i.v.) is given for the first 2 wks (window phase), followed by a biopsy and D/pl plus nab-paclitaxel (nP) 125 mg/m² weekly for 12 wks, followed by D/pl plus epirubicin/cyclophosphamide (EC) q2 wks for 4 cycles. Randomization is stratified by stromal TILs (sTILs) (low (≤10%), intermediate (11-59%), high (≥60%)). Pts with primary cT1b-cT4a-d disease, centrally confirmed TNBC, and sTILs status can be included. Primary objective compares pCR (ypT0 ypN0) rates. Secondary objectives are pCR rates in stratified subpopulations and according to other pCR definitions; response rates; breast conservation rate; toxicity; compliance and survival. Change in sTILs, Ki67 and other immune biomarkers before CT, after the window phase and after CT will be correlated with outcome. The first 10, 20 and 30 pts will be included in safety interim analyses (SIA). Sample size was planned assuming a pCR rate of 48% for pl (nP treated TNBC cohort in GeparSepto) and of 66% for D (as clinically meaningful benefit), requiring 158 pts to show superiority of D (2-sided α = 0.2, 80% power). Assuming a 10% drop-out rate 174 pts will be randomized. Results: Since 6/2016, 50 pts were recruited within 16 sites; data are presented as available until 01/2017. Median age is 49 years; 86% NST and G3 tumors; sTILs categories 40% low, 40% intermediate and 20% high. Blinded SIA was performed. No pt interrupted D/pl, one nP and one EC. Treatment delay was observed in 9 pts (20.0%) in D/pl, 18 (41.9%) in nP and 2 (13.3%) in EC; dose was reduced in 10 pts (23.3%) in nP and in 4 (26.7%) in EC. 10 pts (20%) had at least one grade 3-4 AE: 4 haematological and 6 non-haematological AEs. 4 SAEs and 5 immune related AEs were reported. 2 pts discontinued study treatment prematurely in the EC phase. Conclusions: The addition of D to standard nP-EC is feasible and does not result in an increased toxicity. Clinical trial information: NCT02685059.

Chemotherapy-induced ovarian failure (CIOF) in young women with early breast cancer (EBC)

10068

Background: Women ≤45 years (yrs) treated with chemotherapy (CT) for EBC have a high risk of developing CIOF. Awareness of CIOF is essential for young women. Methods: 740 patients (pts) aged ≤45yrs treated with anthracycline or taxane-based CT for EBC from 4 German neoadjuvant/adjuvant trials were included. Blood samples were collected at baseline (N=740), end of treatment (EOT n=740), 6 (n=177), 12 (n=113), 18 (n=69), 24 (n=47) months (m) after EOT. Only samples collected in a time sequence were included. Estradiol (E2), Follicle-Stimulating Hormone (FSH) and Anti-Müllerian Hormone (AMH) were centrally assessed. CIOF was defined as FSH >12.4IU/l and E2 <52.2ng/ml and was analysed per timepoint and according to clinical and treatment-related variables. Results: Median age was 40yrs (range 21-45); 57.2% had BMI 18.5-<25, 41.1% ≥25; 32% had luminal, 35.9% HER2+, 32.0% triple-negative BC. Median hormone levels at baseline for pts <30yrs vs 30-40yrs vs ≥40yrs were: FSH 5.2IU/I vs 5.6IU/I vs 6.4IU/I; E2 101ng/l vs 86ng/l vs 88ng/l; AMH 2.14ng/ml vs 1.58ng/ml vs 0.53ng/ml. 85.7% of pts had CIOF at EOT, 62.2% at 6m, 54.0% at 12m, 43.5% at 18m, 38.3% at 24m. Similar results were observed in 47 pts with all timepoint samples available. Older vs younger pts had more frequently CIOF at EOT (≥40yrs 94.6%, 30-40yrs 82.0%, <30yrs 50.0%, p<0.001). CIOF at EOT was not influenced by BMI. CT agents impacted the rate of CIOF (p<0.001; Table 1). Higher rate of CIOF was associated with longer CT duration (12w 58.3%, 16-18w 94.5%, 24w 82.1%; p<0.001) and with dose-dense (ddEC-ddD, weekly PM(Cb), intense-dd (idd) EnPC) vs conventional dosed CT (P/nP-EC q3w, P, Cz) (94.5% vs 78.6%; p<0.001). Conclusions: The majority of young women experienced CIOF after CT for EBC. After 2 yrs 62% of the pts returned to premenopausal hormone levels. Age, CT regimen, duration and density influenced the rate of CIOF and should be taken into account when counseling young women who desire to maintain ovarian function. [Table: see text]

