Recently, inhibitors of the phosphatidylinositol-3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) pathway have been introduced into the clinic to overcome endocrine resistance [1, 2]. However, companion diagnostics for these new targeted drugs are lacking. Many molecular alterations in this pathway, as well as in the mitogen-activated protein kinase (MAPK) pathway, leading to its constitutive activation, have been described. Canonic pathway drivers are mutations in the PIK3CA gene [3], loss of expression or genetic alteration in the tumor-suppressor gene PTEN [4], and overexpression of growth factor receptors like human epidermal growth factor receptor 2 (HER2) and insulin-like growth factor 1 receptor (IGF-1R) [5].

PIK3CA mutations occur in about 20% to 25% of invasive ductal breast cancers and in approximately 40% of invasive lobular breast cancers [6], with hotspots in exon 9 (helical domain) and exon 20 (kinase domain). These mutations have been shown to result in in vitro activation of the PI3K/AKT/mTOR pathway [3], leading to endocrine resistance [7]. Nevertheless, the prognostic and predictive value regarding endocrine resistance of these mutations in ERα-positive breast cancer remains unclear. An important limitation of many conflicting clinical studies [8–12] is the analysis of these mutations in consecutive series of endocrine-treated patients, which is unsuitable to discern prognosis from prediction [13]. Only one previous study [14] analyzed these mutations in the context of a clinical trial that randomized between adjuvant tamoxifen and control. In this study, PIK3CA mutations did not predict endocrine resistance, but were associated with a decreased risk for local recurrence. In neoadjuvant endocrine therapy trials, PIK3CA mutation status was not associated with treatment-induced Ki67 changes, a surrogate marker for recurrence-free survival [15], nor with pathologic response [16], whereas the kinase domain mutations were associated with improved overall survival. Several other studies have suggested a relatively favorable survival in patients with kinase domain-mutated breast cancers [8, 17], in comparison with patients without such mutated tumors.

Several other known molecular alterations in the PI3K and or the MAPK pathway have been studied for their validity to predict endocrine resistance. Loss of PTEN, a negative regulator of the PI3K/AKT/mTOR pathway, frequently occurs in breast cancer [18], but did not have clinical validity as a single marker in a previous study [14]. The same holds true for HER2 [19], although the clinical validity of IGF-1R has not been analyzed in the context of a randomized clinical trial.

The aim of our study was to investigate the prognostic and treatment-predictive value of different molecular alterations in the PI3K and/or MAPK pathways in postmenopausal breast cancer patients randomized between adjuvant tamoxifen and no systemic treatment. In addition, we studied the association between these molecular alterations and downstream-activated proteins in the PI3K and/or MAPK pathways.

Our results indicate that PIK3CA mutations are unlikely to have important clinical validity to predict adjuvant tamoxifen resistance in postmenopausal breast cancer patients. In addition, we have shown that the presence of a molecular alteration in the PI3K and/or MAPK pathway is not always associated with high expression of downstream-activated proteins.

The observed frequency of PIK3CA mutations in the current study was in line with those reported in the literature [8, 14, 17]. Similar to others [16, 31], we observed an association with low tumor grade and negative HER2 status. Although others did not observe a significant association between PIK3CA mutations and tumor grade [6, 8, 32], this discordance could be explained by the relatively small number of patients in these studies. In the studies from Buttitta [6] and Barbareschi [8], patients with lobular breast cancer had more often PIK3CA exon 9 mutated tumors compared with nonlobular breast cancer. In our study, we did not observe such enrichment for PIK3CA exon 9 mutations in lobular breast cancer. An important difference between these studies and our study population is that we selectively analyzed ERα-positive postmenopausal breast cancer patients, that are predominantly of low tumor grade. In the other studies [6, 8], patients were younger, and the group of patients with nonlobular breast cancer included hormone receptor-negative patients, who are more often high tumor grade, and are less often PIK3CA-mutated [31]. This might explain why we observed a higher frequency of PIK3CA exon 9 mutations in nonlobular breast cancer compared with the studies from Buttita [6] and Barbareschi [8]. Similar to the results of PIK3CA mutation analysis in almost 2,000 patients who participated in the TEAM trial (treated with adjuvant tamoxifen and/or exemestane) [33], we observed a positive association between PIK3CA kinase domain mutations and PgR status. Altogether, our data indicate that in ERα-positive postmenopausal breast cancer patients, PIK3CA mutations are not enriched in lobular breast cancer, but are associated with favorable prognostic factors like low grade and positive PgR status.

