Resistance to human epidermal growth factor receptor 2 (HER2)-targeted therapies and immuno-oncology agents poses a major challenge in treating HER2-positive breast cancer. Here we demonstrate that p95HER2, a truncated form of HER2, drives immune evasion in HER2-positive female breast cancer, enhancing tumor growth and conferring therapy resistance. This stems from the unique ability of p95HER2 to promote cancer cell-intrinsic programmed death ligand 1 expression and secretion of immunosuppressive mediators including interleukin 6. In preclinical models, this impairs the efficacy of trastuzumab deruxtecan, a HER2-directed antibody–drug conjugate (ADC) that relies on immunogenic responses to cell death for full efficacy. Importantly, we find that neratinib potently directs proteasomal degradation of p95HER2, relieving its immunosuppressive effects, and provide proof of concept that neratinib and/or agents targeting p95HER2 downstream mediators can restore antitumor immunity and trastuzumab deruxtecan efficacy. This study reveals a p95HER2-specific therapy resistance mechanism in HER2-positive female breast cancer and highlights the potential value of targeting p95HER2 to improve outcomes with ADCs or immuno-oncology agents. Hu et al. discovered that the truncated form of human epidermal growth factor receptor 2 (HER2), called p95HER2, drives tumor progression and resistance to the antibody–drug conjugate trastuzumab deruxtecan in HER2+ breast cancer. Blocking p95HER2 restores antitumor immunity.

Patients with KRAS wild-type (wt) metastatic colorectal cancer (mCRC) treated with single-agent cetuximab (C) or panitumumab (P), have improved progression-free survival (PFS) and overall survival (OS) compared to best supportive care but an objective response rate (ORR) of only 13–17

We previously reported our phase Ib trial, testing the safety, tolerability, and efficacy of T-DM1 + neratinib in HER2-positive metastatic breast cancer patients. Patients with ERBB2 amplification in ctDNA had deeper and more durable responses. This study extends these observations with in-depth analysis of molecular markers and mechanisms of resistance in additional patients. Forty-nine HER2-positive patients (determined locally) who progressed on-treatment with trastuzumab + pertuzumab were enrolled in this phase Ib/II study. Mutations and HER2 amplifications were assessed in ctDNA before (C1D1) and on-treatment (C2D1) with the Guardant360 assay. Archived tissue (TP0) and study entry biopsies (TP1) were assayed for whole transcriptome, HER2 copy number, and mutations, with Ampli-Seq, and centrally for HER2 with CLIA assays. Patient responses were assessed with RECIST v1.1, and Molecular Response with the Guardant360 Response algorithm. The ORR in phase II was 7/22 (32

Background:The cardiotoxic effects of doxorubicin, trastuzumab, and other anticancer agents are well known, but molecular genetic testing is lacking for the early identification of patients at risk for therapy-related cardiac toxicity.Methods:Using the Agena Bioscience MassARRAY system, we genotyped TRPC6 rs77679196, BRINP1 rs62568637, LDB2 rs55756123, RAB22A rs707557, intergenic rs4305714, LINC01060 rs7698718, and CBR3 rs1056892 (V244M) (previously associated with either doxorubicin or trastuzumab-related cardiotoxicity in the NCCTG N9831 trial of anthracycline-based chemotherapy ± trastuzumab) in 993 patients with HER2+ early breast cancer from the NSABP B-31 trial of adjuvant anthracycline-based chemotherapy ± trastuzumab. Association analyses were performed with outcomes of congestive heart failure (N = 29) and maximum decline in left ventricular ejection fraction (LVEF) using logistic and linear regression models, respectively, under an additive model with age, baseline LVEF, and previous use of hypertensive medications as covariates.Results:Associations of maximum decline in LVEF in the NCCTG N9831 patients did not replicate in the NSABP B-31 patients. However, TRPC6 rs77679196 and CBR3 rs1056892 were significantly associated with congestive heart failure, p < 0.05, with stronger associations observed in patients treated with chemotherapy only (no trastuzumab) or in the combined analysis of all patients relative to those patients treated with chemotherapy + trastuzumab.Conclusions:TRPC6 rs77679196 and CBR3 rs1056892 (V244M) are associated with doxorubicin-induced cardiac events in both NCCTG N9831 and NSABP B-31. Other variants previously associated with trastuzumab-related decline in LVEF failed to replicate between these studies.

