Monoclonal antibodies (mAbs) against the extracellular domain of the epidermal growth factor receptor (EGFR) have been introduced for the treatment of metastatic colorectal cancer (mCRC). We have reported recently that increased copy number of the EGFR can predict response to anti-EGFR mAbs and that patients might be selected for treatment based on EGFR copy number. Here, we show that mutations activating the RAS/RAF signaling pathway are also predictive and prognostic indicators in mCRC patients, being inversely correlated with response to anti-EGFR mAbs. In cellular models of CRCs, activation of the RAS signaling pathway by introduction of an activated K-RAS allele (Gly12Val) impairs the therapeutic effect of anti-EGFR mAbs. In cancer cells carrying constitutively active RAS, the pharmacologic inhibition of the mitogen-activated protein kinase (MAPK) signaling cascade improves anti-EGFR treatment based on mAbs. These results have implications for the identification of patients who are likely to respond to anti-EGFR treatment. They also provide the rationale for combination therapies, targeted simultaneously to the EGFR and RAS/RAF/MAPK signaling pathways in CRC patients. [Cancer Res 2007;67(6):2643–8]
- colorectal cancer
- targeted therapy
Cetuximab and panitumumab are two monoclonal antibodies (mAbs) directed against the epidermal growth factor receptor (EGFR; ref. 1). Both the chimeric antibody cetuximab and the fully human antibody panitumumab have shown remarkable clinical activity in ∼10% of patients with chemotherapy-resistant metastatic colorectal cancers (mCRC; refs. 2– 5). We have reported previously that most patients with mCRC who achieve tumor shrinkage from treatment with anti-EGFR mAbs exhibit increased EGFR gene copy number ( 2). In these patients, the tumor growth is likely to be driven predominantly by the EGFR pathway ( 6, 7). A fraction of patients with increased EGFR gene copy number respond, whereas patients with normal EGFR gene copy status are unlikely to respond to the therapy ( 8). We did not observe a statistical association between clinical response and the mutational status of the EGFR gene ( 2). This finding is in accordance with other studies, indicating that EGFR mutations are an infrequent event in CRC ( 9).
Regardless of their genetic status, after a variable period, CRCs develop resistance to the mAbs, thus severely impairing their therapeutic potential. For these reasons, it is a clear priority to understand the molecular and cellular basis of primary and acquired resistance to anti-EGFR mAbs, cetuximab and panitumumab. We hypothesized that the lack of primary response or the acquired resistance to anti-EGFR mAbs may be due to constitutive activation of the signaling pathways downstream of the receptor. Our hypothesis is based on the concept that the genes mutated in cancer work in organized pathways. Mutations within a single pathway are often “mutually exclusive,” as predicted if the functional effect of each mutation was similar ( 10, 11). Consequently, only one of the genes involved in a given pathway is generally mutated in any single tumor. This implies that pathways rather than single genes should be the focus of studies aimed at dissecting the molecular basis of targeted therapies. It also suggests that once the “culprit” pathway governing the oncogenic property of a cancer is identified, all the players involved in that pathway (and not only a specific gene) can become a suitable therapeutic target.
Previous work has shown that two main signaling pathways are triggered by activation of the EGFR ( 12). The first is initiated by the small G-protein RAS, which in concert with the protein kinase RAF activates the mitogen-activated protein kinase (MAPK) cascade. The second involves the lipid kinase phosphatidylinositol 3-kinase that triggers activation of the PDK1-AKT signaling machinery ( 13).
It has been reported recently that K-RAS mutations negatively correlate with the response to the mAb cetuximab ( 14). In our previous report, the mutational status of the EGFR signaling effectors (K-RAS, BRAF, or PIK3CA) was not statistically associated with the response, although the K-RAS mutations seemed to be slightly more frequent in nonresponsive patients ( 2).
To assess whether activation of these signaling pathways could be connected to the response to anti-EGFR mAbs, we used two complementary approaches. First, genetic analysis was used to identify which members of the pathway were oncogenically activated in mCRCs from patients that had received the anti-EGFR mAbs. Second, CRC cell models were used to assess whether activation of effectors downstream the EGFR pathway affected the response to anti-EGFR treatment.