In our study, we did not observe an association between PIK3CA mutations and activation of downstream proteins like mTOR and p70S6K. Although in vitro data have shown that PIK3CA mutations result in activation of the PI3K pathway [3], several studies have now shown that this is not necessarily the case in the clinical setting [14, 34]. Perez et al.[14] did not find an association between PIK3CA mutations and p-AKT levels, whereas Loi et al.[34] observed relatively moderate activation of the PI3K pathway in tumors with a PIK3CA exon 20 mutation-associated gene signature. In addition, in a large series of primary breast cancers analyzed with exome sequencing and reverse-phase protein arrays, PI3K pathway activation was not elevated in PIK3CA-mutated luminal A cancers [35]. One of the explanations for this could be an AKT-independent downstream signaling in these PIK3CA mutated tumors [36]. Alternatively, relatively moderate pathway activation could be the result of a feedback mechanism leading to downregulation secondary to pathway activation. Previously it was shown that a negative-feedback loop between mTOR/p70S6K and the IRS protein results in a reduction of the IRS protein in response to activation of mTOR/p70S6K, with subsequent inhibition of the PI3K pathway [37].

In contrast, IGF-1R protein expression was significantly associated with p-p70S6K. Elevated IGF-1R signaling has been shown to result in activation of the PI3K and MAPK pathways in vitro[38]. Surprisingly, in tumors that scored negative for PTEN, we observed relatively low expression of downstream-activated proteins in the PI3K pathway. Although the robustness of PTEN antibodies has been a matter of debate, the reliability of the PTEN antibody we used was previously shown [39]. In addition, we showed a significant association between PTEN immunoscoring and mRNA expression. Similar to our results, Perez et al. [14] observed relatively low expression of AKT in patients whose tumor was negative for PTEN. The underlying mechanism for this unexpected observation remains unclear. An explanation could be that activation of the PI3K pathway in tumors that lack PTEN is relatively low compared with tumors that exhibit PI3K pathway activation from other causes.

In patients randomized to the control arm, we did not observe an association between PIK3CA mutations, or any of the other tested molecular aberrations, and breast cancer prognosis, when corrected for known prognostic factors, such as the PgR status and histologic tumor grade. The relatively low number of HER2-positive breast cancer patients in this series may explain the absence of a significant association between HER2 overexpression and breast cancer prognosis. It is well known that the incidence of HER2 overexpression in the Netherlands is lower than that observed in other countries [40]. With respect to the association of IGF-1R with breast cancer outcome, discordant results have been published [41–45]. These conflicting results may be explained by heterogeneous patient populations (like difference in histologic subtype and treatment) as well as differences in antibodies used.

The association between PIK3CA mutation and breast cancer prognosis has been controversial [31]. Previously, a PIK3CA exon 20 gene signature was associated with favorable outcome in both tamoxifen-treated patients and in patients who did not receive adjuvant systemic treatment [34]. PIK3CA mutation status, as defined with sequencing, did not have prognostic value in this study. Several other studies have suggested a favorable prognosis in patients harboring PIK3CA mutations [8, 14]. A preliminary analysis of PIK3CA mutations in patients participating in the TEAM trial did not find a significant association with outcome [33]. In a recent meta-analysis showing a favorable clinical outcome in ERα-positive postmenopausal breast cancer patients with PIK3CA kinase domain mutations, a potential favorable response to endocrine treatment was suggested as one of the explanations [46]. Our results indicate that this is not likely to be true. A potential bias of these cohort studies is that the prognostic value of PIK3CA mutations is analyzed in breast cancer patients who all received adjuvant endocrine therapy. The prognosis of patients with PIK3CA mutations who have been treated with endocrine therapy is determined by both tumor biology and treatment effect, where the latter can range from substantial to no treatment effect. Hence its tumor biologic prognostic value cannot be deduced from such a study design.

