Prognostic and predictive value of p-Akt, EGFR, and p-mTOR in early breast cancer
- First Online:
- Cite this article as:
- Lazaridis, G., Lambaki, S., Karayannopoulou, G. et al. Strahlenther Onkol (2014) 190: 636. doi:10.1007/s00066-014-0620-6
- 301 Views
Background and purpose
There are scarce data available on the prognostic/predictive value of p-Akt and p-mTOR protein expression in patients with high-risk early breast cancer.
Patients and methods
Formalin-fixed paraffin-embedded (FFPE) tumor tissue samples from 997 patients participating in two adjuvant phase III trials were assessed for EGFR, PTEN, p-Akt, p-mTOR protein expression, and PIK3CA mutational status. These markers were evaluated for associations with each other and with selected patient and tumor characteristics, immunohistochemical subtypes, disease-free survival (DFS), and overall survival (OS).
p-mTOR protein expression was negatively associated with EGFR and positively associated with PTEN, with p-Akt473, and with the presence of PIK3CA mutations. EGFR expression was positively associated with p-Akt473, p-Akt308, and PIK3CA wild-type tumors. Finally, p-Akt308 was positively associated with p-Akt473 expression. In univariate analysis, EGFR (p = 0.016) and the coexpression of EGFR and p-mTOR (p = 0.015) were associated with poor OS. Among patients with p-Akt308-negative or low-expressing tumors, those treated with hormonal therapy were associated with decreased risk for both relapse and death (p = 0.013 and p < 0.001, respectively). In the subgroup of patients with locoregional relapse, positive EGFR and mTOR protein expression was found to be associated with increased (p = 0.034) and decreased (p < 0.001) risk for earlier relapse, respectively. In multivariate analysis, low levels of p-Akt308 and the coexpression of EGFR and p-mTOR retained their prognostic value.
Low protein expression of p-Akt308 was associated with improved DFS and OS among patients treated with hormonal therapy following adjuvant chemotherapy. Coexpression of EGFR and p-mTOR was associated with worse OS.
Keywordsp-mTORp-AktBreast cancerPrognostic valuePredictive value
Prognostischer und prädiktiver Werte von p-Akt, EGFR und p-mTOR beim Mammakarzinom im Frühstadium
Geringe Daten existieren über den prognostischen/prädiktiven Wert der p-Akt- und p-mTOR-Proteinexpression bei Patienten mit “High-risk”-Mammakarzinom im Frühstadium.
Patienten und Methoden
Formalinfixierte und in Paraffin eingebettete (FFPE) Tumorgewebeproben von 997 Patienten, welche im Rahmen von 2 adjuvanten Phase-III-Studien zytostatisch behandelt wurden, wurden auf EGFR, PTEN, p-Akt, pmTOR und PIK3CA-Mutationsstatus untersucht. Diese Marker wurden in Assoziation mit ausgewählten Patienten- und Tumorcharakteristika, immunhistochemischen Mammakarzinomsubtypen, dem krankheitsfreien Überleben (DFS) sowie dem Gesamtüberleben (OS) evaluiert.
Die Expression von p-mTOR war negativ mit der Expression von EGFR, jedoch signifikant positiv mit PTEN und p-Akt473 sowie dem Nachweis von PIK3CA-Mutationen assoziiert. Die EGFR-Expression war signifikant positiv mit der Proteinexpression von p-Akt473 und p-Akt308 sowie PIK3CA-Wildtyp-Tumoren assoziiert. Die p-Akt308-Expression war ebenfalls signifikant positiv mit der Expression von p-Akt473 assoziiert. In der univariaten Analyse war sowohl die EGFR (p = 0,016), als auch die Koexpression von EGFR und p-mTOR (p = 0,015) mit einem schlechteren OS assoziiert. Unter den Patienten mit p-Akt308-negativen oder gering exprimierenden Tumoren konnte bei denjenigen, die mit einer Hormontherapie behandelt wurden, ein signifikant vermindertes Risiko für ein Rezidiv als auch für den Tod (jeweils p = 0,013 und p < 0,001) nachgewiesen werden. In der Subgruppenanalyse von Patienten mit lokoregionärem Rezidiv konnte eine Assoziation von positivem EGFR-Status sowie mTOR-Expression mit einem jeweils erhöhten (p = 0,034) bzw. verminderten (p < 0,001) Risiko für das Auftreten von Frührezidiven nachgewiesen werden. In der multivariaten Analyse behielten die niedrigen p-Akt308-Werte und die Koexpression von EGFR und p-mTOR ihren prognostischen Wert.