A randomised phase III trial comparing two dose-dense, dose-intensified approaches (EPC and PM(Cb)) for neoadjuvant treatment of patients with high-risk early breast cancer (GeparOcto)

518

Background: The sequential use of intense does-dense (idd) epirubicin, paclitaxel, cyclophosphamide (EPC) and weekly paclitaxel/liposomal doxorubicin (+/- carboplatin (Cb) in triple negative breast cancer (TNBC) (PM(Cb)) are considered highly efficient regimens for high-risk early stage breast cancer (BC). Methods: GeparOcto (NCT02125344) patients (pts) received 18 weeks (wks) either EPC (3x E 150mg/m² q2w followed by 3x P225 mg/m² q2w followed by 3x C 2000mg/m² q2) or PM(Cb) (12x P 80mg/m² plus M 20 mg/m² q1w, plus Cb AUC 1.5 q1w in TNBC). For HER2+ BC trastuzumab 6 (8) mg/kg q3w and pertuzumab 420 (840) mg q3w cycles were given concomitantly with P and C. Pts with histologically confirmed, cT1c - cT4a-d BC and central receptor assessment were included. Pts with HER2+ or TNBC were eligible irrespective of nodal status, luminal B-like tumours only if pN+. Primary objective compared pathologic complete response (pCR) rates (ypT0/is ypN0). Sample size calculations assumed a pCR rate of 50% for EPC and 60% for PM(Cb), requiring 950 pts to show superiority of PM(Cb). Secondary objectives compared pCR rates within the stratified subgroups (BC subtype, HER2+ vs HER2- HR+ vs HER2- HR-), amongst others. Results: 961 pts were recruited between 12/2014 and 05/2016, 945 started treatment. Median age was 48 years, 4% T3, 2% T4d, 46% N+, 82% ductal invasive, 66% G3 tumors; 40% were HER2+, 43% TNBC. 347 pts reported SAEs (176 EPC/171 PM(Cb)) and 2 pts died. 35 pneumonias (2 EPC vs 33 PM(Cb)) and 18 pneumonitis (3 EPC vs 15 PM(Cb)) were reported. 16.4% pts with EPC and 33.8% with PM(Cb) discontinued treatment (p<0.001), mainly due to AEs (47 EPC vs 113 PM(Cb)). Mean treatment duration was 17 wks with EPC and 16 wks with PM(Cb). pCR rate was 48.3% with EPC and 47.6% with PM(Cb)(OR 0.97 (95%CI 0.75-1.25), p=0.876). pCR rate in TNBC was 48.5% with EPC and 51.7% with PM(Cb); in HER2+ 62.0% vs 57.4% and in Luminal B 14.1% vs 14.6%. Conclusions: In high-risk early stage breast cancer pts pCR rates of idd EPC compared to weekly PM(Cb) were not significantly different. PM(Cb) appeared to be less feasible. Clinical trial information: NCT02125344.

Breast conservation after neoadjuvant chemotherapy for triple-negative breast cancer: Surgical results from an international randomized trial (BrighTNess)

514

Background: Neoadjuvant systemic therapy (NST) increases the frequency of breast-conserving therapy (BCT) in stage II-III breast cancer, but there is little data on how often it converts patients (pts) from BCT-ineligible (BCT-I) to BCT-eligible (BCT-E) and on the impact of other factors on surgical choices. We collected surgical assessment and management data from an international randomized trial of NST in triple-negative breast cancer (TNBC). Methods: Women with operable TNBC were randomized to veliparib (V) with carboplatin (C) and paclitaxel (P), placebo with C and P or placebo with P followed by doxorubicin and cyclophosphamide. The surgeons assessed BCT candidacy by clinico-radiographic criteria before and after NST; surgical management was at surgeon and patient discretion. We assessed interactions between BCT eligibility pre- and post-NST, germline BRCA mutation ( gBRCA) status, continent of treatment and achievement of pathologic complete response(pCR) and percentage of pts who underwent BCT versus mastectomy. Results: Pre- and post-NST surgical assessments were available for 604 pts who underwent surgery. BCT rates are listed in the Table. The BCT rate was 68% among pts deemed BCT-E after NST. pCR rates were identical between BCT-E pts who chose BCT (55%) vs. mastectomy (53%). Of 141 pts deemed BCT-I at baseline, 75 (53%) converted to BCT-E but only 42 (56%) of these opted for BCT. pCR rates were 49% in BCT-E converts vs. 36% in those remained BCT-I. gBRCA pts (n = 84) were less likely to choose BCT even if they were BCT-E. Pts treated in North America (NA) were less likely to choose BCT (55% vs. 80% for Europe and Asia P<0.0001) even among non- gBRCA considered BCT-E post-NST (61% vs. 85% P<0.0001). Conclusions: This largest prospective analysis of the impact of NST in TNBC demonstrates a conversion rate from BCT-I to BCT-E of 53%. BCT rates were lower in pts with gBRCA; the much higher mastectomy rate among BCT-E pts in NA merits investigation. Clinical trial information: NCT02032277. [Table: see text]