Materials and Methods
Patient characteristics and treatments. We assessed tumors from 48 patients with mCRC enrolled into clinical trials of panitumumab (Amgen, Thousand Oaks, CA) or cetuximab (Erbitux, Merck, Milan, Italy) for treatment of EGFR-expressing mCRC at Ospedale Niguarda Ca' Granda (Milan, Italy). The retrospective analysis of tumor DNA sequences of K-RAS and BRAF was approved by the Ethics Committee of Ospedale Niguarda Ca' Granda. Patients were selected based on the availability of sufficient tumor tissue. All patients had EGFR-expressing mCRC and 1% or more malignant cells that stained for EGFR on immunohistochemical analysis with DAKO EGFR PharmDX kit (DakoCytomation, Glostrup, Denmark) done in the central laboratory of each clinical trial. Demographics of patients, treatment and line of treatment received, objective responses, and durations of response are summarized in Table 1 . In detail, 12 (25.0%) patients received cetuximab monotherapy, 25 (52.1%) received panitumumab monotherapy, and 11 (22.9%) received cetuximab plus irinotecan–based chemotherapy (Camptò, Aventis, Milan, Italy). Cetuximab plus irinotecan were administered to patients with proven refractoriness to irinotecan, defined as documented disease progression during, or within, 3 months of receiving an irinotecan regimen. Tumor response was assessed with computed tomography or magnetic resonance imaging by use of Response Evaluation Criteria in Solid Tumors criteria according to clinical protocols by radiologist investigators at Niguarda Ca' Granda Hospital. All of the patients progressed after treatment and all but one were assessable for progression-free survival analysis.
Mutational analysis. DNA was extracted from paraffin-embedded samples. For each patient, 10-μm sections were prepared. An additional representative 2-μm section was deparaffinized, stained with H&E, and analyzed for detailed morphology. Regions displaying tumor tissues were marked and the tissue was extracted as described previously ( 2). Exon-specific and sequencing primers were designed using Primer3 software ( 15) and synthesized by Invitrogen (Milan, Italy) as previously described ( 2). Conditions to amplify exon-specific regions by PCR from tumor genomic DNA and to identify mutations have been described previously ( 16). PCR was carried out in a volume of 20 μL using a touchdown PCR program as described previously ( 2). Purified PCR products were sequenced using BigDye Terminator version 3.1 Cycle Sequencing kit (Applied Biosystems, Milan, Italy) and analyzed with a 3730 ABI capillary electrophoresis system. Mutational analysis was carried out as described previously. Tumor tissue from patient 29 was limited in quantity and mutational analysis was therefore not technically possible for all exons.
CRIB assays. The RAF-CRIB domain pull-down experiment on cells transfected with mutated K-RAS or the control vector was carried out as described previously by Taylor and Shalloway ( 17). Cells were lysed in a lysis buffer containing 25 mmol/L HEPES (pH 7.3), 0.3 mol/L NaCl, 1 mmol/L MgCl2, 0.5 mmol/L EDTA, 20 mmol/L β-glycerophosphate, 1 mmol/L sodium vanadate, 1 mmol/L DTT, 0.5% Triton X-100, and 5% glycerol containing 0.5 mmol/L phenylmethylsulfonyl fluoride and 5 μg/mL each of aprotinin and pepstatin. Lysates were then incubated with slurry beads previously linked to glutathione S-transferase (GST)-RAF CRIB domain and GST alone for 2 h at 4°C, washed three times in lysis buffer, resuspended, and loaded on a 12% acrylamide gel. Western blots were carried out as described previously ( 2).
Cell culture and transfections. CRC cell lines were obtained from the American Type Culture Collection (Manassas, VA) repository with the exception of the DiFi cells that were supplied by Jose Baselga (Vall d'Hebron University, Barcelona, Spain). Cells were grown in medium supplemented with 10% FCS and antibiotics. The constitutively active K-RAS expression vector used was pDCR-H-RasV12, kindly provided by Letizia Lanzetti (Torino Medical School, Candiolo, Torino, Italy). Empty vectors were used as controls for the transfections. Cells were transiently transfected using LipofectAMINE as suggested by the manufacturer.