Eine niedrige Expression von p-Akt308 bei Patienten, die eine Hormontherapie nach einer adjuvanten Chemotherapie erhielten, war mit einem verbesserten DFS und OS assoziiert. Die Koexpression von EGFR und p-mTOR war mit einem schlechteren OS assoziiert.
Schüsselwörterp-mTORp-AktMammakarzinomPrognostischer WertPrädiktiver Wert
In the era of personalized medicine and targeted agents, there is an urgent need for the identification of prognostic and predictive biomarkers that can help tailor treatment . In breast cancer, despite extensive research, only two validated predictive biomarkers have been identified thus far: ER/PgR  and HER2 . Since then, little progress had been made in the personalized management of breast cancer patients, until the recent recognition of the central role of the PI3K/Akt/mTOR pathway in the regulation, growth, apoptosis, and motility of breast cancer cells [4, 5]. PI3K, Akt. and mTOR inhibitors have been developed and are being tested. The importance of targeting this pathway was apparent after the positive findings of the BOLERO-2 study, in which it was shown that metastatic breast cancer patients with previous exposure to endocrine therapy derive substantial benefit from the addition of an mTOR inhibitor to endocrine therapy .
Although mTOR inhibitors are already in use in daily clinical practice, little is known about the prognostic and predictive value of the expression of p-mTOR (phosphorylated mTOR at Ser-2448), the active form of mTOR. Aberrant kinase (EGFR) signaling, overexpression of Akt, PIK3CA mutations, and loss of PTEN collectively result in dysregulation of the PI3K/Akt/mTOR pathway in breast cancer. There are data supporting the negative prognostic value of EGFR expression in breast cancer patients , while PTEN loss and PIK3CA mutations are thought to confer resistance to anti-HER2 agents in metastatic breast cancer [8, 9] and activation of Akt seems to play an important role in endocrine resistance in metastatic breast cancer . In our study we performed a retrospective analysis of two randomized adjuvant studies (HE10/97 and HE10/00) to evaluate the outcome of patients treated with anthracycline-based adjuvant chemotherapy according to EGFR, PTEN, p-Akt, and p-mTOR protein expression and PIK3CA mutational status. Associations between these effectors, as well as with selected patient and tumor characteristics, were also investigated.
Patients and methods
The HE10/97 trial  was a randomized dose-dense adjuvant chemotherapy phase III trial comparing four cycles of epirubicin (E) followed by four cycles of intensified CMF (E-CMF) with three cycles of E, followed by three cycles of paclitaxel (T) and then by three cycles of intensified CMF (E-T-CMF). The HE10/00 trial [12, 13] was a randomized adjuvant chemotherapy phase III trial comparing dose-dense E-T-CMF (exactly as in the HE10/97 trial) with four cycles of epirubicin/paclitaxel (ET) combination (given on the same day) every 3 weeks followed by three cycles of intensified CMF every 2 weeks (ET-CMF). The baseline characteristics and clinical outcomes of both trials have been published [11–13]. HER2-positive patients received trastuzumab upon relapse, as previously described .
The present translational research protocol was approved by the Bioethics Committee of the Aristotle University of Thessaloniki School of Medicine and patients gave written informed consent for the use of their biological material for future research purposes.
Radiation therapy (RT) was required for all patients who underwent partial mastectomy or those with tumor size ≥ 5 cm and/or > 4 positive lymph nodes, irrespective of the type of surgery (conservative or radical). RT was initiated 3–4 weeks following the completion of chemotherapy.
