Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

pAKT pathway activation is associated with PIK3CA mutations and good prognosis in luminal breast cancer in contrast to p-mTOR pathway activation

pAKT pathway activation is associated with PIK3CA mutations and good prognosis in luminal breast... www.nature.com/npjbcancer ARTICLE OPEN pAKT pathway activation is associated with PIK3CA mutations and good prognosis in luminal breast cancer in contrast to p-mTOR pathway activation 1 2 2 2 2 Amir Sonnenblick , David Venet , Sylvain Brohée , Noam Pondé and Christos Sotiriou Numerous studies have focused on the PI3K/AKT/mTOR pathway in estrogen receptor positive (ER) breast cancer (BC), as a linear signal transduction pathway and reported its association with worse clinical outcomes. We developed gene signatures that reflect the level of expression of phosphorylated-Serine473-AKT (pAKT) and phosphorylated-Serine2448-mTOR (p-mTOR) separately, capturing their corresponding level of pathway activation. Our analysis revealed that the pAKT pathway activation was associated −10 with luminal A BC while the p-mTOR pathway activation was more associated with luminal B BC (Kruskal–Wallis test p <10 ). pAKT pathway activation was significantly associated with better outcomes (multivariable HR, 0.79; 95%CI, 0.74–0.85; p = 2.5 × −10 10 ) and PIK3CA mutations (p = 0.0001) whereas p-mTOR pathway activation showed worse outcomes (multivariable HR,1.1; 95% −4 CI, 1.1–1.2; p = 9.9 × 10 ) and associated with p53 mutations (p = 0.04). in conclusion, our data show that pAKT and p-mTOR pathway activation have differing impact on prognosis and suggest that they are not linearly connected in luminal breast cancers. npj Breast Cancer (2019) 5:7 ; https://doi.org/10.1038/s41523-019-0102-1 INTRODUCTION monotherapy. Therefore, to gain better insight into the relative contribution of each of the signaling pathways which lie down- The phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR-signaling stream to PI3K (namely AKT and mTOR) to BC outcomes, we have pathway mediates key cellular functions, including growth, developed a novel in silico approach which assessed the proliferation, and survival and is frequently involved in carcino- activation of each of these signaling pathways separately, by genesis, tumor progression, and metastases. Numerous studies integrating reverse phase protein array (RPPA) and matched gene have focused on the PI3K/AKT/mTOR pathway in estrogen expression. receptor positive (ER-positive) breast cancer (BC) and have shown that PIK3CA mutations are frequent, that the PI3K/AKT/mTOR- signaling pathway is often dysregulated and that both correlate 2–4 RESULTS with worse clinical outcomes. As a consequence, a large pAKT pathway activated and p-mTOR pathway activated ER- number of drugs targeting the various components of this positive early BCs are associated with distinct and exclusive gene pathway have been developed. Everolimus (an mTOR inhibitor) expression profiles is currently the only approved drug targeting mTOR based on the results of the BOLERO-2 trial. We first derived two distinct signatures whose expression levels While AKT is activated by phospholipid binding and activation could predict AKT and mTOR pathway activation through pAKT loop phosphorylation at Threonine308 by PDK1 and by phosphor- and p-mTOR RPPA levels by computing the differentially ylation within the carboxy terminus at Serine473, mTOR is expressed genes between tumor samples with high and low phosphorylated at Serine2448 via the PI3K-signaling pathway. RPPA levels of pAKT (respectively, activated and inactivated AKT AKT activates the mTOR complex 1 (mTORC1) which in addition to pathway) and p-mTOR proteins (respectively, activated and mTOR contains mLST8, PRAS40, and RAPTOR. This activation inactivated mTOR pathway), using ER-positive tumors from the involves phosphorylation of tuberous sclerosis complex 2 (TSC2), TCGA repository. It is important to note that the two signatures which blocks the ability of TSC2 to act as a GTPase-activating did not share any common genes (Fig. 1a). We next sought to protein, thereby allowing accumulation of Rheb-GTP and mTORC1 assess their biological and clinical relevance in BC. Firstly, we activation. AKT can also activate mTORC1 by PRAS40 phosphor- compared both signatures to the reference classes of the Gene ylation, thereby relieving the PRAS40-mediated inhibition of Ontology and the mSigDB signatures repositories using the Broad 9 10 mTORC1. Institute site. This showed that the pAKT signature was The PI3K/AKT/mTOR pathway is usually considered as a linear significantly enriched in genes up-regulated in less aggressive 11 −27 signal transduction pathway in BC, however in the ER-positive invasive BC tumors (e.g. grade 1 vs. grade 3 ; fdr = 2×10 ). In disease, we have previously shown that PIK3CA mutations were contrast, the p-mTOR signature was enriched in genes expressed 11,12 associated with relatively low mTORC1 functional output and with in mammary stem cells and more aggressive luminal B cancers −7 −5 good outcomes in patients who received adjuvant tamoxifen (fdr = 2×10 , fdr = 3× 10 , respectively). A network clustering 1 2 Oncology Division, Tel Aviv Sourasky Medical Center, and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel and Breast Cancer Translational Research Laboratory, Institut Jules Bordet, Université Libre de Bruxelles, Brussels, Belgium Correspondence: Amir Sonnenblick (amirsonn@gmail.com) or Christos Sotiriou (christos.sotiriou@bordet.be) Received: 27 May 2018 Accepted: 8 January 2019 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. Fig. 1 pAKT and p-mTOR signatures derived from the TCGA. a Venn diagram shows no intersection between the pAKT and p-mTOR gene signatures. b Network representation of the gene signatures. Each node represents the genes up-regulated or down-regulated in the signature. Edges show signatures sharing a significant number of genes. Network clustering shows the tendency of these signatures to cluster together according to their proliferation status. c Integrated analysis of the PIK3CA/pAKT/m-TOR pathway in the TCGA. Luminal breast cancer subtypes differ by pAKT and p-mTOR activity. The panel includes a protein-based (RPPA) proteomic status. Tumors were ordered first by mRNA subtype (luminal A versus B). P values were calculated using the Mann–Whitney test −10 analysis using the pAKT and p-mTOR signatures, as well as other 10 ) (Fig. 2). We next assessed whether pAKT and p-mTOR RPPA-derived signatures that demonstrated a significant intersec- pathway activation were correlated with outcomes (RFS) in ER- tion with them, identified two main sub-networks according positive patients with relapse data available. As shown in Figs. 3 mainly to their proliferation status, namely pAKT-high/p-mTOR- and 5, pAKT pathway activation was significantly associated with low and pAKT-low/p-mTOR high characterized with low and high better outcomes in all luminal patients (multivariable HR, 0.79; −10 proliferation levels, respectively (Fig. 1b). These observations were 95% CI, 0.74–0.85; p = 2.5 × 10 ). Similar results were obtained confirmed when analyzing the TCGA RPPA dataset (Figs. 1c and with a dataset consisting of patients treated with endocrine S1). AKT pathway was more often activated (elevated pAKT therapy only (multivariable HR, 0.82; 95% CI, 0.73–0.93; p = 0.002). expression) in luminal A cancers whereas mTOR pathway was Indeed, patients with pAKT pathway activation had better more often activated (elevated p-mTOR and pS6, an mTOR outcomes irrespective of their specific subtype (luminal A multi- downstream target), in luminal B subtypes (Figs. 1c and S2). Next, variable HR, 0.85; 95% CI, 0.75–0.96; p = 0.01; luminal B HR, 0.91; we sought to determine how the pAKT and p-mTOR signatures 95% CI, 0.83–0.99; p = 0.033). In contrast, patients with p-mTOR correlate with other signatures and RPPA markers of the pathway. pathway activation had significantly worse outcomes in all luminal As shown in Fig. S3, the pAKT signature negatively correlates with −4 patients (multivariable HR, 1.1; 95% CI, 1.1–1.2; p = 9.9 × 10 ) and downstream effectors of the pathway while the p-mTOR signature this remained true when tested in the dataset consisting of positively correlates with them. patients treated with endocrine therapy only (multivariable HR, Altogether, these results demonstrate that the pAKT and p- 1.2; 95% CI, 1.1–1.4; p = 0.004) (Figs. 4 and 5). mTOR pathways, assessed through these RPPA-based gene Next, we assessed the association between the PIK3CA and P53 expression signatures, have exclusive distribution according to mutation status and pAKT and p-mTOR pathway activation in an luminal molecular subtypes and are not necessarily linearly independent set, namely the TCGA BC patients with RNA- connected. sequence gene expression data that were not used to design the signatures. While the pAKT signature was associated with Association of the pAKT and p-mTOR pathway activation with PIK3CA mutations (p = 0.0001), the p-mTOR signature was not clinical outcome in patients with ER-positive early BC (p = 0.22) (Fig. 6a, b). The opposite was true for P53 mutations, To ascertain the impact of each pathway on outcomes in ER- which were positively correlated with p-mTOR pathway activation positive BC, we applied the pAKT and p-mTOR signatures on a (p = 0.04), and negatively correlated with pAKT (p = 0.0003) dataset composed of 38 publicly available microarray datasets. We (Fig. 6i, j). Analysis of the PIK3CA mutations by exon led to similar first assessed whether pAKT or p-mTOR pathway activation were results (Fig. 6), although mutations outside of exons 9 and associated with any particular luminal subtype. As expected, in the 20 seemed less associated with pAKT pathway activation. pooled set