Background: The primary aim of the NRG Oncology/NSABP B-52 clinical trial was to test if estrogen deprivation (ED) administered concomitantly with neoadjuvant docetaxel, carboplatin, trastuzumab, and pertuzumab (TCHP), would improve the pCR rate in patients with HER2+/ER+ early breast cancer. A numerical increase in the pCR rate was observed with ED (46.1% v 40.9%), but the difference was not statistically significant. The purposes of this study were to assess the association of sTILs in pretreatment biopsies with pCR in the total population and within the molecular subtypes of breast cancer and to assess changes in sTILs between pre- and on-treatment biopsies. The secondary endpoints of recurrence-free interval (RFI) and overall survival (OS) are currently being analyzed and will be presented along with association of these endpoints with sTILs in pretreatment biopsies in the total cohort and within molecular subtypes. Methods: Scoring of sTILs on routine H&E slides from pre-treatment biopsies with sufficient tumor from 249 of the 315 patients (79%) entered in B-52 were performed by one of two pathologists (SKM, RSM). Both pathologists scored sTILs on a subset of 64 patients to document concordance. Wilcoxon two-sided test, box and whisker plots, and forest plots were used to assess associations with pCR. Molecular subtypes were determined utilizing RNA-seq data and AIMS subtyping method. On-treatment biopsies were available in 46 patients and were scored and compared to paired baseline samples. Results: Good concordance between pathologists was established with an inter-pathologist difference of ˂20% difference between scores in 92% of cases. sTILs in pre-treatment samples were associated with pCR across both arms of the trial (p=0.0074) and in the TCHP+ED arm (p=0.033), but not in the TCHP arm (p=0.093). The distribution of intrinsic subtypes was 34% luminal B, 29% luminal A, 28% HER2E, 5.8% normal, and 2.7% basal, with no significant differences between the arms. Presence of sTILs showed a trend for association with pCR in HER2E pre-treatment samples (p=0.054) but not in non-HER2E (p=0.75). Similarly, sTILs were associated with pCR in non-luminal tumors (p=0.055) but not in luminal tumors (p=0.44). Stratification by treatment arm and menopausal status suggested sTILs are associated with pCR in premenopausal women treated with TCHP (OR: 1.04, 95% CI=1.00-1.09). Interestingly, decreases in the sTIL scores with treatment were associated with pCR in the TCHP+ED arm (p=0.01) but not in the TCHP arm. Analysis of RFI and OS on B-52 is ongoing and will be presented along with associations of sTILs with intrinsic subtypes for RFI and OS. Conclusions: Although a positive correlation between sTILs and pCR was observed, the clinical utility appears limited because of the extensive overlap in the TIL scores between pCR and non-pCR tumors. Significance for a positive association of sTILs with pCR was detected in HER2E but not in luminal tumors. This may be due to the molecular differences of the subtypes, or the make-up of the TILs, or both. The association of a decrease in sTILs with TCHP+ED treatment needs further investigation. The small number of samples is a limitation of the study; however, the B-52 protocol specified that the collection of the B-52 samples was for the purpose of exploratory analysis. Our results highlight the molecular heterogeneity of the HER+/ER+ patient population and suggests that different treatment strategies may be required in future treatment regimens for this patient population. Support: NSABP Foundation; BCRF; 3U10CA180868-03S2, -180822; UG1CA189867; Genentech. Citation Format: Katherine L. Pogue-Geile, Sai K. Maley, Rim S. Kim, Ying Wang, Roberto Salgado, Corey Lipchik, Huichen Feng, Reena S. Cecchini, Samuel A. Jacobs, Ashok Srinivasan, Eleftherios (Terry) Mamounas, Charles E. Geyer Jr, Priya Rastogi, C. Kent Osborne, Soonmyung Paik, Norman Wolmark, Peter C. Lucas, Mothaffar Rimawi. Association of stromal tumor infiltrating lymphocytes (sTILs) in pretreatment biopsies in different molecular subtypes of HER2+/ER+ breast cancer: Assessment of NRG Oncology/NSABP B-52 [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P1-04-10.

Figure SF3 Comparisons of the expression of the NRG family of ERBB ligands, at the RNA level, between post-anti-EGFR metastatic biopsies collected upon enrollment into FC-7, and post-neratinib residual tumor tissue samples.