The observed absence of a significant interaction between PIK3CA hotspot mutations and tamoxifen-treatment benefit in our study does not rule out a potential reduced effect of tamoxifen in patients carrying these hotspot mutations. Moreover, we cannot exclude that the benefit of tamoxifen might be reduced in patients whose tumors express a rare mutation in the PIK3CA gene or other genes in the PI3K/AKT/mTOR pathway (like AKT1 mutations). Considering their relatively low prevalence [35], we did not determine these mutations, because the power of this study to demonstrate an interaction between these mutations, analyzed as single markers, and tamoxifen treatment is low. For similar reasons, we cannot exclude a potential reduced benefit from tamoxifen in patients whose tumors exhibit one of the canonic pathway drivers, like loss of PTEN and overexpression of HER2 and/or IGF-1R. Nevertheless, when a composed variable was tested, indicating the presence of any of these molecular alterations, again no significant interaction was found. We do not expect that the addition of other (rare) mutations in these analyses would substantially change these results.

In contrast, analysis of p-p70S6K identified a similar amount of patients and showed a highly significant test for interaction [29]. In PIK3CA-mutated tumors that do express p-p70S6K, a reduced effect is likely, but cannot be demonstrated because of lack of power. A meta-analysis of these markers on tumor material available from randomized clinical trials of adjuvant tamoxifen versus nil (for example, NSABP B-14 trial [47], NATO and CRC adjuvant breast trials [48, 49], and the Stockholm trial [50]) might generate more definitive answers regarding these biomarkers.

Considering our observations in ERα-positive early breast cancer patients, it is not likely that PIK3CA hotspot mutation status by itself would be a suitable companion diagnostic predicting benefit from the putative use of PI3K/AKT/mTOR inhibitors in the adjuvant setting. Phase I data have shown a higher response rate for patients with gynecologic malignancies carrying a PIK3CA mutation treated with PI3K/AKT/mTOR inhibitors compared with patients with wild-type tumors [51]. However, whether this is also true for ERα-positive breast cancer must be defined in prospective randomized clinical trials, both in the metastatic and adjuvant settings, with these novel targeted agents. Our data indicate that the selection of patients for these trials should not be restricted to PIK3CA hotspot mutation carriers only.

Novel targeted agents inhibiting the PI3K/AKT/mTOR pathway have a promising role in the treatment of patients with hormone receptor-positive breast cancer resistant to anti-estrogens as single agent. Because PIK3CA hotspot mutations frequently occur and are known to activate the PI3K/AKT/mTOR pathway, these mutations are generally considered a potential predictive biomarker. Our observations indicate that PIK3CA hotspot mutations have limited potential to predict intrinsic tamoxifen resistance in the adjuvant treatment of ERα-positive, postmenopausal breast cancer patients. Furthermore, no clear association between these mutations and activation of downstream proteins in the PI3K/AKT/mTOR pathway has been found in these patients. For identification of companion diagnostics, the focus should switch to the analysis of activated proteins downstream in the PI3K/AKT/mTOR pathway, which are associated with adjuvant tamoxifen resistance [29].

From paraffin-embedded tissue blocks, 10-μm-thick sections were cut and attached to microscope slides. In total, 10 slides per tumor were used for DNA isolation. Slides were deparaffinized in xylene, rehydrated, and stained with hematoxylin. The slides were incubated with sodium thiocyanate overnight. Exact tumor location was circled by a pathologist on an HE-stained slide, which was used as a template. After adding a drop of tissue lysis buffer, tumor tissue was scraped from the slides, added to a 1.5-μl micro centrifuge tube containing a 200-μl mix of tissue lysis buffer/proteinase K. This tube was incubated in a thermomixer at 55°C for 48 hours. An additional 27 μl proteinase K (2 mg/μl) was added after 24 and 36 hours. After 48 hours, the tube was incubated at 80°C for 10 minutes to inactivate proteinase K. After centrifuging, the supernatant was pipetted into a new tube. DNA was purified by using a QIAquick PCR purification kit.