analysis pAKT pathway activation was significantly Finally, in an effort to identify whether these signatures could −10 associated will luminal A cancers (p <10 ) whereas p-mTOR predicts response to mTOR inhibitors, we evaluated another data pathway activation was associated with luminal B cancers (p < set of neo-adjuvant patients treated with Everolimus. Analysis of npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation 1234567890():,; A. Sonnenblick et al. Fig. 2 pAKT (right) and p-mTOR (left) gene signatures expression in publicly available microarray datasets according to the PAM50 breast cancer subtype. Kruskal–Wallis p-value is shown Fig. 3 High pAKT gene signature expression is associated with good prognosis in the luminal subtype. a–f We assessed the prognostic value of tertiles of pAKT gene signature expression in: a all luminal patients treated or not treated (n = 3073), b luminal A (n = 1491), c Luminal B (n = 1582), d all luminal treated with only hormonal therapy (n= 1180), e luminal A treated with only hormonal therapy (n = 491), and f luminal B treated with only hormonal therapy (n= 689). Significance (p-value) of differences in survival between patient groups defined by tertiles of pAKT signature expression is estimated by log-rank test. The analysis presented includes patients with lymph node-negative and lymph node-positive cancers Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 A. Sonnenblick et al. Fig. 4 High p-mTOR gene signature expression is associated with bad prognosis in the luminal subtype. a–f We assessed the prognostic value of tertiles of p-mTOR gene signature expression in: a all luminal patients treated or not treated (n = 3073), b luminal A (n = 1491), c Luminal B (n = 1582), d all luminal treated with only hormonal therapy (n = 1180), e luminal A treated with only hormonal therapy (n = 491), and f luminal B treated with only hormonal therapy (n = 689). Significance (p values) of differences in survival between patient groups defined by tertiles of p-mTOR signature expression is estimated by log-rank test. The analysis presented includes patients with lymph node-negative and lymph node-positive cancers the correlations between the effectiveness of this treatment and tumorigenic effects. Between 30% and 40% of BCs, especially the developed signatures suggests as expected that the pAKT ER-positive tumors, have mutations in PIK3CA. The vast majority signature is associated with less response to Everolimus (r = 0.45; of the PIK3CA mutations are missense mutations which are p = 0.031, Fig. S4). positioned in the helical domain (exon 9, mostly: E545K and Overall, our data suggests that pAKT and p-mTOR pathway E542K) and the kinase domain (exon 20, mostly H1047R) in activation as assessed through the respective signatures, despite hotspot clusters. These mutations have direct effect on AKT being major components of the same overarching pathway (PI3K), phosphorylation. The effect of PIK3CA mutations/pAKT on prog- have distinctly different impacts on disease biology and conse- nosis is mixed in early BC. We found that exons 9 and 20 quently on outcomes in early disease. mutations in PIK3CA were more associated with pAKT than mutations in other exons. We previously reported that PIK3CA mutations were associated DISCUSSION with improved outcome and low levels of signaling through the 4,17 The goal of the present study was to better understand the mTOR pathway in BC. Several possible hypotheses were raised distinct contribution to disease biology and clinical outcomes of regarding the reasons for this. Some data available on PP2A and signaling through the AKT and mTOR downstream pathways, PML, both known to have an inhibitory effect on both AKT and 18,19 which typically occur as part of PI3K pathway activation in luminal mTOR, have suggested that they may be upregulated in BCs. We found that pAKT and p-mTOR were differentially PIK3CA-activated tumors. Negative feedback regulation in PI3K- expressed according to luminal subtypes, implying different mediated cells through the insulin receptor substrate and degrees of pathway activation, and that, more importantly, the relatively weak pathway activation in PIK3CA-mutated cancers pathways were not linearly connected. Additionally, we found that have also been suggested as possible explanations for low levels pAKT pathway activation was positively associated with PIK3CA of signaling through mTOR in ER-positive BC. mutations whereas the opposite was observed with p-mTOR According to our findings only pAKT pathway activation was pathway activation. In contrast, pAKT pathway activation was found to be significantly different between the luminal subtypes associated with good clinical outcome despite its known (A and B) and PIK3CA wildtype versus mutant, whereas p-mTOR npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. N HR CI Akt All 3997 0.72 0.68 to 0.76 0 −4 Akt LumA 1953 0.83 0.75 to 0.92 5 × 10 −5 Akt LumB 2044 0.85 0.79 to 0.91 1 × 10 −8 Akt Hormonal therapy 1234 0.73 0.66 to 0.82 1.5 × 10 Akt LumA Hormonal therapy 595 0.8 0.65 to 0.99 0.037 Akt LumB Hormonal therapy 639 0.92 0.8 to 1 0.21 −13 Akt No hormonal therapy 1392 0.72 0.65 to 0.79 8.9 × 10 Akt LumA No hormonal therapy 678 0.82 0.7 to 0.95 0.0091 −4 Akt LumB No hormonal therapy 714 0.8 0.71 to 0.9 2.5 × 10 −9 mTOR All 3997 1.2 1.1 to 1.3 1.2 × 10 mTOR LumA 1953 1.1 1 to 1.3 0.018 mTOR LumB 2044 1 0.92 to 1.1 0.93 −5 mTOR Hormonal therapy 1234 1.3 1.1 to 1.4 9.2 × 10 mTOR LumA Hormonal therapy 595 1.2 0.95 to 1.5 0.14 mTOR LumB Hormonal therapy 639 1 0.88 to 1.2 0.83 −4 mTOR No hormonal therapy 1392 1.2 1.1 to 1.3 9.9 × 10 mTOR LumA No hormonal therapy 678 1.2 0.99 to 1.4 0.074 mTOR LumB No hormonal therapy 714 0.98 0.87 to 1.1 0.76 0.5 1 2 HR B p N HR CI −10 Akt All 3248 0.79 0.74 to 0.85 2.5 × 10 Akt LumA 1561 0.85 0.75 to 0.96 0.01 Akt LumB 1687 0.91 0.83 to 0.99 0.033 Akt Hormonal therapy 1158 0.82 0.73 to 0.93 0.0021 Akt LumA Hormonal therapy 546 0.86 0.69 to 1.1 0.21 Akt LumB Hormonal therapy 612 0.98 0.85 to 1.1 0.77 −5 Akt No hormonal therapy 1079 0.78 0.69 to 0.88 6.8 × 10 Akt LumA No hormonal therapy 513 0.81 0.66 to 1 0.052 Akt LumB No hormonal therapy 566 0.9 0.77 to 1 0.17 −4 mTOR All 3248 1.1 1.1 to 1.2 9.9 × 10 mTOR LumA 1561 1.1 0.96 to 1.2 0.18 mTOR LumB 1687 0.99 0.91 to 1.1 0.77 mTOR Hormonal therapy 1158 1.2 1.1 to 1.4 0.0041 mTOR LumA Hormonal therapy 546 1.2 0.94 to 1.5 0.13 mTOR LumB Hormonal therapy 612 1 0.89 to 1.2 0.66 mTOR No hormonal therapy 1079 1.1 0.97 to 1.2 0.15 mTOR LumA No hormonal therapy 513 1.2 0.96 to 1.5 0.1 mTOR LumB No hormonal therapy 566 0.89 0.77 to 1 0.11 0.5 1 2 HR Fig. 5 Forest plots showing the hazard ratios of the recurrence free survival of pAKT and p-mTOR gene signatures treated as a continuous variable using Cox univariate a and multivariable analysis b, in the pooled analysis. For multivariate analysis, we considered the following variables: age, tumor size, grade, and nodal status. Signatures with nominal significant effect (p < 0.05) are shown in blue pathway activation was not significant for both. The inconclusive luminal A/PIK3CA mutations and good prognosis, while p-mTOR/ and relative activation of p-mTOR by the mutant PIK3CA may be pS6 is not, suggesting that the presence of HER2 and basal also attributed to the different roles and activators of mTOR and subtype in the primary analysis of the TCGA masked these the fact that mTOR is at the cross section of multiple signaling observations. pathways. Several studies have clearly demonstrated that mTOR is In conclusion our data suggest that the AKT and mTOR a direct substrate for the AKT kinase and identified Serine2448 as pathways are not linearly connected in luminal BCs. pAKT pathway the AKT target site in mTOR. However, additional studies have activation is associated with PIK3CA mutations, luminal A and demonstrated that rapamycin, an inhibitor of mTOR function, good prognosis, while p-mTOR pathway activation is associated blocks serum-stimulated Serine2448 phosphorylation of mTOR in with luminal B, P53 mutations, and bad prognosis. These results an AKT-independent manner and identified S6 kinase as a major may have important clinical implications considering that in low p- effector of mTOR phosphorylation at Serine2448. Indeed, our mTOR BCs, treatment with mTOR inhibitors, such as everolimus, analysis of the TCGA data shows that the S6 kinase (downstream which is highly toxic, will possibly be of lower value since the of mTOR) is associated with luminal B and P53 mutations pathway is not activated. Additionally, pAKT pathway activation, suggesting that while mTOR itself is at the cross section of as measured through our gene signature, can add to presently conflicting pathways its downstream targets are not PIK3CA used outcome prediction tools in both luminal A and luminal B dependent. In addition, there are alternative kinases that can tumors. activate the mTOR pathway independently of AKT, such as RSK which leads to phosphorylation of TSC resulting in increased mTOR signaling and the PDK1–SGK1 axis that can sustain mTOR METHODS 23–25 activity upon AKT suppression. Computation of RPPAs-based signatures The primary TCGA report, which investigated all BC subtypes, We downloaded clinic-pathological, normalized gene expression and RPPA confirmed a high frequency of PIK3CA mutations in luminal BC. data from the publicly available TCGA repository using its online Multiple platforms, which examined the relationship between 26 bioinformatics tools (Fig. S1 flow chart). ER-positive early BCs were PIK3CA mutation and protein expression, have demonstrated that analyzed based on the RPPA proteomic levels. 