Figure SF2: cfDNA mutational analysis cfDNA from plasma collected upon enrollment was analyzed for mutations with a 74-gene Guardant panel (1). Alterations were identified in 30 genes as shown in the figure. Patient IDs: Green=Stable Disease, Red=Progressive Disease, and Black=non-evaluable. Specific patients are discussed below and in the main text.

Figure SF1 Representative IHC and CISH data are shown. HER2 status is negative in the primary tissue (A) but is amplified in the metastatic tissue collected upon enrollment into FC-7, based on CISH (C) and IHC staining (B) with the antibody from Ventana.

Background: The NRG Oncology/NSABP B-52 neoadjuvant clinical trial was conducted to test if the addition of estrogen deprivation (ED) would improve the pCR rate in HER2+/ER+ breast cancer patients (pts) treated with docetaxel, carboplatin, trastuzumab, and pertuzumab (TCHP). A numerical increase in pCR rate was observed with ED (46.1% v 40.9%), but the difference was not statistically significant. We have previously quantitated T cells (CD8, FOXP3), macrophages (CD68), and immune checkpoint proteins (PD-1, PD-L1) with multiplex immunofluorescence in B-52 and shown that CD68 and FOXP3 cells were associated with pCR but not CD8 cells. Our purpose was to determine the associations of FCGR genotypes and immune cells with pCR. Methods: A single baseline, pre-treatment FFFPE tissue section per case (N=181) was used to perform a 7-plex multiplex immunofluorescence procedure using opal fluorophores for staining. The Vectra Pathology System and inForm analysis software (Akoya Biosciences) was used for imaging and quantitation of CD8, CD68, FOXP3, PD-1, and PD-L1 cells in both the tumoral and stromal regions. Stromal data is reported here. Favorable- and unfavorable- FcGγR genotypes for FCGR2A-131H/R and FCGR3A-158V/F alleles were determined via the Sequenom MassARRAY iPLEX platform. Rates of pCR with pts with 1 or 2 favorable alleles was compared to pts who were homozygous for the unfavorable allele. Within each genotype, Wilcoxon rank sum test was used to test the association of markers with pCR and within each treatment. Results: No significant association of FCGR2A and 3A alleles with pCR was detected in the entire B-52 cohort, however, among pts with favorable FCGR genotypes (FCGR2A-131-HH, or H/R, FCGR3A-158- VV, or VF HR) the median value of the % CD68 cells was significantly higher in tumors that achieved pCR v those that did not (p=0.0004, p=0.0006), respectively. In pts who were homozygous for the FCGR2A or FCGR3A unfavorable alleles, there was no significant difference in the median values of the % of CD68 cells between pCR and no-pCR tumors. Further stratification of tumors by treatment showed that pts with an FCGR2A or FCGR3A favorable genotype and whose tumors achieved pCR had a higher median value of CD68 only in the TCHP + ED arm (p=0.0007, p=0.0003), respectively and not in the TCHP arm (p=0.059; p=0.21). Higher levels of PD-L1 were associated with pCR in pts with FCGR3A- favorable genotypes, but higher levels of FOXP3 were associated with pCR regardless of genotype. In contrast to the other cell types, higher PD-1 or CD8 cells showed no association with genotypes. Conclusions: This is an exploratory study examining the potential role of ADCP in HER2+/ER+ breast cancer and supports the notion that ADCP may be one mechanism that promotes the elimination of tumor cells in a subset of pts in the neoadjuvant setting. Tumors that achieve pCR have higher % of CD68 cells, in pts with favorable FCGR2A and 3A genotypes than pts who do not. However, in pts with unfavorable FCGR3A or FCGR2A genotypes there was no difference in the median CD68 levels in pCR v no-pCR tumors. When tumors were further stratified by CD68 levels, FCGR3A genotypes, and treatment, the association of pCR in tumors with high CD68 and FCGR3A favorable genotypes was seen only in the TCHP+ED arm. This may indicate that ED may improve pCR rates in some tumors with more macrophages and favorable genotypes. Macrophages are known to have estrogen receptors, and estrogen has been shown to promote the alternative activation of macrophages, potentially dampening down the immune response. Thus, one could speculate that ED may block the estrogen-induced alternative activation of macrophages, allowing the classically activated macrophages to phagocytize tumor cells. Support: BCRF, U10CA180868 & Admin Sup, U24CA196067, Genentech, NSABP Foun. Citation Format: Katherine L Pogue-Geile, Ying Wang, Huichen Feng, Corey Lipchick, Patrick Gavin, Rim S Kim, Reena S Cecchini, Samuel A Jacobs, Ashok Srinivasan, Sandra M Swain, Eleftherios Mamounas, Charles E Geyer, Jr, Priya Rastogi, Peter C Lucas, C. Kent Osborne, Soonmyung Paik, Norman Wolmark, Mothaffar F Rimawi. Potential role of the antibody-dependent cellular phagocytosis (ADCP) in tumors achieving pCR in NRG Oncology/NSABP B-52 [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 P1-07-04.