Phase 3 study evaluating efficacy and safety of veliparib (V) plus carboplatin (Cb) or Cb in combination with standard neoadjuvant chemotherapy (NAC) in patients (pts) with early stage triple-negative breast cancer (TNBC)

520

Background: Clinical studies suggest that TNBC is sensitive to DNA-damaging agents, including Cb. V is a potent PARP inhibitor that may enhance the antitumor activity of such agents. We present primary response data from a phase 3 randomized, placebo-controlled study (NCT02032277) evaluating the addition of V + Cb or Cb to neoadjuvant paclitaxel (P) followed by doxorubicin + cyclophosphamide (AC). Methods: Pts with histologically confirmed, invasive TNBC (T2–T4 N0–2 or T1 N1–2) amenable to surgical resection were randomized 2:1:1 to (Arm A) P 80 mg/m2 weekly + Cb AUC 6 mg/mL/min q3 weeks + V 50 mg PO BID; (Arm B) P + Cb + PO placebo; or (Arm C) P + IV placebo + PO placebo, for 12 weeks followed by AC (60 mg/m2 or 600 mg/m2 q2 or 3 weeks) × 4. Primary endpoint was pathologic complete response (pCR) in breast and nodes with > 80% power at 2-sided α of 0.05 using pair-wise comparisons for A vs B and A vs C to detect significant treatment effects using Χ2 test; secondary endpoint was rate of conversion to eligibility for breast conservation surgery (BCS). Adverse events (AEs) were assessed with NCI CTCAE V4.0. Results: Six hundred thirty-four pts (median age 50 years; range 22–79) were randomized to Arms A (n = 316), B (n = 160), or C (n = 158). Baseline characteristics were well balanced. No pCR difference was observed between Arms A and B (53.2% vs 57.5% p = 0.36), but pCR in Arm A was higher than Arm C (53.2% vs 31.0% p < 0.001). In non-prespecified analysis, pCR in Arm B was also higher than Arm C (57.5% vs 31.0% p < 0.001). Among pts ineligible for BCS at screening (n = 141), 62% were eligible after NAC in Arm A vs 44% each in Arms B (p = 0.13) and C (p = 0.14). Grade 3–4 AEs (Arms A/B/C, 86%/85%/45%) and serious AEs (30%/27%/14%) neutropenia, thrombocytopenia, anemia, nausea, and vomiting were increased with the addition of Cb; V did not impact toxicity. Median cycles of NAC were not reduced with V + Cb + P or Cb + P vs P. Conclusions: Addition of V to neoadjuvant Cb + P followed by AC did not increase pCR rate in breast and nodes in stage II–III TNBC, while addition of V + Cb or Cb alone to P followed by AC did. Cb (+/– V) increased toxicity but did not impact delivery of NAC. Clinical trial information: NCT02032277.

Survival outcomes of the NeoALTTO study: Updated results of a randomized multicenter phase III neoadjuvant trial

