Conflict of interest: G.P.: grant, Merck Serono; A.L.: honoraria, Merck Serono; O.B.: honoraria, Merck Serono, Amgen; J.-L.M.: research funding, Merck Serono; F.P-L.: honoraria and consult, Merck Serono, Amgen; P. L. P.: honoraria, consultant and research funding, Merck Serono Amgen and Myriad Genetics.
KRAS mutations are a strong predictive marker of resistance to anti-epidermal growth factor receptor (EGFR) antibodies in advanced colorectal cancer (CRC) but only a subset of wild-type (WT) KRAS patients are responders, suggesting the existence of additional markers of resistance to this treatment. The activation of EGFR downstream signaling pathways may be one of these ones. In a series of 42 patients with advanced CRC treated with cetuximab/panitumumab, for whom KRAS status was previously determined, we retrospectively analyzed the intratumor expression of EGFR downstream signaling phosphoproteins of the RAS/MAPK and PI3K/AKT pathways (pERK1/2, pMEK1, pAKT, pP70S6K and pGSK3β) using Bio-Plex® phosphoprotein array. Association with tumor response, progression-free survival (PFS) and overall survival (OS) was assessed. The expression of all the phosphoproteins was higher in KRAS mutated tumors than in WT tumors. The expression of pP70S6K was lower in responders than in nonresponder patients. In univariate analysis, patients with high pMEK1 or pP70S6K expression had a shorter PFS than those with low expression. Patients with high pP70S6K expression also had a shorter OS. In multivariate analysis, PFS was shorter for patients with high pMEK1 or pP70S6K expression, independently of KRAS status, as OS for patients with high pP70S6K expression. Therefore, WT KRAS patients with high pP70S6K expression had a shorter survival than those with low expression. Our results suggest the importance of EGFR downstream signaling phosphoproteins expression in addition to KRAS status to define the subgroup of patients who will not benefit from anti-EGFR therapy.
Recent progress has been made in the treatment of colorectal cancer (CRC) with the introduction of new therapies targeting the epidermal growth factor receptor (EGFR) and the vascular endothelial growth factor. Monoclonal antibodies represent one of the most important options for the inhibition of EGFR.
Cetuximab is a chimeric IgG1 monoclonal antibody, which binds to EGFR with a high specificity and blocks ligand-induced phosphorylation of the receptor. This prevents the phosphorylation of the intracellular tyrosine kinase domain of the receptor which subsequently inhibits activation of downstream signaling pathways like RAS/RAF/MAPK, PI3K/AKT and STAT. Inhibition of these pathways restores normal cell proliferation control and induces apoptosis.
Cetuximab has proved to be active in irinotecan-resistant metastatic CRC expressing EGFR by immunohistochemistry (IHC) in 2 phase II studies.1, 2 Moreover, a randomized phase III study showed a significant improvement of progression-free survival (PFS) and overall survival (OS) with cetuximab when compared to best supportive care in 572 patients previously treated by fluroropyrimidine, irinotecan and oxaliplatin.3 A survival benefit of a combination of chemotherapy plus cetuximab over chemotherapy alone was more recently reported in first-line in 2 randomized studies.4, 5 Similar results have been obtained with the fully human IgG2 anti-EGFR monoclonal antibody panitumumab which was associated with a higher response rate and a longer PFS when compared to best supportive care in chemo-refractory metastatic CRC patients.6 However, only a subset (8–23%) of patients achieved an objective response and benefit from cetuximab or panitumumab and EGFR expression based on IHC has failed to demonstrate any correlation with response to these antibodies.1, 2, 7, 8 It is therefore important to identify new markers of response and survival to anti-EGFR drugs.