265 samples with available pAKT and pS6 were not elevated in PIK3CA-mutated luminal gene expression and RPPA data were considered as luminal (166 Luminal A cancers; instead, they were highly expressed in basal-like and and 99 Luminal B according to PAM50 computed on the cBioPortal HER2 subtypes. Our dataset, which is restricted to luminal cancers, website ). To identify the genes that were differentially expressed demonstrated that pAKT pathway activation is associated with between the low and high expression groups, and to find the genes that Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 A. Sonnenblick et al. 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.00012 p = 0.22 −0.2 No PIK3CA mutation PIK3CA mutation No PIK3CA mutation PIK3CA mutation 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.013 p = 0.19 −0.2 No PIK3CA−ex9 mut PIK3CA−ex9 mut No PIK3CA−ex9 mut PIK3CA−ex9 mut 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.69 p = 0.0082 −0.2 No PIK3CA−ex20 mut PIK3CA−ex20 mut No PIK3CA−ex20 mut PIK3CA−ex20 mut 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.99 p = 0.75 −0.2 No PIK3CA−oth mut PIK3CA−other mut No PIK3CA−oth mut PIK3CA−other mut 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.00034 p = 0.04 −0.2 No P53 mutation P53 mutation No P53 mutation P53 mutation Fig. 6 Expression of pAKT (a, c, e, g, i) and p-mTOR (b–d, f, h, j) gene signatures levels in PIK3CA (all, exon9, exon20 or others) and P53 mutated and wild type samples in an independent RNA sequencing set (n = 309) would optimize the predictive power of our signatures, we used a machine Code availability learning approach as previously described. After this process, we were The expression levels of the signatures in the gene expression datasets left with 69 signatures (Supplementary Data) presenting a relevant AUC for were computed as previously described. In brief, we evaluated using a proteomic status prediction. Among others, p-mTOR achieved an AUC of nested 10-fold cross validation the maximal Benjamini–Hochberg false −6 −11 0.71 (p ~10 ) and pAKT an AUC of 0.77 (p ~10 ) in both luminal A and discovery rate and the minimal gene fold change that would optimize the B cancers. ability of the differentially expressed genes to predict the high/low status npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. Gene-expression data and statistical analyses Table 1. The sources and locations for the 38 gene expression We analyzed 38 gene expression datasets totaling more than 7000 tumors datasets analyzed (detailed in Table 1). To ensure comparability of expression values across multiple data sets, a 0.95 quantile normalization was performed. Dataset No. of Permanent identifier References Differences in expression of pAKT and p-mTOR signatures according to patients subtype were examined using the Kruskal–Wallis test. Survival outcome 31,32 data are presented as recurrence free survival (RFS). Survival plots NKI 337 10.1038/415530a according to the pAKT and p-mTOR signatures tertiles were drawn using 33,34 UCSF 162 GSE123833 the Kaplan–Meier method. Association of the signatures (i.e. pathway STNO2 122 GSE4335 activation) with good or bad outcomes were computed using uni-variate or multi-variate Cox regression analyses and data were presented as forest NCI 99 10.1073/pnas.1732912100 plots. For multivariate analysis, we considered the following variables: age, UNC4 337 GSE18229 tumor size, grade, and nodal status. To assess the correlation between the CAL 118 E-TABM-158 PIK3CA mutation status and AKT and p-mTOR gene pathway activation, as ‡ 39,40 represented by the gene signature scores, we analyzed the TCGA cohort of MDA4 129 (65) GSE123832 RNA sequenced data that was not used for the computation of the KOO 88 GSE123831 signatures (309 samples), and for which both mutational and gene HLP 53 E-TABM-543 expression data were available. Each sample was considered as mutated or 43 not (so a sample with four mutations was considered just like a sample EXPO 353 GSE2109 with one mutation). All mutations were taken into account. PIK3CA 44,45 VDX 344 GSE2034/GSE5327 mutations were also analyzed by specific exons (exons 9, 20, and all others MSK 99 GSE2603 grouped together). ‡ 47 UPP 251 (190) GSE3494 Reporting summary STK 159 GSE1456 ‡ 36,49 Further information on experimental design is available in the Nature UNT 137 (92) GSE2990 Research Reporting Summary linked to this article. DUKE 171 GSE3143 TRANSBIG 198 GSE7390 DATA AVAILABILITY DUKE2 160 GSE6961 The sources and locations for the 38 gene expression datasets analyzed during the MAINZ 200 GSE11121 current study are available in Table 1 and the figshare repository https://doi.org/ LUND2 105 GSE5325 10.6084/m9.figshare.7461776. LUND 143 GSE5325 FNCLCC 150 GSE7017 57 ACKNOWLEDGEMENTS EMC2 204 GSE12276 We would like to thank Carolyn Straehle for her editorial assistance. A.S. is supported MUG 152 GSE10510 by a Clinical Research Career Development Award from the Israel Cancer Research NCCS 183 GSE5364 Fund grants (16-116-CRCDA) and from the Israeli Cancer Research Association (2017- 0140). A.S. was an ESMO translational research fellow. C.S. is supported by the Breast MCCC 75 GSE19177 Research Cancer Foundation (BCRF). EORTC10994 49 GSE1561 DFHCC 115 GSE19615 ‡ 63 AUTHOR CONTRIBUTIONS DFHCC2 84 (75) GSE18864 ‡ 64 A.S. conceived of the study and participated in its design and coordination, analyzed DFHCC3 40 (26) GSE3744 and interpreted the data and wrote the manuscript. S.B. analyzed and interpreted the DFHCC4 129 GSE5460 data, performed the statistical analysis, and wrote the manuscript. N.P. coordinated MAQC2 230 GSE20194 the analysis and drafted the manuscript. D.V. carried out the statistical analysis ‡ 67 analyzed and interpreted the data and drafted the manuscript. C.S. participated in TAM 345 (242) GSE6532/GSE9195 the study design and coordination, analyzed, interpreted the data, and wrote the MDA5 298 GSE17705 manuscript. All authors read and approved the final manuscript. VDX3 136 GSE12093 PNC 248 GSE20713 ADDITIONAL INFORMATION TCGA 517 https://tcga-data.nci.nih.gov/ Supplementary information accompanies the paper on the npj Breast Cancer docs/publications/brca_2012/ website (https://doi.org/10.1038/s41523-019-0102-1). METABRIC 1643 EGAS00000000083 Competing interests: The authors declare no competing interests. Duplicated patients were removed from few datasets for the estimation of concordance and prognostic value Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. of the RPPA in luminal A and B patients together and separately. While the parameters were selected in a 10-fold cross validation, the procedure itself was assessed using a nested cross validation. All analyses were performed REFERENCES using the genefu package of the R (v3.2)/bioconductor (v1.18) statistical 1. Hennessy, B. T., Smith, D. L., Ram, P. T., Lu, Y. & Mills, G. B. Exploiting the PI3K/AKT suite. pathway for cancer drug discovery. Nat. Rev. Drug Discov. 4, 988–1004 (2005). 2. Koboldt, D. C. et al. Comprehensive molecular portraits of human breast tumours. Nature 490,61–70 (2012). 3. Campbell, I. G. Mutation of the PIK3CA gene in ovarian and breast cancer. Cancer Network analysis and clustering Res. 64, 7678–7681 (2004). 4. Loi, S. et al. PIK3CA mutations associated with gene signature of low Network analysis and clustering was performed as previously described. mTORC1 signaling and better outcomes in estrogen receptor-positive breast The MCL graph clustering algorithm was applied. Visualization of the cancer. Proc. Natl Acad. Sci. USA 107, 10208–10213 (2010). network has been rendered using the yED software. Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 A. Sonnenblick et al. 5. Dienstmann, R., Rodon, J., Serra, V. & Tabernero, J. Picking the point of inhibition: 34. Korkola, J. E. et al. Differentiation of lobular versus ductal breast carcinomas by a comparative review of PI3K/AKT/mTOR pathway inhibitors. Mol. Cancer Ther. 13, expression microarray analysis. Cancer Res. 63, 7167–7175 (2003). 1021–1031 (2014). 35. Sorlie, T. et al. Repeated observation of breast tumor subtypes in independent 6. Baselga, J. et al. Everolimus in postmenopausal hormone-receptor–positive gene expression data sets. Proc. Natl Acad. Sci. USA 100, 8418–8423 (2003). advanced breast cancer. N. Engl. J. Med. 366, 520–529 (2012). 36. Sotiriou, C. et al. Breast cancer classification and prognosis based on gene 7. Navé, B. T., Ouwens, M., Withers, D. J., Alessi, D. R. & Shepherd, P. R. Mammalian expression profiles from a population-based study. Proc. Natl Acad. Sci. USA 100, target of rapamycin is a direct target for protein kinase B: identification of a 10393–10398 (2003). convergence point for opposing effects of insulin and amino-acid deficiency on 37. Prat, A. et al. Phenotypic and molecular characterization of the claudin-low protein translation. Biochem. J. 344, 427–431 (1999). intrinsic subtype of breast cancer. Breast Cancer Res. 12, R68 (2010). 8. Guertin, D. A. & Sabatini, D. M. Defining the role of mTOR in cancer. Cancer Cell 38. Chin, K. et al. Genomic and transcriptional aberrations linked to breast cancer 12,9–22 (2007). pathophysiologies. Cancer Cell 10, 529–541 (2006). 9. Manning, B. D. & Cantley, L. C. AKT/PKB signaling: navigating downstream. Cell 39. Hess, K. R. et al. Pharmacogenomic predictor of sensitivity to preoperative che- 129, 1261–1274 (2007). motherapy with paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide 10. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based in breast cancer. J. Clin. Oncol. 24, 4236–4244 (2006). approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. 40. Liedtke, C. et al. Response to neoadjuvant therapy and long-term survival in USA 102, 15545–15550 (2005). patients with triple-negative breast cancer. J. Clin. Oncol. 26, 1275–1281 (2008). 11. Sotiriou, C. et al. Gene expression profiling in breast cancer: understanding the 41. Huang, E. et al. Gene expression predictors of breast cancer outcomes. Lancet molecular basis of histologic grade to improve prognosis. J. Natl. Cancer Inst. 98, Lond. Engl. 361, 1590–1596 (2003). 