The aim of breast irradiation was to treat the skin, muscles, and lymphatics in the axilla and the supraclavicular fossa (if there were > 4 positive lymph nodes), the entire surgical scar, and any remaining breast tissue. Dose to the underlying lung tissue, cervical spine, and brachial plexus was to be kept to a minimum.
The patient lay supine, with the ipsilateral arm abducted to 90° and the head away from the affected side. RT was given with megavoltage machines. A treatment planning was prepared for each individual patient, using either a simulator (in which case an outline was taken with plaster of Paris) or computed tomography (CT) slices taken in the treatment position.
As standard, a four-field technique excluding the internal mammary nodes was applied (two tangential fields, medial and lateral, one anterior and one posterior field covering the axilla and the supraclavicular fossa). The upper border of the tangential fields lay at the level of the manubrium sterni and the lower border 1–2 cm below the margin of the breast tissue. The lateral border lay at least in the midaxillary line and the medial approximately in the midline. The axilla and supraclavicular fossa were treated using an anterior field whose inferior border lay 0.5 cm above and parallel to the superior border of the tangential fields. The medial border lay 1 cm lateral to the anterior midline. Superiorly, only the supraclavicular nodes were covered. The posterior field covered either only the axilla or both the supraclavicular fossa and the axilla.
As an alternative to the four-field technique, a three-field technique could be applied (two tangential and one anterior-oblique high-axillary supraclavicular field, angled at approximately 10° medially). Scar extensions, not covered by the standard fields, were treated by electron beams.
A total dose of 50–55 Gy was given to the chest wall, with another 50–55 Gy given to the regional lymph nodes (at a depth of 1 cm for the supraclavicular nodes). The dose per fraction was in the range of 1.8–2.0 Gy. Five fractions were given per week. The underlying chest wall tissues did not receive more than 70 % of the prescribed dose at a depth of 2 cm. A dose exceeding 105 % (hot spot) was not accepted, with wedges or compensators used to reduce hot spots. Radiation of the axilla was not recommended.
An additional dose of 10 Gy was delivered to the tumor bed. The technique depended on the type of operation, i.e., radical mastectomy: 10 MeV electrons; lumpectomy: external beam radiotherapy (photons or electrons) or brachytherapy.
Tumor tissue samples, processing, methods, and biomarkers
For all methods, formalin-fixed paraffin-embedded (FFPE) tumor tissue samples were used. The samples were collected retrospectively in the first trial (HE10/97) and prospectively in the second one (HE10/00). For in situ methods (immunohistochemistry and FISH), tissue microarray technology was used, whereas for PIK3CA mutations, single-nucleotide polymorphism (SNP) genotyping was employed. The REMARK diagram  for the study is shown in Fig. 1. A detailed description of all methods employed in this study is provided as supplementary material (Supplementary material, File S1), while selected immunohistochemistry (IHC) images can be seen in Fig. 2.
Categorical data are displayed as frequencies and corresponding percentages, with continuous data as median and range. Comparison of categorical data was performed by Fisher’s exact or Pearson chi-square tests. Disease-free survival (DFS) and overall survival (OS) were calculated according to the STEEP system . Time-to-event distributions were estimated using the Kaplan–Meier method and compared using the log-rank test.
Cox regression analyses were performed for DFS and OS, to assess the prognostic or predictive significance of biomarkers for paclitaxel treatment. A backward selection procedure with a removal criterion of p > 0.10 was performed in order to identify significant factors among menopausal status, paclitaxel treatment, subtype classification, involved axillary lymph nodes, tumor grade and size, type of surgery, histological type, hormonal therapy (HT), RT, EGFR, PTEN, p-mTOR, p-Akt308, and p-Akt473 protein expression, and PIK3CA mutational status. The final model was adjusted for paclitaxel treatment and immunophenotypical subtypes. Significant clinical combinations of the examined markers were also assessed in univariate and multivariate analyses. Results of this study were presented according to reporting recommendations for tumor marker prognostic studies . The statistical analysis was conducted using the statistical software SPSS (version 18.0, IBM Corporation, Armonk, N.Y., USA) and SAS (version 9.3, SAS Institution Inc., Cary, N.C., USA).