KRAS mutations have been reported to be associated with lack of response to cetuximab and/or poorer survival in chemo-refractory metastatic CRC patients in several independent studies.9–15 The predictive and prognostic value of KRAS mutations was also confirmed in 2 randomized phase III trials comparing cetuximab or panitumumab monotherapy with best supportive care.9, 16 The hypothesis is that KRAS mutation could be responsible for an acquired activation of the RAS/MAPK signaling pathway downstream of the EGFR which would be independent of the ligand binding to the receptor, and which could therefore, induce resistance to anti-EGFR antibodies, as it has been demonstrated in vitro.10 However, only 28% to 64% wild-type (WT) KRAS patients respond to cetuximab,10–15 which suggests the existence of other molecular markers of resistance to this treatment. More recently, it has been suggested that BRAF mutation could also be responsible of resistance to anti-EGFR monoclonal antibodies.10, 17
As the key mechanism involved in resistance to anti-EGFR therapy appears to come from the constitutive activation of EGFR downstream signaling pathways, the aim of this study was to evaluate the expression of EGFR downstream signaling phosphoproteins of the MAPK (pERK1/2, pMEK1) and PI3K/AKT (pAKT, pP70S6K, pGSK3β) pathways as potential new markers of response to anti-EGFR treatment in metastatic CRC patients. The expression of these selected phosphoproteins was investigated by Bioplex® phosphoprotein array (Bio-Plex®, Bio-Rad, Hercules, CA), which has been demonstrated to have a good reproducibility and to be significantly correlated with western blot analysis.18–20
Material and Methods
In the present study, 42 patients (24 males; mean age: 61.8 years) with histologically proven metastatic colorectal adenocarcinoma treated with cetuximab (n = 41) or panitumumab (n = 1) were analyzed. These 42 patients were those from a previous series of 114 patients15 evaluated for tumor response for whom protein material was available. An analysis of EGFR expression was performed by IHC on at least 1 tumor fragment and was considered EGFR positive if at least 1% malignant cells stained (Zymed Laboratories Inc., San Francisco, CA, or Dako, Glostrup, Denmark). This retrospective study was performed according to the French ethics laws.
One patient received panitumumab in monotherapy (6 mg/kg every 2 weeks) and 41 patients received cetuximab, in monotherapy (n = 2) or in combination with irinotecan (alone, n = 37; FOLFIRI regimen, n = 2) (initial dose of 400 mg/m2 followed by weekly infusions of 250 mg/m2). The anti-EGFR treatment was given as second-line, third-line, fourth-line, fifth-line or more in 16, 17, 5 and 3 cases, respectively. One patient received cetuximab as first line without any chemotherapy. All the patients, except one, had progressed under an irinotecan-based chemotherapy and were therefore treated by cetuximab according to the US Food and Drug Administration guidelines. The median follow-up was 10.1 months.
Tumor response was evaluated by computerized tomodensitometry according to the RECIST (Response Evaluation Criteria in Solid Tumors) criteria.21 In this study, patients with complete response (CR) and partial response (PR) were classified as responders, and those with stable disease (SD) and progressive disease (PD) as nonresponders.
DNA extraction and KRAS, BRAF and PIK3CA mutation analysis
DNA was extracted from tumor samples obtained from primary colorectal tumor or metastatic tissue as previously described.15KRAS mutation status was determined retrospectively on tumor samples obtained before the anti-EGFR treatment by an allelic discrimination assay and checked by direct sequencing of exon 2 as previously described.15, 22
BRAF V600E mutation detection was also assessed by allelic discrimination using Taqman probes following the same protocol as KRAS mutation as previously described.23 Probes and protocol are available on request.
The entire coding sequence of PIK3CA was analyzed by direct sequencing using a Big Dye Terminator cycle sequencing kit and analyzed on an ABI Prism 3100 DNA analyzer automated sequencer (Applied Biosystems, Courtaboeuf, France). Primers and PCR conditions are available upon request.
All somatic mutations found were further validated by conducting a new independent amplification and sequencing procedure.
Protein extraction and EGFR downstream signaling phosphoprotein expression
Whole cell protein extraction was performed from 5-mm frozen tumor tissues using the Kit RIPA lysis Buffer 1X (Tebu-bio, Le Perray en Yvelines, France) prepared with inhibitors (PMSF, protease inhibitor cocktail and sodium orthonavate) (Dutscher, Issy-les-Moulineaux, France) according to the manufacturer's recommendations. BPA assay requires 15–20 mg of tissue containing more than 50% of tumor material24 or equivalent amount of protein extract (i.e., 25 μg total protein per assay in triplicate).
The expression of key phosphorylated proteins of the downstream EGFR signaling pathway (pMEK1, pERK1/2, pAKT, pP70S6K and pGSK3β) was analyzed using phosphoprotein array (Bio-Plex®, Marnes-la-Coquette, France). This technique is based on multiplex sandwich bead immunoassays.20 Protein extracts were transferred into 96-well dishes and diluted with 25 μl buffered solution. Fluorescent capture beads coupled to antibodies directed against the phosphoproteins (pAKT, pGSK3β, pP70S6K, pMEK1, pERK1/2) were mixed, and added into each well and incubated overnight. Following incubation, the plates were washed and incubated with biotinylated antibodies fixing each target protein. Streptavidin–phycoerythrin solution was then added. The analysis consisted in a double laser fluorescence detection allowing simultaneous identification of the target protein through the red fluorescence emission signal of the bead and quantification of the target protein through the fluorescence intensity of phycoerythrin. Results were recorded as mean fluorescence intensities and compared to negative controls. Positive controls consisting of standard protein extracts from cell lines were added to each series. All results were normalized through the different batches of analyses by the same mutated tumor sample. The expression level of each phosphoprotein was given in an arbitrary unit.