262–272 (2006). 42. Natrajan, R. et al. An integrative genomic and transcriptomic analysis reveals 12. Lim, E. et al. Transcriptome analyses of mouse and human mammary cell sub- molecular pathways and networks regulated by copy number aberrations in populations reveal multiple conserved genes and pathways. Breast Cancer Res. basal-like, HER2 and luminal cancers. Breast Cancer Res. Treat. 121, 575–589 12, R21 (2010). (2010). 13. Sabine, V. S. et al. Gene expression profiling of response to mTOR inhibitor 43. Bittner M. (2005) Expression Project for Oncology (expO). www.intgen.org/expo. everolimus in pre-operatively treated post-menopausal women with oestrogen Accessed 3 Nov 2010. receptor-positive breast cancer. Breast Cancer Res. Treat. 122, 419–428 (2010). 44. Wang, Y. et al. Gene-expression profiles to predict distant metastasis of lymph- 14. Curtis, C. et al. The genomic and transcriptomic architecture of 2,000 breast node-negative primary breast cancer. Lancet Lond. Engl. 365, 671–679 (2005). tumours reveals novel subgroups. Nature 486, 346–352 (2012). 45. Minn, A. J. et al. Lung metastasis genes couple breast tumor size and metastatic 15. Samuels, Y. et al. High frequency of mutations of the PIK3CA gene in human spread. Proc. Natl Acad. Sci. USA 104, 6740–6745 (2007). cancers. Science 304, 554 (2004). 46. Minn, A. J. et al. Genes that mediate breast cancer metastasis to lung. Nature 436, 16. Yang, S. X., Polley, E. & Lipkowitz, S. New insights on PI3K/AKT pathway altera- 518–524 (2005). tions and clinical outcomes in breast cancer. Cancer Treat. Rev. 45,87–96 (2016). 47. Miller, L. D. et al. An expression signature for p53 status in human breast cancer 17. Kalinsky, K. et al. PIK3CA mutation associates with improved outcome in breast predicts mutation status, transcriptional effects, and patient survival. Proc. Natl cancer. Clin. Cancer Res. 15, 5049–5059 (2009). Acad. Sci. USA 102, 13550–13555 (2005). 18. Trotman, L. C. et al. Identification of a tumour suppressor network opposing 48. Pawitan, Y. et al. Gene expression profiling spares early breast cancer patients nuclear Akt function. Nature 441, 523–527 (2006). from adjuvant therapy: derived and validated in two population-based cohorts. 19. Bernardi, R. et al. PML inhibits HIF-1alpha translation and neoangiogenesis Breast Cancer Res. 7, R953–R964 (2005). through repression of mTOR. Nature 442, 779–785 (2006). 49. Loi, S. et al. Definition of clinically distinct molecular subtypes in estrogen 20. Paz, K. et al. A molecular basis for insulin resistance. Elevated serine/threonine receptor-positive breast carcinomas through genomic grade. J. Clin. Oncol. 25, phosphorylation of IRS-1 and IRS-2 inhibits their binding to the juxtamembrane 1239–1246 (2007). region of the insulin receptor and impairs their ability to undergo insulin-induced 50. Bild, A. H. et al. Oncogenic pathway signatures in human cancers as a guide to tyrosine phosphorylation. J. Biol. Chem. 272, 29911–29918 (1997). targeted therapies. Nature 439, 353–357 (2006). 21. Sekulić, A. et al. A direct linkage between the phosphoinositide 3-kinase-AKT 51. Desmedt, C. et al. Strong time dependence of the 76-gene prognostic signature for node-negative breast cancer patients in the TRANSBIG multicenter inde- signaling pathway and the mammalian target of rapamycin in mitogen- stimulated and transformed cells. Cancer Res. 60, 3504 (2000). pendent validation series. Clin. Cancer Res. 13, 3207–3214 (2007). 22. Chiang, G. G. & Abraham, R. T. Phosphorylation of mammalian target of rapa- 52. Bonnefoi, H. et al. Validation of gene signatures that predict the response of mycin (mTOR) at Ser-2448 is mediated by p70S6 kinase. J. Biol. Chem. 280, breast cancer to neoadjuvant chemotherapy: a substudy of the EORTC 10994/BIG 25485–25490 (2005). 00-01 clinical trial. Lancet Oncol. 8, 1071–1078 (2007). 23. Roux, P. P., Ballif, B. A., Anjum, R., Gygi, S. P. & Blenis, J. Tumor-promoting phorbol 53. Schmidt, M. et al. The humoral immune system has a key prognostic impact in esters and activated Ras inactivate the tuberous sclerosis tumor suppressor node-negative breast cancer. Cancer Res. 68, 5405–5413 (2008). complex via p90 ribosomal S6 kinase. Proc. Natl Acad. Sci. USA 101, 13489–13494 54. Saal, L. H. et al. Poor prognosis in carcinoma is associated with a gene expression (2004). signature of aberrant PTEN tumor suppressor pathway activity. Proc. Natl Acad. 24. Castel, P. et al. PDK1-SGK1 signaling sustains AKT-independent mTORC1 activa- Sci. USA 104, 7564–7569 (2007). tion and confers resistance to PI3Kα inhibition. Cancer Cell 30, 229–242 (2016). 55. Niméus-Malmström, E. et al. Gene expression profiling in primary breast cancer 25. Ma, L., Chen, Z., Erdjument-Bromage, H., Tempst, P. & Pandolfi, P. P. Phosphor- distinguishes patients developing local recurrence after breast-conservation ylation and functional inactivation of TSC2 by Erk implications for tuberous surgery, with or without postoperative radiotherapy. Breast Cancer Res. 10, R34 sclerosis and cancer pathogenesis. Cell 121, 179–193 (2005). (2008). 26. Cancer Genome Atlas Network. Comprehensive molecular portraits of human 56. Campone, M. et al. Prediction of metastatic relapse in node-positive breast breast tumours. Nature 490,61–70 (2012). cancer: establishment of a clinicogenomic model after FEC100 adjuvant regimen. 27. Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles Breast Cancer Res. Treat. 109, 491–501 (2008). using the cBioPortal. Sci. Signal. 6, pl1 (2013). 57. Bos, P. D. et al. Genes that mediate breast cancer metastasis to the brain. Nature 28. Sonnenblick, A. et al. p-STAT3 in luminal breast cancer: integrated RNA-protein 459, 1005–1009 (2009). pooled analysis and results from the BIG 2-98 phase III trial. Int. J. Oncol. 52, 58. Calabrò, A. et al. Effects of infiltrating lymphocytes and estrogen receptor on 424–432 (2018). gene expression and prognosis in breast cancer. Breast Cancer Res. Treat. 116, 29. Enright, A. J., Van Dongen, S. & Ouzounis, C. A. An efficient algorithm for large- 69–77 (2009). scale detection of protein families. Nucleic Acids Res. 30, 1575–1584 (2002). 59. Yu, K. et al. A precisely regulated gene expression cassette potently modulates 30. Sylvain, S. B., Sonnenblick, A., & Venet, D. Gene expression data sources for in metastasis and survival in multiple solid cancers. PLoS Genet. 4, e1000129 (2008). silico approach to assessing activation of AKT/mTOR signalling pathway signal- 60. Waddell, N. et al. Subtypes of familial breast tumours revealed by expression and ling pathways ER-positive early BC. figshare. Fileset https://doi.org/10.6084/m9. copy number profiling. Breast Cancer Res. Treat. 123, 661–677 (2010). figshare.7461776 (2018). 61. Farmer, P. et al. Identification of molecular apocrine breast tumours by micro- 31. van de Vijver, M. J. et al. A gene-expression signature as a predictor of survival in array analysis. Oncogene 24, 4660–4671 (2005). breast cancer. N. Engl. J. Med. 347, 1999–2009 (2002). 62. Li, Q. et al. Minimising immunohistochemical false negative ER classification 32. van’t Veer, L. J. et al. Gene expression profiling predicts clinical outcome of breast using a complementary 23 gene expression signature of ER status. PloS ONE 5, cancer. Nature 415, 530–536 (2002). e15031 (2010). 33. Korkola, J. E. et al. Identification of a robust gene signature that predicts breast 63. Silver, D. P. et al. Efficacy of neoadjuvant Cisplatin in triple-negative breast cancer. cancer outcome in independent data sets. BMC Cancer 7, 61 (2007). J. Clin. Oncol. 28, 1145–1153 (2010). npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. 64. Richardson, A. L. et al. X chromosomal abnormalities in basal-like human breast Open Access This article is licensed under a Creative Commons cancer. Cancer Cell 9, 121–132 (2006). Attribution 4.0 International License, which permits use, sharing, 65. Lu, X. et al. Predicting features of breast cancer with gene expression patterns. adaptation, distribution and reproduction in any medium or format, as long as you give Breast Cancer Res. Treat. 108, 191–201 (2008). appropriate credit to the original author(s) and the source, provide a link to the Creative 66. Popovici, V. et al. Effect of training-sample size and classification difficulty on the Commons license, and indicate if changes were made. The images or other third party accuracy of genomic predictors. Breast Cancer Res. 12, R5 (2010). material in this article are included in the article’s Creative Commons license, unless 67. Loi, S. et al. Predicting prognosis using molecular profiling in estrogen receptor- indicated otherwise in a credit line to the material. If material is not included in the positive breast cancer treated with tamoxifen. BMC Genom. 9, 239 (2008). article’s Creative Commons license and your intended use is not permitted by statutory 68. Symmans, W. F. et al. Genomic index of sensitivity to endocrine therapy for breast regulation or exceeds the permitted use, you will need to obtain permission directly cancer. J. Clin. Oncol. 28, 4111–4119 (2010). from the copyright holder. To view a copy of this license, visit http://creativecommons. 69. Zhang, Y. et al. The 76-gene signature defines high-risk patients that benefit from org/licenses/by/4.0/. adjuvant tamoxifen therapy. Breast Cancer Res. Treat. 116, 303–309 (2009). 70. Dedeurwaerder, S. et al. DNA methylation profiling reveals a predominant © The Author(s) 2019 immune component in breast cancers. EMBO Mol. Med. 3, 726–741 (2011). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png npj Breast Cancer Springer Journals

pAKT pathway activation is associated with PIK3CA mutations and good prognosis in luminal breast cancer in contrast to p-mTOR pathway activation

Loading next page...
