Impact of somatic PI3K pathway and ERBB family mutations on pathological complete response (pCR) in HER2-positive breast cancer patients who received neoadjuvant HER2-targeted therapies

Approximately 20% of primary breast cancers have an alteration, usually amplification, of the human epidermal growth factor receptor 2 (HER2) gene. HER2-amplified cancers have an inferior prognosis, with more frequent and more rapid metastatic relapse [1]. The addition of trastuzumab to conventional chemotherapy has significantly improved the outcomes for patients with both early-stage and metastatic HER2-altered breast cancer, with significantly improved survival in both settings [2]. Although a small minority of patients with HER2-positive metastatic disease achieve durable complete remissions, approximately 80% to 90% develop progressive cancer. Addition of trastuzumab to pre-operative chemotherapy results in 30% to 50% of patients achieving a pathological complete response (pCR) [2]. Despite this, approximately 20% of patients with HER2-positive early-stage disease and 90% of patients with metastatic disease will die from disease which is resistant to trastuzumab. Newer targeted anti-HER2 agents have been studied in an attempt to overcome trastuzumab resistance [3–5].

One such agent is lapatinib, a small-molecule inhibitor of epidermal growth factor receptor (EGFR) and receptor tyrosine-protein kinase erbB-2 (ERBB2). In a large randomised trial, the addition of lapatinib to capecitabine in patients whose cancer had progressed following treatment with anthracycline, taxanes and trastuzumab was associated with statistically prolonged progression-free survival compared with capecitabine alone [6]. Evidence suggests that combined targeting of both the extracellular domain of HER2 with trastuzumab and the kinase domain with lapatinib may further improve response. We have shown that combined treatment with trastuzumab and lapatinib improves response to chemotherapy in SKBR3 breast cancer cells and decreases tumour growth in BT-474 xenografts [7], with the combination also showing improved response in the neoadjuvant treatment of HER2-positive breast cancer in some clinical trials [8–11]. However, the benefit of combined HER2-targeted therapy has not been shown consistently. The National Surgical Adjuvant Breast and Bowel Project protocol B-41 demonstrated no significant difference in pCR rates between patients receiving the combination of trastuzumab and lapatinib and those receiving either single agent [12], while the EORTC 10054 study demonstrated a numerically higher but non-significant benefit of double anti-HER2 blockade with trastuzumab and lapatinib [13]. Thus, to determine for ourselves whether the addition of lapatinib to, or the substitution of lapatinib for, trastuzumab would improve pCR, we initiated protocol ICORG 10-05, a prospective, randomised trial with stage Ic/II/III HER2-positive breast cancer patients (NCT01485926) [14].

Activating somatic mutations in the phosphatidylinositol 3-kinase (PI3K)/AKT pathway are present in a range of tumour types [15, 16]. Mutations in PIK3CA occur in approximately one-third of breast cancers [17], and these mutations have been implicated in the development of trastuzumab resistance [18, 19]. Ligand binding to ERBB family members activates intracellular signalling pathways such as the PI3K/AKT pathway [20]. Trastuzumab and lapatinib block this signalling, either by binding ERBB2 at the cell surface or by directly inhibiting the kinase activity of both EGFR and ERBB2 [20]. Possible resistance mechanisms include constitutive activation of the PI3K/AKT pathway through somatic mutations in the PI3K pathway or altered intracellular signalling involving loss of phosphatase and tensin homolog (PTEN) [18].

Mutations in EGFR, ERBB2, Erb-B2 receptor tyrosine kinase 3 (ERBB3) and Erb-B2 receptor tyrosine kinase 4 (ERBB4) (referred to hereinafter as ERBB family mutations) either occur alone or co-occur with PIK3CA mutations in 19% of HER2-positive breast cancers (n = 58) [21, 22]. Because ERBB-mediated effects are dependent on PI3K/AKT signalling and ERBB family mutations can activate the PI3K/AKT pathway, it is likely that they have similar canonical signalling effects to PI3K pathway mutations and PTEN loss. Therefore, the primary aim of our study was to investigate the association of pCR with ERBB family and PIK3CA mutations and PTEN loss (defined as PI3K pathway activation) in primary HER2-positive breast cancer treated with one or two HER2-targeting agents.