χ2 test was used to calculate the p value for association between KRAS mutation and response to cetuximab. Expression levels of each phosphoprotein were compared between group (i.e., KRAS mutated and nonmutated tumors or between responders and nonresponder patients) by the Wilcoxon ranksum test. The PFS was calculated as the period from the first day of cetuximab treatment to the date of tumor progression, to death from any cause or to the date of the last follow-up at which data point was censored. The OS time was calculated as the period from the first day of cetuximab treatment until death of any cause or until the date of the last follow-up, at which data point was censored. We dichotomized each phosphoprotein expression, by choosing the threshold optimizing the log-rank test for PFS and leaving at least 10 patients in both groups (low expression vs. high expression). Taking into account the number of tests performed (23 thresholds tested per phosphoprotein), we retained a threshold of significance of 0.001 for the log-rank test, corresponding to an adjusted p value of 0.023. Therefore with this threshold both PFS and OS were estimated by the Kaplan-Meier method and compared using the log-rank test. A univariate survival analysis of these dichotomized variables was performed using a Cox model. A multivariate Cox model was used to estimate the effect of KRAS mutation and phosphoprotein expression. Analysis was carried out using the STATA software (College Station, TX).
Tumor response and survival according to KRAS mutation status
An objective response to cetuximab was obtained in 28.6% of the patients (CR: 1, PR: 11). A KRAS mutation was present in 45% of the tumors (n = 19) and was significantly associated with the absence of response to cetuximab (p < 0.001; Table 1). In univariate analysis, median PFS and OS were longer when tumor was not mutated: 32 weeks, 95% Confidence interval (CI) [14.7–46] versus 8 weeks, 95% CI [6.1–9], (log-rank test, p < 10−4) and 13.9 months, 95% CI [6.5–21.6] versus 6.4 months, 95% CI [2.8–10.1] (log-rank test, p = 0.02), respectively.
Table 1. Response to anti-EGFR antibodies according to KRAS, BRAF and PIK3CA mutational status
Phosphoprotein expression and KRAS mutation status
The median [range value] values of protein expression were 49.2 [0–963], 54.18 [0–373.7], 28.1 [1.3–826.3], 22 [6.7–154.8] and 25.7 [0–251.9] for pAKT, pGSK3β, pMEK1, pERK1/2 and pP70S6K in arbitrary unit, respectively.
Phosphoprotein expression was evaluated according to KRAS mutation status. The expression of all the phosphorylated proteins was higher in KRAS mutated tumors compared to WT tumors, but the difference was statistically significant for pP70S6K, pGSK3β and pMEK1 only (Fig. 1).
Furthermore, 1 tumor that contained a BRAF mutation also demonstrated the highest expression value for pMEK1 in the group of KRAS nonmutated tumors (536 arbitrary units; Fig. 1). The 6 tumors that contained a PI3KCA mutation did not have significant distinct phosphoprotein signatures, even if high phosphoproteins expression was observed in KRAS mutated patients with a PIK3CA mutation (all but 2 PIK3CA mutated patients were also KRAS mutated; Fig. 1). As for KRAS mutations, all the patients carrying a BRAF or a PIK3CA mutation were nonresponders (Table 1).
Tumor response and survival according to phosphoprotein expression
A correlation between phosphoprotein expression and tumor response to cetuximab was found only for pP70S6K. The expression was significantly lower in responders compared to nonresponder patients (20.5 vs. 50 arbitrary units, respectively; p = 0.024; Fig. 2).
In Cox univariate analysis, PFS was longer for patients with low expression of pAKT, pGSK3β, pMEK1, pERK1/2 and pP70S6K (Table 2). A cut-off value for dichotomizing each variable was determined to maximize the PFS difference between the 2 groups with a minimal size of 10 patients in 1 group. The cut-off value was 37.9, 59.25, 79.3, 42, 16.3 for pAKT, pGSK3β, pMEK1, pERK1/2 and pP70S6K, respectively. Considering the number of tests used for determining the optimal cutoff, we retained the threshold of significance at 0.001. Patients with a pMEK1 expression level higher than the cut-off value demonstrated a shorter PFS than those with a low expression level (median PFS: 7 ± 0.1 weeks, 95% CI [4.4–12] vs. 20 ± 3.4 weeks, 95% CI [8.6–33]; p = 0.0001). Patients with a pP70S6K expression level higher than the cut-off value demonstrated a shorter PFS than those with a low expression level (median PFS: 8 ± 0.1 weeks, 95% CI8–16vs. 33.1 ± 0.04 weeks, 95% CI [14.8–58.1]; p = 0.0006; Fig. 3).