 
/lp/springer-journals/pakt-pathway-activation-is-associated-with-pik3ca-mutations-and-good-RUnbM8666d

References (79)

Publisher
Springer Journals
Copyright
Copyright © 2019 by The Author(s)
Subject
Biomedicine; Biomedicine, general; Cancer Research; Oncology; Human Genetics; Cell Biology
eISSN
2374-4677
DOI
10.1038/s41523-019-0102-1
Publisher site
See Article on Publisher Site

Abstract

www.nature.com/npjbcancer ARTICLE OPEN pAKT pathway activation is associated with PIK3CA mutations and good prognosis in luminal breast cancer in contrast to p-mTOR pathway activation 1 2 2 2 2 Amir Sonnenblick , David Venet , Sylvain Brohée , Noam Pondé and Christos Sotiriou Numerous studies have focused on the PI3K/AKT/mTOR pathway in estrogen receptor positive (ER) breast cancer (BC), as a linear signal transduction pathway and reported its association with worse clinical outcomes. We developed gene signatures that reflect the level of expression of phosphorylated-Serine473-AKT (pAKT) and phosphorylated-Serine2448-mTOR (p-mTOR) separately, capturing their corresponding level of pathway activation. Our analysis revealed that the pAKT pathway activation was associated −10 with luminal A BC while the p-mTOR pathway activation was more associated with luminal B BC (Kruskal–Wallis test p <10 ). pAKT pathway activation was significantly associated with better outcomes (multivariable HR, 0.79; 95%CI, 0.74–0.85; p = 2.5 × −10 10 ) and PIK3CA mutations (p = 0.0001) whereas p-mTOR pathway activation showed worse outcomes (multivariable HR,1.1; 95% −4 CI, 1.1–1.2; p = 9.9 × 10 ) and associated with p53 mutations (p = 0.04). in conclusion, our data show that pAKT and p-mTOR pathway activation have differing impact on prognosis and suggest that they are not linearly connected in luminal breast cancers. npj Breast Cancer (2019) 5:7 ; https://doi.org/10.1038/s41523-019-0102-1 INTRODUCTION monotherapy. Therefore, to gain better insight into the relative contribution of each of the signaling pathways which lie down- The phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR-signaling stream to PI3K (namely AKT and mTOR) to BC outcomes, we have pathway mediates key cellular functions, including growth, developed a novel in silico approach which assessed the proliferation, and survival and is frequently involved in carcino- activation of each of these signaling pathways separately, by genesis, tumor progression, and metastases. Numerous studies integrating reverse phase protein array (RPPA) and matched gene have focused on the PI3K/AKT/mTOR pathway in estrogen expression. receptor positive (ER-positive) breast cancer (BC) and have shown that PIK3CA mutations are frequent, that the PI3K/AKT/mTOR- signaling pathway is often dysregulated and that both correlate 2–4 RESULTS with worse clinical outcomes. As a consequence, a large pAKT pathway activated and p-mTOR pathway activated ER- number of drugs targeting the various components of this positive early BCs are associated with distinct and exclusive gene pathway have been developed. Everolimus (an mTOR inhibitor) expression profiles is currently the only approved drug targeting mTOR based on the results of the BOLERO-2 trial. We first derived two distinct signatures whose expression levels While AKT is activated by phospholipid binding and activation could predict AKT and mTOR pathway activation through pAKT loop phosphorylation at Threonine308 by PDK1 and by phosphor- and p-mTOR RPPA levels by computing the differentially ylation within the carboxy terminus at Serine473, mTOR is expressed genes between tumor samples with high and low phosphorylated at Serine2448 via the PI3K-signaling pathway. RPPA levels of pAKT (respectively, activated and inactivated AKT AKT activates the mTOR complex 1 (mTORC1) which in addition to pathway) and p-mTOR proteins (respectively, activated and mTOR contains mLST8, PRAS40, and RAPTOR. This activation inactivated mTOR pathway), using ER-positive tumors from the involves phosphorylation of tuberous sclerosis complex 2 (TSC2), TCGA repository. It is important to note that the two signatures which blocks the ability of TSC2 to act as a GTPase-activating did not share any common genes (Fig. 1a). We next sought to protein, thereby allowing accumulation of Rheb-GTP and mTORC1 assess their biological and clinical relevance in BC. Firstly, we activation. AKT can also activate mTORC1 by PRAS40 phosphor- compared both signatures to the reference classes of the Gene ylation, thereby relieving the PRAS40-mediated inhibition of Ontology and the mSigDB signatures repositories using the Broad 9 10 mTORC1. Institute site. This showed that the pAKT signature was The PI3K/AKT/mTOR pathway is usually considered as a linear significantly enriched in genes up-regulated in less aggressive 11 −27 signal transduction pathway in BC, however in the ER-positive invasive BC tumors (e.g. grade 1 vs. grade 3 ; fdr = 2×10 ). In disease, we have previously shown that PIK3CA mutations were contrast, the p-mTOR signature was enriched in genes expressed 11,12 associated with relatively low mTORC1 functional output and with in mammary stem cells and more aggressive luminal B cancers −7 −5 good outcomes in patients who received adjuvant tamoxifen (fdr = 2×10 , fdr = 3× 10 , respectively). A network clustering 1 2 Oncology Division, Tel Aviv Sourasky Medical Center, and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel and Breast Cancer Translational Research Laboratory, Institut Jules Bordet, Université Libre de Bruxelles, Brussels, Belgium Correspondence: Amir Sonnenblick (amirsonn@gmail.com) or Christos Sotiriou (christos.sotiriou@bordet.be) Received: 27 May 2018 Accepted: 8 January 2019 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. Fig. 1 pAKT and p-mTOR signatures derived from the TCGA. a Venn diagram shows no intersection between the pAKT and p-mTOR gene signatures. b Network representation of the gene signatures. Each node represents the genes up-regulated or down-regulated in the signature. Edges show signatures sharing a significant number of genes. Network clustering shows the tendency of these signatures to cluster together according to their proliferation status. c Integrated analysis of the PIK3CA/pAKT/m-TOR pathway in the TCGA. Luminal breast cancer subtypes differ by pAKT and p-mTOR activity. The panel includes a protein-based (RPPA) proteomic status. Tumors were ordered first by mRNA subtype (luminal A versus B). P values were calculated using the Mann–Whitney test −10 analysis using the pAKT and p-mTOR signatures, as well as other 10 ) (Fig. 2). We next assessed whether pAKT and p-mTOR RPPA-derived signatures that demonstrated a significant intersec- pathway activation were correlated with outcomes (RFS) in ER- tion with them, identified two main sub-networks according positive patients with relapse data available. As shown in Figs. 3 mainly to their proliferation status, namely pAKT-high/p-mTOR- and 5, pAKT pathway activation was significantly associated with low and pAKT-low/p-mTOR high characterized with low and high better outcomes in all luminal patients (multivariable HR, 0.79; −10 proliferation levels, respectively (Fig. 1b). These observations were 95% CI, 0.74–0.85; p = 2.5 × 10 ). Similar results were obtained confirmed when analyzing the TCGA RPPA dataset (Figs. 1c and with a dataset consisting of patients treated with endocrine S1). AKT pathway was more often activated (elevated pAKT therapy only (multivariable HR, 0.82; 95% CI, 0.73–0.93; p = 0.002). expression) in luminal A cancers whereas mTOR pathway was Indeed, patients with pAKT pathway activation had better more often activated (elevated p-mTOR and pS6, an mTOR outcomes irrespective of their specific subtype (luminal A multi- downstream target), in luminal B subtypes (Figs. 1c and S2). Next, variable HR, 0.85; 95% CI, 0.75–0.96; p = 0.01; luminal B HR, 0.91; we sought to determine how the pAKT and p-mTOR signatures 95% CI, 0.83–0.99; p = 0.033). In contrast, patients with p-mTOR correlate with other signatures and RPPA markers of the pathway. pathway activation had significantly worse outcomes in all luminal As shown in Fig. S3, the pAKT signature negatively correlates with −4 patients (multivariable HR, 1.1; 95% CI, 1.1–1.2; p = 9.9 × 10 ) and downstream effectors of the pathway while the p-mTOR signature this remained true when tested in the dataset consisting of positively correlates with them. patients treated with endocrine therapy only (multivariable HR, Altogether, these results demonstrate that the pAKT and p- 1.2; 95% CI, 1.1–1.4; p = 0.004) (Figs. 4 and 5). mTOR pathways, assessed through these RPPA-based gene Next, we assessed the association between the PIK3CA and P53 expression signatures, have exclusive distribution according to mutation status and pAKT and p-mTOR pathway activation in an luminal molecular subtypes and are not necessarily linearly independent set, namely the TCGA BC patients with RNA- connected. sequence gene expression data that were not used to design the signatures. While the pAKT signature was associated with Association of the pAKT and p-mTOR pathway activation with PIK3CA mutations (p = 0.0001), the p-mTOR signature was not clinical outcome in patients with ER-positive early BC (p = 0.22) (Fig. 6a, b). The opposite was true for P53 mutations, To ascertain the impact of each pathway on outcomes in ER- which were positively correlated with p-mTOR pathway activation positive BC, we applied the pAKT and p-mTOR signatures on a (p = 0.04), and negatively correlated with pAKT (p = 0.0003) dataset composed of 38 publicly available microarray datasets. We (Fig. 6i, j). Analysis of the PIK3CA mutations by exon led to similar first assessed whether pAKT or p-mTOR pathway activation were results (Fig. 6), although mutations outside of exons 9 and associated with any particular luminal subtype. As expected, in