Biomarker analysis is necessary to identify those patients most likely to respond to neoadjuvant chemotherapy. In HER2-positive breast cancer, HER2 status is still the only recognised predictive marker to select patients for anti-HER2-targeted therapy; however, in recent years, a number of other biomarkers have emerged as potential predictors for response. PI3K pathway activation is the most common signal transduction pathway alteration in breast cancer [17, 25]. It mostly results from somatic PIK3CA mutations or PTEN loss [26], but it can also be a result of mutations in ERBB family genes [20, 27]. Although a number of studies have reported on PI3K pathway activation in HER2-positive breast cancer, they have been focused mainly on its impact on response to anti-HER2 therapy in metastatic disease or in the adjuvant setting [18, 28–30]. In the present study, we sought to evaluate associations between activation of the PI3K pathway and the efficacy of trastuzumab and lapatinib therapy in the neoadjuvant setting in early HER2-positive breast cancer.

Whereas pre-clinical studies have shown that PI3K pathway activation contributes to resistance to anti-HER2-targeted therapies [18, 31], clinical studies have failed to give a clear answer. In the Neoadjuvant Lapatinib and/or Trastuzumab Treatment Optimization Trial (NeoALTTO), tumours with PIK3CA mutations had lower pCR rates after treatment with neoadjuvant paclitaxel plus HER2-targeted therapy, with pCR rates decreasing from 34.5% in PIK3CA WT tumours to 21.3% in mutated tumours [32]. Results from two sequential neoadjuvant studies (NCT00133796 and NCT00206427) demonstrated that patients with tumours with low PTEN expression or PI3K pathway mutations were less likely to achieve a pCR after neoadjuvant trastuzumab/docetaxel than patients with tumours with no PI3K pathway mutations or high PTEN expression (18.2% vs. 66.7%; p = 0.015) [33]. In the Chemotherapy, Herceptin and Lapatinib in Operable Breast cancer (CHER-LOB) study, similar pCR rates after neoadjuvant paclitaxel for 12 weeks followed by fluorouracil, epirubicin and cyclophosphamide for four courses every 3 weeks plus HER2-targeted therapy were seen in patients with PIK3CA-WT and PIK3CA-mutated breast cancers (33.3% vs. 22.7%; p = 0.323) [34]. A recent meta-analysis of pooled data from five neoadjuvant clinical trials demonstrated lower pCR rates in the PIK3CA mutant cohort than in the WT cohort (16.2% vs. 29.6%; p < 0.001) [35].

In the present study, somatic PIK3CA mutations were found in 24.3% of tumours, and ERBB family mutations were found in 10.8% of tumours, with a similar distribution in ER-positive or ER-negative tumours, which confirms results of previous studies [17, 21, 22]. Reduced PTEN expression was found in 31.1% of tumours. Combining these aberrations which result in activation of the PI3K pathway, we find 55.6% of tumours with PI3K pathway activation (Fig. 2c). However, there was no correlation between activation of the PI3K pathway and response to anti-HER2 neoadjuvant therapy in all study patients as measured by pCR (Fig. 3). The rate of pCR was similar among tumours with activation of the PI3K pathway (defined as a PIK3CA mutation and/or low PTEN expression and/or an ERBB family mutation) and among tumours without a PIK3CA or ERBB family mutation or low PTEN expression (44% vs. 50%; p = 0.769) (Fig. 3).

Pre-clinical studies suggest that PTEN loss may be a potential mechanism of resistance to HER2-targeted therapies [36, 37]. In the neoadjuvant setting, results are more variable. Dave et al. [33] showed that patients with low PTEN expression who received trastuzumab had lower pCR rates than those with high PTEN expression (15.4% vs. 44.4%). However, in those patients who received lapatinib, low expression of PTEN was significantly associated with higher pCR rates than high expression (92.3% vs. 41.2%; p = 0.007). In the GeparQuattro study, low PTEN expression was also associated with lower response to trastuzumab-based chemotherapy compared with high PTEN expression (27.6% vs. 57.1%; p = 0.010) [38]. In our present study, we did not find any association between PTEN expression and pCR, similar to the results of the NeoALTTO study [39]. The differences between the studies may be attributable to a lack of standardisation of PTEN detection and scoring methods, as well as to the lack of a standardised definition for low and high PTEN expression.