512

Background: In the neoadjuvant NeoALTTO trial dual HER2 blockade with lapatinib (L) plus trastuzumab (T) combined with weekly paclitaxel significantly increased the pathologic complete response rate (pCR) compared with either anti-HER2 agent alone plus paclitaxel. At first analysis pts with pCR had a better event free survival (EFS) and overall survival (OS) after median follow-up of 3.84 yrs. Methods: 455 pts with operable HER2-positive breast cancer were randomized to receive either L (n=154) 1500mg/day, T 4mg/kg loading dose followed by 2mg/kg/wk (n=149) or L 1000mg/day plus T (n=152) for 6 weeks followed by the assigned anti-HER2 treatment combined with paclitaxel weekly x 12. Following surgery pts received 3 cycles fluorouracil, epirubicin and cyclophosphamide q 3 weeks. The assigned anti-HER2 treatment was continued for 34 weeks thereafter. Primary endpoint was pCR (ypT0/is), secondary endpoints were EFS and OS and the association between pCR and OS analyzed by landmark analysis 30 weeks after randomization. Median follow-up was 6.7 years. Results: 6-yrs EFS rate was 67%/ 67%/74% with L/T/TL, respectively (L vs T HR 0.98 [95% CI 0.64–1.51] p=0.93; TL vs T HR 0.81 [95% CI 0.52–1.26] p=0.35). In the hormone receptor negative group 6- yrs EFS rate was 61%/ 63%/74% for the 3 groups, respectively (L vs T HR 1.09 [95% CI 0.61–1.95] p=0.76; TL vs T HR 0.81 [95% CI 0.44–1.51] p=0.52). OS at 6 yrs was 82%/79%/85% for L, T and TL, respectively (L vs T: HR 0.85 [95% CI 0.49-1.46] p=0.56; TL vs T HR 0.72 [95% CI 0.41-1.27] p=0.26). In landmark analyses, pts with a pCR had significantly higher 6-yrs EFS (77% /65%) and OS (89% /77%) compared to those without pCR, both overall and for the hormone receptor negative cohort. Conclusions: The updated results of the NeoALTTO study confirm the sustained survival benefits for pts who achieve a pCR. EFS and OS after 6 yrs did not differ significantly between the 3 treatment groups. The combination of T and L showed numerically higher EFS compared to T, especially in the hormone-receptor negative group. Clinical trial information: NCT00553358. [Table: see text]

Impact in delay of start of chemotherapy and surgery on pCR and survival in breast cancer: A pooled analysis of individual patient data from six prospectively randomized neoadjuvant trials

571

Background: Time interval from diagnostic biopsy to neoadjuvant chemotherapy (NACT) start (TBC) and from last chemotherapy application to surgery (TCS) are influenced by many factors. It is unclear whether a delay of systemic therapy or surgery impacts patients (pts) outcome. Methods: 9127 pts with early BC from 6 German neoadjuvant trials receiving an anthracycline-taxane based NACT were included. pCR (ypT0/is ypN0), disease free survival (DFS) and overall survival (OS) were compared according to TBC and TCS length (cut-off of ≤4 vs >4weeks (w)), overall and in subgroups (BC subtypes [luminal, HER2+, triple-negative breast cancer (TNBC)] and pCR [yes vs no] for survival endpoints) adjusted by study. Results: Data on TBC were available for 8072 pts, on TCS for 6420, on follow-up (FU) for 7889. Median age was 49 yrs, 25.6% had cT3-4, 48.6% N+, 44.1% G3, 46.0% luminal, 26.4% TNBC, 27.6% HER2+ tumors. Median (m) FU-time was 65 months [0-201]. mTBC was 23 days [0-228] (67.5% ≤4w vs 32.5% >4w), mTCS was 28d [0-204] (53.7% ≤4w vs 46.3% >4w), with inter-study variability for mTBC ranging from 14 to 32d and for mTCS ranging from 24 to 29d from the oldest to the most recently conducted study. TBC did not influence the pCR rate, neither in all patients nor in subgroups. At multivariable logistic regression analysis TBC length did not independently predict pCR. TBC did not influence DFS or OS, neither in all patients nor in subgroups. TCS<4w was associated with a trend towards a better DFS in all patients (HR=1.11 95%CI (0.99-1.24), p=0.08) and in pts not achieving pCR (HR=1.12, 95%CI (0.99-1.26), p=0.08). No difference was observed within BC subtypes. OS was not impacted by TCS length. At multivariable Cox regression analysis TBC or TCS≤4 vs >4w did not independently influence DFS or OS. Conclusions: A delay in starting NACT does not impact the pCR rate, DFS or OS and results are independent of the subgroup. However, early surgery after NACT in pts without pCR seems to influence outcome. Our analysis is explorative, but indicates for the first time, that time interval of starting NACT and undergoing surgery might be uncritical. Further research is ongoing.