Basic clinicopathological characteristics of the study population and their association with the examined biomarkers are provided as supplementary material (Supplementary material, File S2). DFS and OS did not differ significantly between treatment groups. At a median follow-up of 105 months (range 0.1–167), the 5-year DFS rates were 74, 70, and 74 %, while the OS rates were 88, 82, and 86 % for the E-T-CMF, E-CMF, and ET-CMF groups, respectively.
Distribution of the examined markers in the total study population
In the total cohort, EGFR protein expression was positive in 16.5 % while PTEN loss was observed in 53 % of the patients. Cytoplasmic p-Akt473 protein expression was positive in 56.1 %, nuclear p-Akt473 in 22.1 %, while any p-Akt473 positivity (cytoplasmic or nuclear or both) was present in 63.7 % of the cases. p-Akt308 was positive in 80.9 % and p-mTOR in 74 % of the patients. PIK3CA mutations were found in 23.7 % of the cases (Table 1).
Distribution of the examined markers in the total study population
N = 997
No loss ( > 10 %)
Positive (either cytoplasmic or nuclear)
Association of the markers with breast cancer subtypes
The association between the examined markers and breast cancer subtypes are presented in Table 2. Of note, EGFR protein expression was positive in the majority of the triple-negative breast cancer (TNBC) patients (62.7 %), while the rest of the subgroups had significantly lower percentages (luminal A < luminal B < luminal-HER2 < HER2-enriched, p < 0.001). p-mTOR protein expression was positive in the majority of the patients across all breast cancer subtypes (63.8–79.9 %), being most often positive in the luminal-HER2 subtype and less frequently positive in the TNBC subtype (p = 0.046). PIK3CA mutations were more common in the luminal A and luminal B subtypes, and this was a less common event in HER2-enriched tumors (p < 0.001). Interestingly PTEN loss was more common in the luminal A and TNBC subtypes (p < 0.001), while high p-Akt473 protein expression was most commonly seen in the HER2-enriched subtype (p = 0.007).
Association of markers with breast cancer subtype classificationa
No loss (≥ 10 %)
Positive (cytoplasmic or nuclear)
Associations between markers
The associations between the examined markers are presented in Table 3. p-mTOR protein expression was negatively associated with EGFR expression (p = 0.041) and positively associated with PTEN protein expression (p < 0.001), p-Akt473 (both cytoplasmic and combined; p < 0.001) and PIK3CA mutations (p < 0.001). EGFR was positively associated with cytoplasmic p-Akt473, p-Akt308, and PIK3CA wild-type tumors (p = 0.007, p = 0.035, and p < 0.001, respectively).
PTEN was more commonly expressed when p-Akt (nuclear, cytoplasmic, or combined) was activated and in PIK3CA mutated tumors (p = 0.005). Furthermore, p-Akt308 protein expression was significantly positively associated with p-Akt473 expression (nuclear, cytoplasmic, and combined, p < 0.001, p = 0.036, and p < 0.001, respectively).
In univariate analysis, only EGFR expression among all examined markers was associated with OS. More specifically, EGFR-positive patients were associated with poor OS (HR = 1.42, 95 % CI 1.06–1.92, Wald’s p = 0.021), although no statistically significant association with DFS was observed (Fig. 3). Examining the coexpression of EGFR and p-mTOR, it was found to be associated with increased risk for death (HR = 1.53, 95 % CI 1.09–2.16, p = 0.015) and a trend for increased risk for relapse (HR = 1.32, 95 % CI 0.97–1.80, p = 0.079; Fig. 4).