Increased EGFR (epidermal growth factor receptor) expression has been reported in many types of human cancer and its levels are positively associated with advanced cancers. Recently, upregulation of Id-1 (inhibitor of differentiation or DNA binding) protein was found in over 70% of ovarian cancer samples and correlated with poor survival of ovarian cancer patients. However, the molecular mechanisms responsible for the role of Id-1 in ovarian cancer are not clear. The aim of this study was to investigate the effect of Id-1 on ovarian cancer proliferation and its association with the EGFR pathway. To achieve this, we transfected an Id-1 expression vector into three ovarian cancer cell lines and examined cell proliferation rate by flow cytometry and bromodeoxyuridine staining. We found that ectopic Id-1 expression led to increased cell proliferation demonstrated by increased BrdU incorporation rate and S-phase fraction. The Id-1-induced cell growth was associated with upregulation of EGFR at both transcriptional and protein levels. In contrast, inactivation of Id-1 through transfection of an Id-1 antisense vector resulted in downregulation of EGFR. Our results indicate that increased Id-1 in ovarian cancer cells may promote cancer cell proliferation through upregulation of EGFR. Our findings also implicate that Id-1 may be a potential target for the development of novel strategies in the treatment of ovarian cancer.

Abstract Purpose: In metastatic colorectal cancer (mCRC), HER2 (ERBB2) gene amplification is implicated in anti-EGFR therapy resistance. We sought to determine the recommended phase II dose (RP2D) and efficacy of neratinib, a pan-ERBB kinase inhibitor, combined with cetuximab, in patients with progressive disease (PD) on anti-EGFR treatment. Patients and Methods: Twenty-one patients with quadruple-wild-type, refractory mCRC enrolled in this 3+3 phase Ib study. Standard dosage cetuximab was administered with neratinib at 120 mg, 160 mg, 200 mg, and 240 mg/day orally in 28-day cycles. Samples were collected for molecular and pharmacokinetic studies. Results: Sixteen patients were evaluable for dose-limiting toxicity (DLT). 240 mg was determined to be the RP2D wherein a single DLT occurred (1/7 patients). Treatment-related DLTs were not seen at lower doses. Best response was stable disease (SD) in 7 of 16 (44%) patients. HER2 amplification (chromogenic in situ IHC) was detected in 2 of 21 (9.5%) treatment-naïve tumors and 4 of 16 (25%) biopsies upon trial enrollment (post-anti-EGFR treatment and progression). Compared with matched enrollment biopsies, 6 of 8 (75%) blood samples showed concordance for HER2 CNV in circulating cell-free DNA. Five SD patients had HER2 amplification in either treatment-naïve or enrollment biopsies. Examination of gene-expression, total protein, and protein phosphorylation levels showed relative upregulation of ≥2 members of the HER-family receptors or ligands upon enrollment versus matched treatment-naïve samples. Conclusions: The RP2D of neratinib in this combination was 240 mg/day, which was well tolerated with low incidence of G3 AEs. There were no objective responses; SD was seen at all neratinib doses. HER2 amplification, detectable in both tissue and blood, was more frequent post-anti-EGFR therapy.