the 20 seemed less associated with pAKT pathway activation. pooled set analysis pAKT pathway activation was significantly Finally, in an effort to identify whether these signatures could −10 associated will luminal A cancers (p <10 ) whereas p-mTOR predicts response to mTOR inhibitors, we evaluated another data pathway activation was associated with luminal B cancers (p < set of neo-adjuvant patients treated with Everolimus. Analysis of npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation 1234567890():,; A. Sonnenblick et al. Fig. 2 pAKT (right) and p-mTOR (left) gene signatures expression in publicly available microarray datasets according to the PAM50 breast cancer subtype. Kruskal–Wallis p-value is shown Fig. 3 High pAKT gene signature expression is associated with good prognosis in the luminal subtype. a–f We assessed the prognostic value of tertiles of pAKT gene signature expression in: a all luminal patients treated or not treated (n = 3073), b luminal A (n = 1491), c Luminal B (n = 1582), d all luminal treated with only hormonal therapy (n= 1180), e luminal A treated with only hormonal therapy (n = 491), and f luminal B treated with only hormonal therapy (n= 689). Significance (p-value) of differences in survival between patient groups defined by tertiles of pAKT signature expression is estimated by log-rank test. The analysis presented includes patients with lymph node-negative and lymph node-positive cancers Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 A. Sonnenblick et al. Fig. 4 High p-mTOR gene signature expression is associated with bad prognosis in the luminal subtype. a–f We assessed the prognostic value of tertiles of p-mTOR gene signature expression in: a all luminal patients treated or not treated (n = 3073), b luminal A (n = 1491), c Luminal B (n = 1582), d all luminal treated with only hormonal therapy (n = 1180), e luminal A treated with only hormonal therapy (n = 491), and f luminal B treated with only hormonal therapy (n = 689). Significance (p values) of differences in survival between patient groups defined by tertiles of p-mTOR signature expression is estimated by log-rank test. The analysis presented includes patients with lymph node-negative and lymph node-positive cancers the correlations between the effectiveness of this treatment and tumorigenic effects. Between 30% and 40% of BCs, especially the developed signatures suggests as expected that the pAKT ER-positive tumors, have mutations in PIK3CA. The vast majority signature is associated with less response to Everolimus (r = 0.45; of the PIK3CA mutations are missense mutations which are p = 0.031, Fig. S4). positioned in the helical domain (exon 9, mostly: E545K and Overall, our data suggests that pAKT and p-mTOR pathway E542K) and the kinase domain (exon 20, mostly H1047R) in activation as assessed through the respective signatures, despite hotspot clusters. These mutations have direct effect on AKT being major components of the same overarching pathway (PI3K), phosphorylation. The effect of PIK3CA mutations/pAKT on prog- have distinctly different impacts on disease biology and conse- nosis is mixed in early BC. We found that exons 9 and 20 quently on outcomes in early disease. mutations in PIK3CA were more associated with pAKT than mutations in other exons. We previously reported that PIK3CA mutations were associated DISCUSSION with improved outcome and low levels of signaling through the 4,17 The goal of the present study was to better understand the mTOR pathway in BC. Several possible hypotheses were raised distinct contribution to disease biology and clinical outcomes of regarding the reasons for this. Some data available on PP2A and signaling through the AKT and mTOR downstream pathways, PML, both known to have an inhibitory effect on both AKT and 18,19 which typically occur as part of PI3K pathway activation in luminal mTOR, have suggested that they may be upregulated in BCs. We found that pAKT and p-mTOR were differentially PIK3CA-activated tumors. Negative feedback regulation in PI3K- expressed according to luminal subtypes, implying different mediated cells through the insulin receptor substrate and degrees of pathway activation, and that, more importantly, the relatively weak pathway activation in PIK3CA-mutated cancers pathways were not linearly connected. Additionally, we found that have also been suggested as possible explanations for low levels pAKT pathway activation was positively associated with PIK3CA of signaling through mTOR in ER-positive BC. mutations whereas the opposite was observed with p-mTOR According to our findings only pAKT pathway activation was pathway activation. In contrast, pAKT pathway activation was found to be significantly different between the luminal subtypes associated with good clinical outcome despite its known (A and B) and PIK3CA wildtype versus mutant, whereas p-mTOR npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. N HR CI Akt All 3997 0.72 0.68 to 0.76 0 −4 Akt LumA 1953 0.83 0.75 to 0.92 5 × 10 −5 Akt LumB 2044 0.85 0.79 to 0.91 1 × 10 −8 Akt Hormonal therapy 1234 0.73 0.66 to 0.82 1.5 × 10 Akt LumA Hormonal therapy 595 0.8 0.65 to 0.99 0.037 Akt LumB Hormonal therapy 639 0.92 0.8 to 1 0.21 −13 Akt No hormonal therapy 1392 0.72 0.65 to 0.79 8.9 × 10 Akt LumA No hormonal therapy 678 0.82 0.7 to 0.95 0.0091 −4 Akt LumB No hormonal therapy 714 0.8 0.71 to 0.9 2.5 × 10 −9 mTOR All 3997 1.2 1.1 to 1.3 1.2 × 10 mTOR LumA 1953 1.1 1 to 1.3 0.018 mTOR LumB 2044 1 0.92 to 1.1 0.93 −5 mTOR Hormonal therapy 1234 1.3 1.1 to 1.4 9.2 × 10 mTOR LumA Hormonal therapy 595 1.2 0.95 to 1.5 0.14 mTOR LumB Hormonal therapy 639 1 0.88 to 1.2 0.83 −4 mTOR No hormonal therapy 1392 1.2 1.1 to 1.3 9.9 × 10 mTOR LumA No hormonal therapy 678 1.2 0.99 to 1.4 0.074 mTOR LumB No hormonal therapy 714 0.98 0.87 to 1.1 0.76 0.5 1 2 HR B p N HR CI −10 Akt All 3248 0.79 0.74 to 0.85 2.5 × 10 Akt LumA 1561 0.85 0.75 to 0.96 0.01 Akt LumB 1687 0.91 0.83 to 0.99 0.033 Akt Hormonal therapy 1158 0.82 0.73 to 0.93 0.0021 Akt LumA Hormonal therapy 546 0.86 0.69 to 1.1 0.21 Akt LumB Hormonal therapy 612 0.98 0.85 to 1.1 0.77 −5 Akt No hormonal therapy 1079 0.78 0.69 to 0.88 6.8 × 10 Akt LumA No hormonal therapy 513 0.81 0.66 to 1 0.052 Akt LumB No hormonal therapy 566 0.9 0.77 to 1 0.17 −4 mTOR All 3248 1.1 1.1 to 1.2 9.9 × 10 mTOR LumA 1561 1.1 0.96 to 1.2 0.18 mTOR LumB 1687 0.99 0.91 to 1.1 0.77 mTOR Hormonal therapy 1158 1.2 1.1 to 1.4 0.0041 mTOR LumA Hormonal therapy 546 1.2 0.94 to 1.5 0.13 mTOR LumB Hormonal therapy 612 1 0.89 to 1.2 0.66 mTOR No hormonal therapy 1079 1.1 0.97 to 1.2 0.15 mTOR LumA No hormonal therapy 513 1.2 0.96 to 1.5 0.1 mTOR LumB No hormonal therapy 566 0.89 0.77 to 1 0.11 0.5 1 2 HR Fig. 5 Forest plots showing the hazard ratios of the recurrence free survival of pAKT and p-mTOR gene signatures treated as a continuous variable using Cox univariate a and multivariable analysis b, in the pooled analysis. For multivariate analysis, we considered the following variables: age, tumor size, grade, and nodal status. Signatures with nominal significant effect (p < 0.05) are shown in blue pathway activation was not significant for both. The inconclusive luminal A/PIK3CA mutations and good prognosis, while p-mTOR/ and relative activation of p-mTOR by the mutant PIK3CA may be pS6 is not, suggesting that the presence of HER2 and basal also attributed to the different roles and activators of mTOR and subtype in the primary analysis of the TCGA masked these the fact that mTOR is at the cross section of multiple signaling observations. pathways. Several studies have clearly demonstrated that mTOR is In conclusion our data suggest that the AKT and mTOR a direct substrate for the AKT kinase and identified Serine2448 as pathways are not linearly connected in luminal BCs. pAKT pathway the AKT target site in mTOR. However, additional studies have activation is associated with PIK3CA mutations, luminal A and demonstrated that rapamycin, an inhibitor of mTOR function, good prognosis, while p-mTOR pathway activation is associated blocks serum-stimulated Serine2448 phosphorylation of mTOR in with luminal B, P53 mutations, and bad prognosis. These results an AKT-independent manner and identified S6 kinase as a major may have important clinical implications considering that in low p- effector of mTOR phosphorylation at Serine2448. Indeed, our mTOR BCs, treatment with mTOR inhibitors, such as everolimus, analysis of the TCGA data shows that the S6 kinase (downstream which is highly toxic, will possibly be of lower value since the of mTOR) is associated with luminal B and P53 mutations pathway is not activated. Additionally, pAKT pathway activation, suggesting that while mTOR itself is at the cross section of as measured through our gene signature, can add to presently conflicting pathways its downstream targets are not PIK3CA used outcome prediction tools in both luminal A and luminal B dependent. In addition, there are alternative kinases that can tumors. activate the mTOR pathway independently of AKT, such as RSK which leads to phosphorylation of TSC resulting in increased mTOR signaling and the PDK1–SGK1 axis that can sustain mTOR METHODS 23–25 activity upon AKT suppression. Computation of RPPAs-based signatures The primary TCGA report, which investigated all BC subtypes, We downloaded clinic-pathological, normalized gene expression and RPPA confirmed a high frequency of PIK3CA mutations in luminal BC. data from the publicly available TCGA repository using its online Multiple platforms, which examined the relationship between 26 bioinformatics tools (Fig. S1 flow chart). ER-positive early BCs were PIK3CA mutation and protein expression, have demonstrated that analyzed based on the RPPA proteomic levels. 