Furthermore, a trend for a significant interaction of p-mTOR with paclitaxel treatment was observed for DFS. More specifically, among positive p-mTOR tumors, patients treated with paclitaxel had marginally lower risk for relapse compared with the patients not treated with paclitaxel (HR = 0.76, 95 % CI 0.57–1.02, Wald’s p = 0.065), while in the negative p-mTOR subgroup no difference between paclitaxel- and non-paclitaxel-treated patients was found (HR = 1.66, 95 % CI 0.76–3.60, p = 0.20). Finally, a significant interaction of p-Akt308 with adjuvant HT was found for both DFS and OS. More specifically, among p-Akt308-negative tumors, patients who were treated with HT were associated with decreased risk for relapse (HR = 0.42, 95 % CI 0.22–0.84, Wald’s p = 0.013) and death (HR = 0.26, 95 % CI 0.12–0.55, p < 0.001). Among p-Akt308-positive tumors, no significant differences were observed for patients treated with adjuvant HT (p = 0.58 for DFS and p = 0.22 for OS). None of the examined markers was of any prognostic or predictive value according to RT and immunophenotypical breast cancer subtypes (p values for interaction > 0.05 for both DFS and OS).
In multivariate analysis presented by forest plots (Fig. 5), large size (> 5 cm vs. < 2 cm) and four or more positive nodes were found to be significant negative prognostic indicators for both DFS and OS. Adjuvant HT was favorable for both DFS and OS in the p-Akt308 negative subgroup, while the coexpression of EGFR and p-mTOR was a negative prognostic indicator for OS.
Analysis of locoregional control
In the total cohort, locoregional relapse was observed in 50 patients (5.0 %), distant metastases in 267 patients (26.8 %), while both locoregional and distant metastases were found in 21 patients (2.1 %). In an attempt to evaluate the possible association with RT, we examined the prognostic/predictive significance of all markers in the subgroup of patients with locoregional relapse. In univariate analysis, EGFR and mTOR protein expression among all examined markers was associated with DFS (log-rank, p = 0.032 and p < 0.001, respectively, Fig. 6) in such patients; however, no statistically significant associations with OS were observed. More specifically, EGFR-positive patients with locoregional relapse were associated with increased risk for earlier relapse (HR = 1.92, 95 % CI 1.05–3.53, Wald’s p = 0.034), while mTOR-positive patients were associated with decreased risk for earlier relapse (HR = 0.27, 95 % CI 0.13–0.59, p < 0.001), as shown in the Kaplan–Meier curves of Fig. 6.
In the present study, we evaluated the association of the protein expression of EGFR, PTEN, p-Akt (308 and 473), and p-mTOR and the PI3KCA mutational status in FFPE tissue samples from 997 breast cancer patients who participated in two adjuvant breast cancer studies with anthracycline-based chemotherapy, aiming to evaluate the prognostic or predictive value of these markers and the possible associations between them. Adjustments were made according to treatment (paclitaxel- vs. no-paclitaxel-containing regimens) and to the five well-known breast cancer subtypes (luminal A vs. luminal B vs. luminal-HER2 vs. HER2-enriched vs. TNBC). As stated, patients received trastuzumab upon relapse only, since trastuzumab received approval in the adjuvant setting in 2005.
In a recent study, p-mTOR protein expression correlated with poor outcome in early-stage triple-negative breast cancer . In our cohort, none of the examined markers were of any prognostic or predictive value according to immunophenotypical breast cancer subtypes or to treatment (paclitaxel- vs. no-paclitaxel-containing).
Regarding EGFR, several retrospective immunohistochemical studies have highlighted EGFR protein expression as a negative prognostic indicator for OS [7, 18–20], whereas other similar studies have failed to confirm this finding [21–23]. These studies collectively reported an overall incidence of EGFR protein expression in breast cancer of 18–35 %. An important issue is that immunohistochemistry is not the best way to evaluate the role of EGFR in breast cancer, since increased EGFR signaling is often associated with increased EGFR turnover .
In our study, EGFR, as assessed by immunohistochemistry, was positive in 16.5 % of the total study population. Most of the TNBC cases were EGFR-positive (62.7 %), while in the luminal A and luminal B subtypes the corresponding percentages were 2.5 % and 7.9 %. In univariate analysis adjusted to treatment, EGFR expression was a negative prognostic factor for OS, although no statistical association with DFS was apparent. In multivariate analysis, such prognostic value for OS could not be confirmed.