BACKGROUND: Colorectal cancer (CRC) pts with deficient mismatch repair (dMMR) tumors have greater immune cell infiltration than pts with proficient (pMMR) tumors and is thought to be partly responsible for the robust response of dMMR pts to PD-1/PD-L1 blockade. Not all pts with dMMR tumors respond to PD-1 blockade. Therefore, we conducted exploratory analyses to understand the complexity of the tumor immune microenvironment. We developed a gene expression signature predictive for MMR status and tested it for associations with prognosis.METHODS: The one-sided Wilcoxon signed-rank test was used to select genes that were significantly over expressed in dMMR v pMMR tumors in a randomly selected discovery cohort in NSABP C-08 (N=500). Six chemokine genes were selected to build a predictive signature for MMR status because they were among the top 20 most significantly differentially expressed genes and were part of an immune chemo-attractant pathway. Addition of other genes did not improve prediction accuracy. The signature was tested in an independent cohort of C-08 (N=454) and in three other data sets: NSABP C-07 (N=1603), TCGA colon (N=619), and the NSABP Molecular Profiling Registry (MPR-1) (N=263). MPR-1 tissue samples were from primary tumors in pts who had recurred. Survival analyses were performed in C-07, C-08, and TCGA data, using Cox proportional hazards models. MMR status was determined for MPR-1 tissues with IHC using antibodies for MSH2, MLH1, PMS2, and MSH6.RESULTS: Higher chemokine scores (CKS) were significantly associated with dMMR tumors in all three independent data sets composed of treatment-naïve early-stage CRC pts (P<0.01). However, in primary tumors of dMMR pts who developed recurrent disease, CKS was non-significantly lower than pts with pMMR tumors (P=0.072). As expected, high CKS was associated with a better prognosis in all three data sets, composed of early-stage CRC and was a more robust prognostic indicator in all 3 datasets than was MMR status. When CKS was combined with MMR status in early-stage pts, pts with high CKS had a better prognosis than pts with low CKS regardless of MMR status.CONCLUSIONS: The CKS score has a strong association with MMR status in three different data sets and is associated with a better prognosis in early-stage CRC. In these pts, dMMR tumors had significantly higher CKS than pMMR tumors, but primary tumors from MPR1 pts who had all become metastatic showed a trend for association in the reverse direction. We conclude that dMMR primary tumors, which are likely to become metastatic, have a very different immune microenvironment than those that do not because the CKS is composed of chemo attractants for a variety of immune cells including NK, T, and B cells. Such a signature could be useful for prospectively identifying dMMR pts with early-stage CRC who are at high risk for relapse. SUPPORT: Genentech; sanofi; U10CA180868, UG1CA189867, U24CA196067, PA DoH; NSABP Foundation.Citation Format: Ying Wang, Rim S. Kim, Corey Lipchik, Ashok Srinivasan, Huichen Feng, Nan Song, Carmen J. Allegra, Patrick G. Gavin, Samuel A. Jacobs, Norman Wolmark, Peter C. Lucas, Katherine L. Pogue-Geile. Development of a chemokine signature identifying dMMR patients (pts) with poor prognosis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3407.

After the publication of this work [1] the authors have reported that in Table 3 The letter "T" in columns 5 and 7 should not be there.