265 samples with available pAKT and pS6 were not elevated in PIK3CA-mutated luminal gene expression and RPPA data were considered as luminal (166 Luminal A cancers; instead, they were highly expressed in basal-like and and 99 Luminal B according to PAM50 computed on the cBioPortal HER2 subtypes. Our dataset, which is restricted to luminal cancers, website ). To identify the genes that were differentially expressed demonstrated that pAKT pathway activation is associated with between the low and high expression groups, and to find the genes that Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 A. Sonnenblick et al. 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.00012 p = 0.22 −0.2 No PIK3CA mutation PIK3CA mutation No PIK3CA mutation PIK3CA mutation 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.013 p = 0.19 −0.2 No PIK3CA−ex9 mut PIK3CA−ex9 mut No PIK3CA−ex9 mut PIK3CA−ex9 mut 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.69 p = 0.0082 −0.2 No PIK3CA−ex20 mut PIK3CA−ex20 mut No PIK3CA−ex20 mut PIK3CA−ex20 mut 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.99 p = 0.75 −0.2 No PIK3CA−oth mut PIK3CA−other mut No PIK3CA−oth mut PIK3CA−other mut 1.2 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 p = 0.00034 p = 0.04 −0.2 No P53 mutation P53 mutation No P53 mutation P53 mutation Fig. 6 Expression of pAKT (a, c, e, g, i) and p-mTOR (b–d, f, h, j) gene signatures levels in PIK3CA (all, exon9, exon20 or others) and P53 mutated and wild type samples in an independent RNA sequencing set (n = 309) would optimize the predictive power of our signatures, we used a machine Code availability learning approach as previously described. After this process, we were The expression levels of the signatures in the gene expression datasets left with 69 signatures (Supplementary Data) presenting a relevant AUC for were computed as previously described. In brief, we evaluated using a proteomic status prediction. Among others, p-mTOR achieved an AUC of nested 10-fold cross validation the maximal Benjamini–Hochberg false −6 −11 0.71 (p ~10 ) and pAKT an AUC of 0.77 (p ~10 ) in both luminal A and discovery rate and the minimal gene fold change that would optimize the B cancers. ability of the differentially expressed genes to predict the high/low status npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. Gene-expression data and statistical analyses Table 1. The sources and locations for the 38 gene expression We analyzed 38 gene expression datasets totaling more than 7000 tumors datasets analyzed (detailed in Table 1). To ensure comparability of expression values across multiple data sets, a 0.95 quantile normalization was performed. Dataset No. of Permanent identifier References Differences in expression of pAKT and p-mTOR signatures according to patients subtype were examined using the Kruskal–Wallis test. Survival outcome 31,32 data are presented as recurrence free survival (RFS). Survival plots NKI 337 10.1038/415530a according to the pAKT and p-mTOR signatures tertiles were drawn using 33,34 UCSF 162 GSE123833 the Kaplan–Meier method. Association of the signatures (i.e. pathway STNO2 122 GSE4335 activation) with good or bad outcomes were computed using uni-variate or multi-variate Cox regression analyses and data were presented as forest NCI 99 10.1073/pnas.1732912100 plots. For multivariate analysis, we considered the following variables: age, UNC4 337 GSE18229 tumor size, grade, and nodal status. To assess the correlation between the CAL 118 E-TABM-158 PIK3CA mutation status and AKT and p-mTOR gene pathway activation, as ‡ 39,40 represented by the gene signature scores, we analyzed the TCGA cohort of MDA4 129 (65) GSE123832 RNA sequenced data that was not used for the computation of the KOO 88 GSE123831 signatures (309 samples), and for which both mutational and gene HLP 53 E-TABM-543 expression data were available. Each sample was considered as mutated or 43 not (so a sample with four mutations was considered just like a sample EXPO 353 GSE2109 with one mutation). All mutations were taken into account. PIK3CA 44,45 VDX 344 GSE2034/GSE5327 mutations were also analyzed by specific exons (exons 9, 20, and all others MSK 99 GSE2603 grouped together). ‡ 47 UPP 251 (190) GSE3494 Reporting summary STK 159 GSE1456 ‡ 36,49 Further information on experimental design is available in the Nature UNT 137 (92) GSE2990 Research Reporting Summary linked to this article. DUKE 171 GSE3143 TRANSBIG 198 GSE7390 DATA AVAILABILITY DUKE2 160 GSE6961 The sources and locations for the 38 gene expression datasets analyzed during the MAINZ 200 GSE11121 current study are available in Table 1 and the figshare repository https://doi.org/ LUND2 105 GSE5325 10.6084/m9.figshare.7461776. LUND 143 GSE5325 FNCLCC 150 GSE7017 57 ACKNOWLEDGEMENTS EMC2 204 GSE12276 We would like to thank Carolyn Straehle for her editorial assistance. A.S. is supported MUG 152 GSE10510 by a Clinical Research Career Development Award from the Israel Cancer Research NCCS 183 GSE5364 Fund grants (16-116-CRCDA) and from the Israeli Cancer Research Association (2017- 0140). A.S. was an ESMO translational research fellow. C.S. is supported by the Breast MCCC 75 GSE19177 Research Cancer Foundation (BCRF). EORTC10994 49 GSE1561 DFHCC 115 GSE19615 ‡ 63 AUTHOR CONTRIBUTIONS DFHCC2 84 (75) GSE18864 ‡ 64 A.S. conceived of the study and participated in its design and coordination, analyzed DFHCC3 40 (26) GSE3744 and interpreted the data and wrote the manuscript. S.B. analyzed and interpreted the DFHCC4 129 GSE5460 data, performed the statistical analysis, and wrote the manuscript. N.P. coordinated MAQC2 230 GSE20194 the analysis and drafted the manuscript. D.V. carried out the statistical analysis ‡ 67 analyzed and interpreted the data and drafted the manuscript. C.S. participated in TAM 345 (242) GSE6532/GSE9195 the study design and coordination, analyzed, interpreted the data, and wrote the MDA5 298 GSE17705 manuscript. All authors read and approved the final manuscript. VDX3 136 GSE12093 PNC 248 GSE20713 ADDITIONAL INFORMATION TCGA 517 https://tcga-data.nci.nih.gov/ Supplementary information accompanies the paper on the npj Breast Cancer docs/publications/brca_2012/ website (https://doi.org/10.1038/s41523-019-0102-1). METABRIC 1643 EGAS00000000083 Competing interests: The authors declare no competing interests. Duplicated patients were removed from few datasets for the estimation of concordance and prognostic value Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. of the RPPA in luminal A and B patients together and separately. While the parameters were selected in a 10-fold cross validation, the procedure itself was assessed using a nested cross validation. All analyses were performed REFERENCES using the genefu package of the R (v3.2)/bioconductor (v1.18) statistical 1. Hennessy, B. T., Smith, D. L., Ram, P. T., Lu, Y. & Mills, G. B. Exploiting the PI3K/AKT suite. pathway for cancer drug discovery. Nat. Rev. Drug Discov. 4, 988–1004 (2005). 2. Koboldt, D. C. et al. Comprehensive molecular portraits of human breast tumours. Nature 490,61–70 (2012). 3. Campbell, I. G. Mutation of the PIK3CA gene in ovarian and breast cancer. Cancer Network analysis and clustering Res. 64, 7678–7681 (2004). 4. Loi, S. et al. PIK3CA mutations associated with gene signature of low Network analysis and clustering was performed as previously described. mTORC1 signaling and better outcomes in estrogen receptor-positive breast The MCL graph clustering algorithm was applied. Visualization of the cancer. Proc. Natl Acad. Sci. USA 107, 10208–10213 (2010). network has been rendered using the yED software. Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7 A. Sonnenblick et al. 5. Dienstmann, R., Rodon, J., Serra, V. & Tabernero, J. Picking the point of inhibition: 34. Korkola, J. E. et al. Differentiation of lobular versus ductal breast carcinomas by a comparative review of PI3K/AKT/mTOR pathway inhibitors. Mol. Cancer Ther. 13, expression microarray analysis. Cancer Res. 63, 7167–7175 (2003). 1021–1031 (2014). 35. Sorlie, T. et al. Repeated observation of breast tumor subtypes in independent 6. Baselga, J. et al. Everolimus in postmenopausal hormone-receptor–positive gene expression data sets. Proc. Natl Acad. Sci. USA 100, 8418–8423 (2003). advanced breast cancer. N. Engl. J. Med. 366, 520–529 (2012). 36. Sotiriou, C. et al. Breast cancer classification and prognosis based on gene 7. Navé, B. T., Ouwens, M., Withers, D. J., Alessi, D. R. & Shepherd, P. R. Mammalian expression profiles from a population-based study. Proc. Natl Acad. Sci. USA 100, target of rapamycin is a direct target for protein kinase B: identification of a 10393–10398 (2003). convergence point for opposing effects of insulin and amino-acid deficiency on 37. Prat, A. et al. Phenotypic and molecular characterization of the claudin-low protein translation. Biochem. J. 344, 427–431 (1999). intrinsic subtype of breast cancer. Breast Cancer Res. 12, R68 (2010). 8. Guertin, D. A. & Sabatini, D. M. Defining the role of mTOR in cancer. Cancer Cell 38. Chin, K. et al. Genomic and transcriptional aberrations linked to breast cancer 12,9–22 (2007). pathophysiologies. Cancer Cell 10, 529–541 (2006). 9. Manning, B. D. & Cantley, L. C. AKT/PKB signaling: navigating downstream. Cell 39. Hess, K. R. et al. Pharmacogenomic predictor of sensitivity to preoperative che- 129, 1261–1274 (2007). motherapy with paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide 10. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based in breast cancer. J. Clin. Oncol. 24, 4236–4244 (2006). approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. 40. Liedtke, C. et al. Response to neoadjuvant therapy and long-term survival in USA 102, 15545–15550 (2005). patients with triple-negative breast cancer. J. Clin. Oncol. 26, 1275–1281 (2008). 11. Sotiriou, C. et al. Gene expression profiling in breast cancer: understanding the 41. Huang, E. et al. Gene expression predictors of breast cancer outcomes. Lancet molecular basis of histologic grade to improve prognosis. J. Natl. Cancer Inst. 98, Lond. Engl. 361, 1590–1596 (2003). 