Combination chemotherapy with nintedanib/placebo for patients with advanced or recurrent endometrial cancer: The NSGO ENGOT-EN1/FANDANGO trial

TPS5611

Background: Endometrial cancer (EC) patients with advanced and recurrent disease relapse despite treatment with combination chemotherapy and have a short progression-free survival (PFS). With the emerging clinical data on anti-angiogenic agents and with promising results of nintedanib in ovarian cancer, it is apparent to explore its role in EC. Nintedanib is a potent, orally available triple receptor tyrosine kinase inhibitor targeting VEGFR 1-3, PDGFR α/β, and FGFR 1-3. This placebo-controlled, multicenter, two-arm, phase 2 trial compares nintedanib versus placebo as concomitant and maintenance therapy in combination with chemotherapy in patients with advanced or recurrent EC. Methods: The primary objective of this trial is to evaluate efficacy of nintedanib against placebo in combination with chemotherapy, defined by PFS. Key eligibility criteria include: histologically confirmed EC, stage 3C 2 or 4 A & B or relapsed after adjuvant therapy for stage 1-3 disease; prior surgery; adjuvant chemotherapy; radiation therapy; hormonal therapy are permitted; measurable/non-measurable disease. 148 patients will be randomized 1:1 to receive nintedanib 200mg twice daily or placebo days 2-21 during chemotherapy (six cycles of Carboplatin (AUC5) and paclitaxel (175mg/m2) every 21 days) and continuously in maintenance phase. Nintedanib/placebo is continued until disease progression, unacceptable toxicity, or withdrawal. Secondary endpoints include PFS in sub-populations, PFS2, disease specific survival, time to first subsequent therapy, time to second subsequent therapy, overall survival, objective response, disease control rate, patient reported outcoms (assessed via EORTC QLQ-C30 and EORTC QLQ-EN24) and safety. Trial is enrolling patients. The following cooperative groups are participating: NSGO (DK, FIN, SWE, NOR), NOGGO (GER), BGOG (BEL), & GINECO (FRA). Clinical trial information: NCT02730416.

The prognostic and predictive impact of genomic grade index (GGI) versus central grade or molecular class in intermediate-risk breast cancer (BC): Results from the EC-Doc trial

1092

Background: Potential markers for adjuvant taxane-based chemotherapy (CTx) in early, intermediate-risk BC include histologic grade (HG), Ki-67, Genomic Grade (GG) or molecular classification. The randomized EC-Doc trial demonstrated improvements in DFS and OS for EC-Doc vs. FEC in patients with 1-3 positive LN. Methods: Centrally assessed protein expression data by IHC, histology/HG (n=772) and GG (n=476) were obtained. Luminal A/B classes were defined as: ER/PR+/KI-67<20/>20% and/or HER2+. Correlations/concordance of these factors were estimated; impact on DFS and value for predicting taxane-based CTx benefit was assessed. Results: Low, equivocal and high GG (GG-1/-EQ/-3) categories were observed among 54 (11%); 60 (13%); 358 patients (76%) and associated with decreasing 5-yr-DFS rates of 100%, 92% and 82% (p < 0.001).There is only 60% concordance between local (L)/central(C) HG assessments and 63% between GG and C-HG. 37.7% of GG-3 tumors were L- HG3 vs. 72% C-HG3; 79% of C-HG2 and 83% of L-HG2 tumors were re-classified (56-71% to GGI3). Only 5.6-6% of GG1 were HG1 by L/C-HG respectively.GG was prognostic only in L-HG subgroups.In univariate subgroup analyses, EC-Doc was significantly superior to FEC: C-G3 (HR=0.58, 0.39-0.93), high Ki-67 (HR=0.55, 0.32E0.92) and GG-3 (HR=0.58, 0.34-0.99) subgroups. In multivariate analyses of HR+ disease (including age, therapy, GGI category, tumor size, LN status, central/local HG, Ki-67, HER2), either C-HG3 / high Ki-67 (as dichotomous variables) or C-HG3 / GGI (as continuous variable) were identified as independent prognostic factors.In interaction analysis, only C-HG, luminal A subtype and interaction of luminal-B/therapy were significant. If local HG was included instead of C-HG, only GG was as independent prognostic factor. Conclusions: These data support GGI as independent prognostic factor in early HR+ BC and as predictive marker regarding taxane benefit by univariate analysis. There is heterogeneity between L/C-HG and GGI. C-HG and Ki-67 assessment appear similarly informative. A predictive effect regarding benefit of taxane-containing CTx was seen in the IHC-luminal B subtype.