Table 2. Univariate Cox analysis for progression free survival according to phosphoproteins expression
In Cox multivariate analysis, the 2 models including KRAS and pMEK1 or KRAS and pP70S6K showed that the expression of these 2 phosphoproteins are associated with PFS independently of KRAS status (Table 3). The inclusion of both pMEK1 and pP70S6K expression level did not significantly improve the model. Finally PFS was significantly associated with pMEK1 and with pP70S6K when stratified by KRAS status (p = 0.002 and p = 0.01, respectively; Fig. 4). The expression level of pMEK1 and pP70S6K were investigated for association with the OS. In univariate analysis, a significant association was only observed between pP70S6K and OS. Patients with low levels of pP70S6K were associated with better OS (median OS: 6.5 ± 0.1 months, 95% CI [3.8–10.5] for the high expression level of pP70S6K versus 21.6 months, 95% CI [6.1–26.3] for the low expression level group of pP70S6K; p = 0.003). The log-rank stratified on KRAS status is still significant (p = 0.01; Fig. 5).
Table 3. Multivariate analysis for PFS including KRAS status and phosphoproteins expression
The RAS/RAF/MEK/ERK and PI3K/PTEN/AKT signaling cascades play critical role in the transmission of signals from growth factor receptors to regulate gene expression and prevent apoptosis. The RAF/MEK/ERK pathway is intimately linked with the PI3K/PTEN/AKT pathway. For example, it has been shown that KRAS activating mutations regulate activation of both pathways, but the effect of KRAS activating mutation in human colorectal tumors on the level of phosphorylation of different proteins of both pathways is not well known. Our study brings new information suggesting that KRAS activating mutations are linked to a significant increase of MEK1, P70S6K and GSK3β phosphorylation as compared to nonmutated tumors. Previous data, linking KRAS mutations with phosphoprotein expression, were derived from cell lines or xenograft tumors. In a recent article, Haigis et al. showed that MEK was highly phosphorylated in tumors expressing K-RasG12D, but pERK was not detectably upregulated in those same tumors.25 It is known from a previous report that KRAS-mutated tumor cells also induce activation of the PI3K-AKT cascade.26 Therefore, our results obtained from human CRC are in line with those expected from previous comparisons of KRAS mutated tumors with nonmutated. The development of multiple efficient chemotherapies in advanced CRC necessitates the identification of markers able to predict response to the different drugs. Monoclonal antibodies against EGFR represent one of the most promising agents in advanced CRC patients, either in chemotherapy-refractory patients but also in first-line treatment and have been the subject of several studies to identify predictive markers of response. Most of these studies9–15, 22, 27–35 have focused their research on somatic or germline molecular markers more or less directly involved in the EGFR pathway or in antibody dependant cell mediated cytotoxicity (ADCC), which are the 2 mechanisms of cetuximab anti-tumor effects. The most relevant findings concern somatic KRAS mutations for which the data are convergent in up to 1,000 analyzed patients in the literature, and show that somatic KRAS mutations are associated with resistance to cetuximab or panitumumab and with a shorter survival of the patients treated by these antibodies.4, 5, 9–15, 28, 31, 32 Although KRAS mutations appear to be a reliable marker for predicting resistance to anti-EGFR therapies, the absence of KRAS mutation is not systematically associated with response to these treatments since only 26% to 64% of WT KRAS patients are responders.10–15, 28, 31 These results imply the existence of other markers of resistance to this treatment.