262–272 (2006). 42. Natrajan, R. et al. An integrative genomic and transcriptomic analysis reveals 12. Lim, E. et al. Transcriptome analyses of mouse and human mammary cell sub- molecular pathways and networks regulated by copy number aberrations in populations reveal multiple conserved genes and pathways. Breast Cancer Res. basal-like, HER2 and luminal cancers. Breast Cancer Res. Treat. 121, 575–589 12, R21 (2010). (2010). 13. Sabine, V. S. et al. Gene expression profiling of response to mTOR inhibitor 43. Bittner M. (2005) Expression Project for Oncology (expO). www.intgen.org/expo. everolimus in pre-operatively treated post-menopausal women with oestrogen Accessed 3 Nov 2010. receptor-positive breast cancer. Breast Cancer Res. Treat. 122, 419–428 (2010). 44. Wang, Y. et al. Gene-expression profiles to predict distant metastasis of lymph- 14. Curtis, C. et al. The genomic and transcriptomic architecture of 2,000 breast node-negative primary breast cancer. Lancet Lond. Engl. 365, 671–679 (2005). tumours reveals novel subgroups. Nature 486, 346–352 (2012). 45. Minn, A. J. et al. Lung metastasis genes couple breast tumor size and metastatic 15. Samuels, Y. et al. High frequency of mutations of the PIK3CA gene in human spread. Proc. Natl Acad. Sci. USA 104, 6740–6745 (2007). cancers. Science 304, 554 (2004). 46. Minn, A. J. et al. Genes that mediate breast cancer metastasis to lung. Nature 436, 16. Yang, S. X., Polley, E. & Lipkowitz, S. New insights on PI3K/AKT pathway altera- 518–524 (2005). tions and clinical outcomes in breast cancer. Cancer Treat. Rev. 45,87–96 (2016). 47. Miller, L. D. et al. An expression signature for p53 status in human breast cancer 17. Kalinsky, K. et al. PIK3CA mutation associates with improved outcome in breast predicts mutation status, transcriptional effects, and patient survival. Proc. Natl cancer. Clin. Cancer Res. 15, 5049–5059 (2009). Acad. Sci. USA 102, 13550–13555 (2005). 18. Trotman, L. C. et al. Identification of a tumour suppressor network opposing 48. Pawitan, Y. et al. Gene expression profiling spares early breast cancer patients nuclear Akt function. Nature 441, 523–527 (2006). from adjuvant therapy: derived and validated in two population-based cohorts. 19. Bernardi, R. et al. PML inhibits HIF-1alpha translation and neoangiogenesis Breast Cancer Res. 7, R953–R964 (2005). through repression of mTOR. Nature 442, 779–785 (2006). 49. Loi, S. et al. Definition of clinically distinct molecular subtypes in estrogen 20. Paz, K. et al. A molecular basis for insulin resistance. Elevated serine/threonine receptor-positive breast carcinomas through genomic grade. J. Clin. Oncol. 25, phosphorylation of IRS-1 and IRS-2 inhibits their binding to the juxtamembrane 1239–1246 (2007). region of the insulin receptor and impairs their ability to undergo insulin-induced 50. Bild, A. H. et al. Oncogenic pathway signatures in human cancers as a guide to tyrosine phosphorylation. J. Biol. Chem. 272, 29911–29918 (1997). targeted therapies. Nature 439, 353–357 (2006). 21. Sekulić, A. et al. A direct linkage between the phosphoinositide 3-kinase-AKT 51. Desmedt, C. et al. Strong time dependence of the 76-gene prognostic signature for node-negative breast cancer patients in the TRANSBIG multicenter inde- signaling pathway and the mammalian target of rapamycin in mitogen- stimulated and transformed cells. Cancer Res. 60, 3504 (2000). pendent validation series. Clin. Cancer Res. 13, 3207–3214 (2007). 22. Chiang, G. G. & Abraham, R. T. Phosphorylation of mammalian target of rapa- 52. Bonnefoi, H. et al. Validation of gene signatures that predict the response of mycin (mTOR) at Ser-2448 is mediated by p70S6 kinase. J. Biol. Chem. 280, breast cancer to neoadjuvant chemotherapy: a substudy of the EORTC 10994/BIG 25485–25490 (2005). 00-01 clinical trial. Lancet Oncol. 8, 1071–1078 (2007). 23. Roux, P. P., Ballif, B. A., Anjum, R., Gygi, S. P. & Blenis, J. Tumor-promoting phorbol 53. Schmidt, M. et al. The humoral immune system has a key prognostic impact in esters and activated Ras inactivate the tuberous sclerosis tumor suppressor node-negative breast cancer. Cancer Res. 68, 5405–5413 (2008). complex via p90 ribosomal S6 kinase. Proc. Natl Acad. Sci. USA 101, 13489–13494 54. Saal, L. H. et al. Poor prognosis in carcinoma is associated with a gene expression (2004). signature of aberrant PTEN tumor suppressor pathway activity. Proc. Natl Acad. 24. Castel, P. et al. PDK1-SGK1 signaling sustains AKT-independent mTORC1 activa- Sci. USA 104, 7564–7569 (2007). tion and confers resistance to PI3Kα inhibition. Cancer Cell 30, 229–242 (2016). 55. Niméus-Malmström, E. et al. Gene expression profiling in primary breast cancer 25. Ma, L., Chen, Z., Erdjument-Bromage, H., Tempst, P. & Pandolfi, P. P. Phosphor- distinguishes patients developing local recurrence after breast-conservation ylation and functional inactivation of TSC2 by Erk implications for tuberous surgery, with or without postoperative radiotherapy. Breast Cancer Res. 10, R34 sclerosis and cancer pathogenesis. Cell 121, 179–193 (2005). (2008). 26. Cancer Genome Atlas Network. Comprehensive molecular portraits of human 56. Campone, M. et al. Prediction of metastatic relapse in node-positive breast breast tumours. Nature 490,61–70 (2012). cancer: establishment of a clinicogenomic model after FEC100 adjuvant regimen. 27. Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles Breast Cancer Res. Treat. 109, 491–501 (2008). using the cBioPortal. Sci. Signal. 6, pl1 (2013). 57. Bos, P. D. et al. Genes that mediate breast cancer metastasis to the brain. Nature 28. Sonnenblick, A. et al. p-STAT3 in luminal breast cancer: integrated RNA-protein 459, 1005–1009 (2009). pooled analysis and results from the BIG 2-98 phase III trial. Int. J. Oncol. 52, 58. Calabrò, A. et al. Effects of infiltrating lymphocytes and estrogen receptor on 424–432 (2018). gene expression and prognosis in breast cancer. Breast Cancer Res. Treat. 116, 29. Enright, A. J., Van Dongen, S. & Ouzounis, C. A. An efficient algorithm for large- 69–77 (2009). scale detection of protein families. Nucleic Acids Res. 30, 1575–1584 (2002). 59. Yu, K. et al. A precisely regulated gene expression cassette potently modulates 30. Sylvain, S. B., Sonnenblick, A., & Venet, D. Gene expression data sources for in metastasis and survival in multiple solid cancers. PLoS Genet. 4, e1000129 (2008). silico approach to assessing activation of AKT/mTOR signalling pathway signal- 60. Waddell, N. et al. Subtypes of familial breast tumours revealed by expression and ling pathways ER-positive early BC. figshare. Fileset https://doi.org/10.6084/m9. copy number profiling. Breast Cancer Res. Treat. 123, 661–677 (2010). figshare.7461776 (2018). 61. Farmer, P. et al. Identification of molecular apocrine breast tumours by micro- 31. van de Vijver, M. J. et al. A gene-expression signature as a predictor of survival in array analysis. Oncogene 24, 4660–4671 (2005). breast cancer. N. Engl. J. Med. 347, 1999–2009 (2002). 62. Li, Q. et al. Minimising immunohistochemical false negative ER classification 32. van’t Veer, L. J. et al. Gene expression profiling predicts clinical outcome of breast using a complementary 23 gene expression signature of ER status. PloS ONE 5, cancer. Nature 415, 530–536 (2002). e15031 (2010). 33. Korkola, J. E. et al. Identification of a robust gene signature that predicts breast 63. Silver, D. P. et al. Efficacy of neoadjuvant Cisplatin in triple-negative breast cancer. cancer outcome in independent data sets. BMC Cancer 7, 61 (2007). J. Clin. Oncol. 28, 1145–1153 (2010). npj Breast Cancer (2019) 7 Published in partnership with the Breast Cancer Research Foundation A. Sonnenblick et al. 64. Richardson, A. L. et al. X chromosomal abnormalities in basal-like human breast Open Access This article is licensed under a Creative Commons cancer. Cancer Cell 9, 121–132 (2006). Attribution 4.0 International License, which permits use, sharing, 65. Lu, X. et al. Predicting features of breast cancer with gene expression patterns. adaptation, distribution and reproduction in any medium or format, as long as you give Breast Cancer Res. Treat. 108, 191–201 (2008). appropriate credit to the original author(s) and the source, provide a link to the Creative 66. Popovici, V. et al. Effect of training-sample size and classification difficulty on the Commons license, and indicate if changes were made. The images or other third party accuracy of genomic predictors. Breast Cancer Res. 12, R5 (2010). material in this article are included in the article’s Creative Commons license, unless 67. Loi, S. et al. Predicting prognosis using molecular profiling in estrogen receptor- indicated otherwise in a credit line to the material. If material is not included in the positive breast cancer treated with tamoxifen. BMC Genom. 9, 239 (2008). article’s Creative Commons license and your intended use is not permitted by statutory 68. Symmans, W. F. et al. Genomic index of sensitivity to endocrine therapy for breast regulation or exceeds the permitted use, you will need to obtain permission directly cancer. J. Clin. Oncol. 28, 4111–4119 (2010). from the copyright holder. To view a copy of this license, visit http://creativecommons. 69. Zhang, Y. et al. The 76-gene signature defines high-risk patients that benefit from org/licenses/by/4.0/. adjuvant tamoxifen therapy. Breast Cancer Res. Treat. 116, 303–309 (2009). 70. Dedeurwaerder, S. et al. DNA methylation profiling reveals a predominant © The Author(s) 2019 immune component in breast cancers. EMBO Mol. Med. 3, 726–741 (2011). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2019) 7

Journal

npj Breast CancerSpringer Journals

Published: Jan 31, 2019

There are no references for this article.