Background: NRG Oncology/NSABP B-52 neoadjuvant clinical trial was conducted to test if the addition of estrogen deprivation (ED) would improve the pCR rate in HER2+/ER+ breast cancer pts treated with docetaxel, carboplatin, trastuzumab, and pertuzumab (TCHP). A numerical increase in the pCR rate was observed with ED (46.1% v 40.9%), but the difference was not statistically significant. The purpose of this study was to determine the utility of using stromal tumor infiltrating lymphocytes (sTILs), mutations, established, and novel signatures, to assess their value for predicting pCR, particularly in the TCHP + ED arm. Methods: Whole transcriptome RNA-Seq and Ampli-Seq libraries were sequenced on the Ion Torrent platform. Mutations were assessed with a custom Ampli-Seq panel of 117 genes, including HER2-activated pathways and/or trastuzumab (T)-resistance markers. The 8-gene T-benefit signature was prospectively tested for association with pCR, and was previously validated in the adjuvant setting in B-31 and NCCTG N9831. Hot-spot mutations, sTILs, and signatures for immune cells, intrinsic subtypes, risk of recurrence proliferation (RORP), and MammaPrint, were also tested for associations with pCR. RNA-Seq data was also used to identify ED-predictive genes. Differential expression was assessed in normalized RNA-Seq data using DESeq2 and each sample was subtyped with the AIMS classifier. Wilcoxon two-sided test, chi-square, or Fisher's exact tests were used to assess associations with pCR. Results: Subtypes were determined with RNA-Seq data from pretreatment biopsies (N=230). The 8-gene T-benefit signature was associated with pCR. The high-, medium-, and low- T-benefit groups had pCR rates of 55%, 44%, and 6.8%, respectively (p=1.3e-13). Intrinsic subtypes were associated with pCR by comparing HER2E to all other subtypes combined (69% v 28%, p=3.9e-08). Hot spot mutations in PIK3CA alone (p=0.014), or combined with hot spot mutations in ERBB2, ERBB3, AKT1, PTEN, and MAP3K1, were associated with no pCR (p=0.0007) and may be useful as resistance markers. sTILs, MammaPrint, RORP, and some immune cell signatures also showed statistically significant associations with pCR but did not identify a subset of pts with an increased pCR rate in the TCHP + ED arm. In contrast, expression of the SH3BP2 gene was associated with pCR only in the TCHP + ED arm (interaction p=0.03). Conclusion: The previously validated 8-gene T-benefit signature identifies a subset of pts with very low pCR rate (6.8%) with TCHP. PIK3CA and other activating mutations were associated with no pCR. These findings may identify pts with resistant disease who may require a different treatment. Exploratory analyses suggest that SH3BP2 expression may identify pts who may benefit from TCHP + ED, but validation is required for clinically utility. Support: BCRF, U10CA180868, -180822; UG1CA189867, Genentech, PUMA Biother Citation Format: Katherine L. Pogue-Geile, Ying Wang, Huichen Feng, Corey Lipchik, Rim S. Kim, Reena S. Cecchini, Samuel A. Jacobs, Ashok Srinivasan, Joseph P. Costantino, Eleftherios P. Mamounas, Charles E. Geyer, Priya Rastogi, Peter C. Lucas, Soonmyung Paik, C. Kent Osborne, Norman Wolmark, Mothaffar F. Rimawi. Association of molecular signatures, mutations, and sTILs, with pCR in breast cancer patients in NRG Oncology/NSABP B-52 [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4064.

3503 Background: MOSAIC and C-07 showed that oxaliplatin (OX) added to 5-fluorouracil plus leucovorin significantly improved disease free survival (DFS). However, OX is associated with neurotoxicity and the vast majority of patients do not receive OX-benefit; highlighting the importance of an OX-benefit predictor. In C-07, colon tumors with a CRCA (Colorectal Cancer Assigner) stem-like subtype were associated with a poor prognosis and no OX-benefit, but stage III patients with an enterocyte subtype did. We tested the association of CRCA and CMS molecular subtypes with prognosis and OX-benefit in stage III MOSAIC patients. Also, recombination proficiency scores (RPS) were tested for the same associations. RPS scores quantify the efficiency of DNA-damage repair. Low RPS have been associated with inferior overall survival (OS) in non-small cell lung carcinoma patients treated with surgery alone but better OS for patients who received surgery plus chemotherapy. Methods: Gene expression profiles from 590 stage III pts with follow up were successfully profiled by a custom designed nCounter code set. CRCA subtypes were determined by a locked down algorithm based on a re-estimated centroid using 72 genes (Song et al 2016). CMS subtypes were determined by modified single sample predictor (SSP) using 84 genes. RPS scores were determined as previously described by Pitroda et al (2014). Signature predictions were made while blinded to clinical outcome and signature performance was evaluated while blinded to gene expression. Results: The stem-like subtype was associated with a very poor prognosis (Stem-like vs others HR=1.56, p<0.01) and no OX-benefit. CRCA and CMS subtypes did not associate with OX-benefit. Using a median cut point, stage III patients with low RPS scores received significant OX-benefit (HR=0.67, p=0.033, N=290) and patients with high RPS scores did not (HR=1.2, p=0.32, N=300) with significant interaction p=0.025. Conclusions: The observation in C-07, that the stem-like subtype has the poorest prognosis and did not receive OX-benefit, was validated in MOSAIC, identifying patients who need new therapies. RPS scores may help to identify the subset of patients with OX-benefit in stage III CC, confirmation of this observation is currently being investigated in C-07. Support: PA DoH; NSABP; Sanofi.