Considering the multiplicity of signaling effectors, a more global and direct approach of evaluating the functionality of these signaling pathways may be achieved by measuring of the expression level of phosphorylated proteins that reflects the activation of the RAS/MAPK and PI3K/AKT pathways. To our knowledge, the present study is the first to analyze the expression of such phosphoproteins and to evaluate their clinical relevance, in addition to KRAS mutations in patients treated by anti-EGFR therapy. In the present study, the expression of the selected signaling phosphoproteins was investigated by multiplex Luminex® bead-immunoanalysis. Recent comparative studies36 established that Luminex® technology can be applied to detect multiple antigen-antibody reactions in patient samples in which these interactions are traditionally assessed individually. This method proved more sensitive than multiple individual ELISA determinations therefore validating the use of Luminex® bead-immunoanalysis for multiplex determination of cytokines. This technique was proposed and cross-validated with western blot analysis for determination of phosphoprotein in cell and tumor protein extracts in breast,37 head and neck38 and prostate cell lines18 exposed to cetuximab as well as in clinical frozen diagnostic specimens from breast and head and neck carcinomas.19, 20 Recently, using optimized standard operating procedures regarding sample size and total protein concentration range and monoclonal antibodies used for immunoanalysis, and on the basis of the US Food and Drug Administration guidelines, Bioplex® phosphoprotein array intraassay and interassay coefficients of variation revealed good reproducibility of the technique and the results achieved using Bioplex® phosphoprotein array analyses significantly correlated (p < 0.001) with those obtained with numerized western blot analyses.19 Furthermore, Bland-Altman analyses clearly demonstrated that Bioplex® phosphoprotein array could be used instead of western blot providing a unique way of analyzing multiple phosphoprotein expression in small clinical specimens. Although it opens up new perspectives in the comprehension of resistance mechanisms to anti-EGFR therapy, this technology remains sophisticated, difficult to standardize and so to be used in clinical practice, which is a major limit. Moreover, it requires fresh frozen tumor tissue, which is often difficult to obtain, as evidenced by the low number of tumors analyzed in this study.
The response rate to cetuximab or panitumumab therapy is linked to the level of expression of pP70S6K protein, suggesting that the activation of the PI3K/PTEN/AKT pathway is a critical factor in resistance to this type of therapy. Moreover, both pMEK1 and pP70S6K were significantly associated with PFS independently of KRAS status. Patients with high expression of these 2 phosphoproteins in their tumor had a shorter survival than those with a low expression. A high expression level of pP70S6K was also significantly associated with a shorter OS regardless of the KRAS status. The expression of pMEK1 and pP70S6K reflects the activation level of the MAPK and PIK3/AKT pathways, respectively. These preliminary data would suggest that a global analysis of the functionality of intracellular signaling pathways involved in carcinogenesis may give more information than analysis of single genes analysis in understanding the molecular mechanisms underlying resistance to anti-EGFR therapies. This enhanced understanding is of clinical relevance as it adds to KRAS mutation status in defining whose patients will not benefit from cetuximab. Indeed, patients with a WT KRAS tumor but with a high level of pP70S6K or pMEK1 expression had a significantly shorter survival than those with a low-level expression of these phosphoproteins. This may explain why a proportion of patients without KRAS mutation do not benefit from cetuximab. These findings also suggest that in these patients RAS/MAPK or PI3K/AKT signaling pathways are activated by mechanisms other than KRAS mutations. One possible mechanism could be activating mutations in other effectors such as BRAF or PI3KCA, present in approximately 10–15% and 15–25% of CRC, respectively, and which have already been reported to be “gain of function” mutations activating the RAS/MAPK and PI3K/AKT pathways, respectively.39–43 Our results showing high levels of phosphoproteins expression in the patient with a BRAF mutation and in KRAS mutated patients with a PI3KCA mutation support the hypothesis that mutations BRAF and PIK3CA mutations behave as KRAS mutations from a functional point of view. Recently, the loss of PTEN expression, a key negative regulator of PI3K/AKT pathway has also been shown linked to resistance to cetuximab or panitumumab.13 This result is consistent with the association between resistance to anti-EGFR therapies and the high expression of pP70S6K, the terminal effector of the PI3K/AKT pathway.
Nevertheless, our data should be considered with caution, first, because of all the limits inherent to the Bio-Plex® phosphoprotein array technique we outlined above, second, because these data derive from a small, retrospective cohort, but also because the level of phosphoproteins may be different between primary tumors that were analyzed in the present study and metastatic sites which are the targets of anti-EGFR therapy.
In conclusion, these experimental preliminary results, which have to be confirmed in the future and obtained by a technique used in routine, suggest the interest of measuring EGFR downstream signaling phosphoproteins expression in addition to KRAS mutation analysis for the prediction of response to cetuximab and the survival of advanced CRC patients. The expression of these phosphoproteins allows a global analysis of signaling pathways functionality and is likely to offer great insight than single gene analysis for understanding of molecular mechanisms underlying resistance to anti-EGFR therapies. Such an approach could offer valuable additional information to the KRAS status of tumor cells.
We thank Institut National du Cancer (PL06_119), the Ligue Nationale Contre le Cancer, the “Ligue Contre le Cancer, Comités Lorrains,” the Region Ile-de-France.