Randomized, open-label, phase II study comparing the efficacy and the safety of cabazitaxel versus weekly paclitaxel given as neoadjuvant treatment in patients with operable triple-negative or luminal b/HER2 normal breast cancer (GENEVIEVE)

TPS1138

Background: Cabazitaxel is a new taxoid promoting the tubulin assembly in vitro and stabilizing microtubules against cold-induced depolymerization as efficiently as docetaxel. It has shown superior survival against mitoxantrone plus prednisone in docetaxel pre-treated hormone refractory metastatic prostate cancer pts leading to its registration. It showed a favorable toxicity profile with a low rate of alopecia. In the GENEVIEVE study it will be compared to weekly paclitaxel, which is currently most widely used in breast cancer (BC) pts. Methods: This is a prospective multicenter, randomized, open label study investigating efficacy and safety of cabazitaxel. Pts with uni- or bilateral primary BC (stage cT3/T2/T1c and cN+/T1c and pNSLN+), tumor lesion ≥ 2cm (palpation) or ≥ 1cm (sonography) and centrally confirmed TNBC or luminal B/HER2- can be included. Pts will be randomized to four q3w cabazitaxel (25mg/m² i.v.) vs. 12 q1w paclitaxel (80mg/m² i.v.). Randomization will be stratified by nodal status and subtype. Treatment will be given until surgery, disease progression, unacceptable toxicity or withdrawal of consent. The primary objective is pathologic complete response (pCR) (ypT0/is ypN0/+). Secondary objectives are pCR in stratified subgroups and by other definitions, objective response rate, pCR and local recurrence free survival in pts with clinical complete response and neg. core biopsy before surgery, breast conservation rate, toxicity, compliance, survival rates, biomarkers predicting response. Assuming 15% pCR in controls and targeting a smallest clinical improvement of 10% (i.e. pCR = 25% in experimental arm), a total of 326 pts (163/arm) are required for the one-sided proportion comparison test (α=0.1) with 80% power. The trial is registered under NCT01779479. Results: Recruitment is planned for 12 mths in 45 (+10 back-up) sites in Germany. 1st pt in is planned for Feb 2013. Conclusion: GENEVIEVE will rapidly and precisely compare efficacy and tolerability of cabacitaxel vs. paclitaxel to decide if further development in BC is reasonable. Clinical trial information: NCT01779479.

Impact of expression levels of mRNA HER2 and ESR1 on the pathologic complete remission (pCR) rate after neoadjuvant treatment with anthracycline-taxane containing chemotherapy in combination with trastuzumab in the GeparQuattro trial

523

Background: Recent data suggest that benefit from (neo)adjuvant trastuzumab might be related to expression of HER2 and estrogen receptor (ESR1). We investigated mRNA levels of HER2 and ESR1 in core biopsies in HER2 positive tumors of the GeparQuattro trial and compared it to response to neoadjuvant treatment. Methods: In the GeparQuattro trial 445 HER2+ pts were included based on local testing and received neoadjuvant trastuzumab and chemotherapy (either 4 x EC →4 x docetaxel (D) or 4 x EC → 4 x D/capecitabine (C) or 4 x EC →4x D →4 X C). In 217 available pretherapeutic core biopsies HER2 levels were analysed by IHC, SISH and quantitative RT-PCR using predefined cutoffs; pCR was defined as ypT0is;ypN0. Results: Only 73% of the tumors (158 of 217) were centrally HER2 positive (cHER2+) by IHC/SISH, while 59 tumors (27%) were centrally HER2 negative (cHER2-). As all pts had received neoadjuvant trastuzumab, this gave us the possibility to evaluate response of HER2- tumors to trastuzumab. The pCR rate of cHER2+ tumors was significantly increased (46.8%, p<0.0005) compared to cHER2-, who had a pCR rate of only 20.3%. Similar results were obtained if HER2 positivity was determined by RT-PCR (pCR rate 50% vs. 17.4%, p<0.0005). Assessment of HER2 status by RT-PCR showed a concordance of 86.2% with the IHC/SISH status. In uni- and multivariate logistic regression analysis of cHER2+ cases including continuous HER2 and ESR1 mRNA levels the HER2 mRNA expression was significantly associated with prediction for a pCR in (univ.: OR 1.43, 95% CI 1.11-1.83, p=0.005; multiv.: OR 1.42, 95% CI 1.11-1.83, p=0.006). In the cHER2+ tumors, the pCR rate was similar for ESR1 mRNA positive (47.4%) and ESR1 negative tumors (46.3%). Conclusions: Only pts with cHER2+ tumors independently of the method used have an additional benefit in terms of a pCR from adding trastuzumab to chemotherapy. In cHER2-negative patients the pCR rate is comparable to the pCR rate in the non trastuzumab treated population. Increasing HER2 mRNA levels were associated with a better response to trastuzumab based treatment in cHER2 + tumors.