In the TCH arm, as in all study patients, the pCR rates were not significantly affected by PIK3CA or ERBB family mutations or by PTEN or PI3K activation status. In contrast, in the TCHL arm, patients with PIK3CA and/or ERBB family mutated tumours were more likely to achieve a pCR than patients with PIK3CA and ERBB WT tumours (77.8% vs. 35%; p = 0.05) (Fig. 3). Our results are in contrast to NeoALTTO, where patients treated with a combination of weekly paclitaxel, trastuzumab and lapatinib who had PIK3CA-WT tumours had a pCR rate of 53.1%, which decreased to 28.6% in patients with tumours harbouring PIK3CA mutations [32]. The difference in the impact of PIK3CA mutations in the two studies may be attributable to the low sample size in the TCHL cohort. Our study also included analysis of additional biomarkers of PI3K/AKT activation, including ERBB family mutations, which suggests that a trend may exist within the data towards a benefit of TCHL treatment in patients with PIK3CA and/or ERBB family mutations, and this warrants further study in a larger cohort. Furthermore, it is worth noting that the neoadjuvant regimens used are different in the relevant study arms. TCHL for six cycles was used in our study versus HL (trastuzumab and lapatinib) followed by weekly paclitaxel for 12 weeks with HL in NeoALLTO; therefore, sensitivity to the chemotherapy part of the regimen itself cannot be ruled out.

The finding that activating mutations in a signalling pathway can increase sensitivity to targeted therapies is not unusual. The tyrosine kinase inhibitors erlotinib and gefitinib show significant clinical responses in lung adenocarcinoma patients harbouring EGFR activating mutations [40], whereas dacomitinib, a pan-HER tyrosine kinase inhibitor has been shown to be beneficial in HER2-mutated lung tumours [41]. ERBB4 has been shown to be highly mutated in melanoma, where ERBB4 mutations increase ERBB4 kinase activity despite similar expression of the HER4 protein and sensitise cells to lapatinib [42]. In some, though not all, clinical studies, the combination of trastuzumab and lapatinib has been shown to be more effective than each drug given alone, likely because acquired drug resistance that results from the activation of alternate pro-survival pathways can be overcome by combination treatments [43]. In vitro, while PIK3CA mutated HER2-positive BT474 cells stably expressing the ERBB2-T798M mutation were resistant to trastuzumab, the addition of lapatinib or cetuximab restored sensitivity [44], likely through blocking heterodimer formation [44].

A limitation of our study and other studies is the absence of a group not treated with anti-HER2 therapy. Given that all patients received chemotherapy in association with anti-HER2 treatment, and also that there was considerable heterogeneity in the chemotherapy treatment regimens used between the studies, we cannot exclude the possibility that some of our findings are contributed to by sensitivity to the chemotherapy part of the regimen itself. We also acknowledge that our study is small, and it may be useful to validate these results in a larger cohort, where there will be a greater power to demonstrate a positive interaction between mutation status and pCR.

To the best of our knowledge this is the first study looking at the effects of PI3K pathway activation through a combination of PIK3CA gene mutations, ERBB family gene mutations and low PTEN expression on pCR in the neoadjuvant treatment of HER2-positive breast cancer. Our study confirms the high prevalence of mutations that result in activation of the PI3K pathway in HER2-positive breast cancer and their potential utility in predicting patients most likely to respond to combination anti-HER2 therapy. Because this is also the first report on the impact of PIK3CA mutations on responsiveness to neoadjuvant taxane plus carboplatin-based HER2-targeted therapy regimens, with other studies such as NeoALLTO involving neoadjuvant taxane- or taxane/anthracycline-based HER2-targeted therapy regimens, our findings need validation in other studies.