PTEN loss was associated with low p-mTOR, p-Akt473, and p-Akt308 protein expression and wild-type PI3KCA tumors. In univariate and multivariate Cox regression analyses, no prognostic or predictive value was identified for PTEN loss. These findings suggest that Akt activation was not the result of PTEN loss, which appears to be in contradiction to some of the existing literature. However, similar findings with the ones reported here have been observed in other studies [24, 25], hypothesizing that PTEN loss activates Rho family proteins, leading to increased cell proliferation through pathways independent of Akt. Furthermore, Akt and its downstream proteins are activated in about 30 % of breast cancer cases, independently of PTEN loss. Two distinct studies suggested HER2 and ER as upstream positive regulators of Akt [24–26].
Many studies have reported PTEN loss and PIK3CA mutations as being mutually exclusive, while others suggested just the opposite . In our study, PTEN loss and PIK3CA mutations were not mutually exclusive, but PTEN loss was a more common event among PIK3CA wild-type tumors.
Low expression of p-Akt308 was a positive prognostic factor for DFS and OS in multivariate analysis for patients who received hormonal therapy after completing adjuvant chemotherapy (p < 0.001). High expression of p-Akt308 was of no prognostic value. This is in agreement with other reported series underlying the role of Akt as a predictor of resistance to endocrine therapy , while supporting a role for Akt or mTOR inhibitors in reversing this resistance. To our knowledge, our study, although retrospective, is the biggest one in terms of the number of cases being evaluated. p-Akt473, in univariate and multivariate analyses, was of no prognostic value.
When considered alone, p-mTOR was of no prognostic or predictive value. An important finding of our study is that, in univariate and multivariate analyses, coexpression of EGFR and p-mTOR versus no-coexpression was associated with worse OS (HR = 1.49, 95 % CI 1.03–2.15, p = 0.015), while there was no statistical association with DFS. In addition, when the significance of all markers was examined in the subgroup of patients with locoregional relapse, which might reflect a possible predictive value for RT [28, 29], positive EGFR protein expression was found to be associated with increased risk for earlier relapse, while positive mTOR protein expression was associated with decreased risk for earlier relapse.
Biomarkers and surrogate end-points have great potential for use in clinical oncology, but their statistical validation presents major challenges and very few biomarkers have been robustly confirmed [30, 31]. With the emergence of the use of mTOR inhibitors, it would be of interest to retrospectively evaluate p-mTOR protein expression in studies like BOLERO-II, trying to show possible associations of p-mTOR protein expression with response to such agents. As a second step, it would be necessary, in order to reach the highest level of evidence, that a randomized trial with an “interaction design” (one in which all patients are stratified according to biomarker expression and then randomized to one of two treatments) be undertaken.
In early breast cancer, low protein expression of p-Akt308 was associated with improved DFS and OS among patients treated with hormonal therapy following adjuvant chemotherapy. Coexpression of EGFR and p-mTOR was found to be associated with worse OS. In the subgroup of patients with locoregional relapse, positive EGFR protein expression was found to be associated with increased risk for earlier relapse, while positive mTOR expression was associated with decreased risk for earlier relapse.
The authors are indebted to all patients and their families for their trust and participation in the HE10/97 and HE10/00 trials and for the provision of biological material for research purposes. The authors also wish to thank all HeCOG personnel (data managers, research assistants, and monitors) for their dedication, M. Moschoni for data coordination, T. Spinari for collection of FFPE tissue blocks, and S. Dallidou for secretarial assistance.
Conflict of interest
G. Lazaridis, S. Lambaki, G. Karayannopoulou, A.G. Eleftheraki, I. Papaspirou, M. Bobos, I. Efstratiou, G. Pentheroudakis, N. Zamboglou, and G. Fountzilas state they are supported by an internal Hellenic Cooperative Oncology Group (HeCOG) translational research grant (HE TRANS_BR).