Prognostic and predictive impact of Ki-67 before and after neoadjuvant chemotherapy on PCR and survival: Results of the GeparTrio trial

1023

Background: We previously reported as a result of the GeparTrio phase III trial that response-guided neoadjuvant chemotherapy (CT) with TACx8 or TAC/NX, compared to TACx6, can improve survival especially in hormone-receptor (HR)-positive tumors. As this benefit could not be predicted by pathological complete response (pCR), better surrogate response markers are warranted. Methods: 2072 patients with operable or locally advanced breast cancer were treated with 2 cycles TAC before interim response assessment. Responders were randomized to additional TACx4 or TACx6 and non-responders to TACx4 or NXx4. We centrally measured Ki-67 in1165 pre-CT core biopsies and in 676 post-CT surgical samples. Counting patients with a pCR as having 0% Ki-67, 757 pre-/ post-CT pairs were available. Ki-67 percentage levels were grouped to low (0-15%), moderate (15.01-35%), and high (35.01-100%) according to cut-point finding analysis in a training and validation cohort. Results: pCR rates were 4.2%, 12.9%, and 29.0% in tumors with low, moderate, and high pre-CT Ki-67 levels (p<0.0001). Pre-CT Ki-67 levels significantly predicted disease-free survival (DFS) (log rank p<0.0001) overall, in the HR+ (p<0.0001), but not in the HR- (p=0.5) subgroup. Post-CT Ki-67 levels correlated with DFS (p<0.0001). Patients with low post-CT Ki-67 levels showed comparable outcome to patients with pCR. Patients with increased Ki-67 levels from before to after CT showed an impaired outcome compared to patients with stable or decreased Ki-67 levels (p<0.0001). However, post-CT Ki-67 levels appeared to have more prognostic relevance than Ki-67 changes. Low post-CT Ki-67 levels were not more frequent after response-guided treatments (response-guided vs conventional: p=0.153; TACx6 vs TACx8: p=0.335; TACx6 vs TAC/NX: p=0.420). Similar negative results were found for HR+ and HR- subgroups. Conclusions: Pre-CT Ki-67 levels are predictive for pCR and prognostic for DFS. Post-CT Ki-67 levels and changes between pre-and post-CT-Ki-67 levels are prognostic for DFS. As neither could predict different treatment effects on DFS, Ki67 cannot replace pCR as a surrogate marker for outcome after neo-adjuvant CT.

Oral administration of an estrogen metabolite-induced potentiation of radiation antitumor effects in presence of wild-type p53 in non-small-cell lung cancer

PURPOSE

The purpose of this study was to investigate the efficacy of 2-methoxyestradiol as an antitumor and radiosensitizing agent for the treatment of human malignancy.

METHODS AND MATERIALS

Two cancer cell lines with wild-type p53 status were exposed first to irradiation and then to an oral formulation of the nontoxic metabolite 2-methoxyestradiol (2ME) to stabilize p53 levels.

RESULTS

Cell growth was inhibited via G1 growth and apoptosis. Subsequent in vitro growth and Tunel assays indicated that this combination was superior to radiation alone at inducing p53 protein accumulation, stabilizing p53 protein levels, and substantially reducing long-term tumor cell growth (approximately 80%) and colony formation (approximately 95%) in vitro, and inducing apoptosis. However, harboring mutated p53, H322 cell line, was relatively insensitive to such a treatment regimen. Western blot analysis revealed that growth inhibition was associated with increased levels of p53 and p21 protein accumulation. Experiments with subcutaneous tumor in a nu/nu mouse showed the combination treatment to be superior to radiation alone at reducing tumor growth ( approximately 50% reduction as compared to radiation alone) in vivo.

CONCLUSION

Thus, our studies confirmed a unique strategy whereby oral administration of a nontoxic estrogen metabolite, 2ME, significantly enhanced the radiation effect on a subcutaneous tumor without any toxicity and suggesting that this strategy may be clinically useful as an adjuvant therapy.