Proliferation inhibition assay and Western blotting. For cell viability assays, cells were grown in medium supplemented with 2% FCS and incubated for 5 days with increasing concentrations of cetuximab (Komtur Pharmaceuticals, Freiburg, Germany) and/or PD98059 (Calbiochem, Milan, Italy). CRC cells were seeded in 96-well plates, and at the end of the drug incubation period, a tetrazolium salt–based reagent (CellTiter96 Aqueous One Solution, Promega, Milan, Italy) was added to each well according to the instructions provided by the manufacturer. After an incubation of 2 h, absorbance was read at 490 nm on a Beckman Coulter (Milan, Italy) DTX 880 plate reader. Values for control cells were considered as 100% viability. The triplicate values were all within 5% and the mean values were calculated and plotted with error bars representing the SD of triplicate samples from three independent experiments. Western blotting was done as described ( 2).
Statistics. Data have been analyzed using Stata 9.1 (Stata Corp., College Station, TX); all significance levels were set at P < 0.05. Qualitative comparisons of objective response to therapy (progressive disease, stable disease, and partial response) and gene mutations as predictors were done by the Fisher's exact test to check possible significance; the response to therapy was then analyzed by logistic regression using the presence of gene mutations as regressors. All logistic analyses were checked by means of pseudo R2 value and post tested with Hosmer and Lemeshow goodness-of-fit test followed by receiver operating characteristic analysis to verify sensitivity, specificity, and positive/negative predictive values of each model. The time-to-progression (TTP) analysis was done after Kaplan-Meier and then survivor curves were compared by means of log-rank test. TTP was defined as the time from random assignment of each protocol (see patients characteristics and treatments) until first documented tumor progression or death.
Mutational profiling of the EGFR signaling pathways in patients treated with cetuximab or panitumumab. The mutational status of K-RAS (exon 2) and BRAF (exons 15 and 21) was assessed in a series of 48 patients who had received either cetuximab or panitumumab ( Table 1 and 2 ). Informative sequences were obtained for all cases with the exception of one patient (patient 29), for which the mutational status of BRAF could not be ascertained due to insufficient starting material. K-RAS mutations were detected in 16 of the 48 (33.3%) tumors, and BRAF mutations were detected in 6 of the 48 (12.5%) tumors ( Table 2). The most frequent K-RAS alterations in our samples were single amino acid substitutions in codon 12 [10 of the 16 (62.5%) mutated tumors]. Single amino acid substitutions were also detected in codon 13, but less frequently [6 of 16 (37.5%)]. The only BRAF mutation found in our samples was the previously described V600E substitution that was detected in six patients. Taken together, 45.8% of the analyzed tumors carried a mutation either in K-RAS or BRAF genes. K-RAS and BRAF mutations were mutually exclusive exactly as expected if they were acting in the same signaling pathway.
RAS/RAF mutations are negatively associated with response to cetuximab or panitumumab. Next, we attempted to assess whether the mutational status of K-RAS or BRAF was associated with the clinical response to anti-EGFR mAbs ( Table 2). The presence of K-RAS mutations was not significantly linked to objective response to therapy, with a trend toward a negative association with response (1 of 11 mutations versus 15 of 37 mutations for responders versus nonresponders; P = 0.073). BRAF mutations alone were also not significantly associated with objective response to therapy (P = 0.312). Importantly, however, the presence of K-RAS and/or BRAF mutations was negatively associated with partial response (P = 0.005) and the logistic regression confirmed this association (odds ratio, 0.071; 95% confidence interval, 0.08–0.619; P = 0.017). These results suggest that patients carrying either K-RAS or BRAF mutations are less likely to achieve partial response to therapy with anti-EGFR mAbs. Our data also indicate that signaling pathways rather than single genes should be the focus of genetic studies aimed at dissecting the molecular basis of targeted therapies.
RAS/RAF mutations negatively correlate with progression-free survival of mCRC patients treated with cetuximab or panitumumab. Next, we assessed whether in patients treated with anti-EGFR mAbs the oncogenic activation of downstream signaling pathways was associated with time to tumor progression, which is a more compelling evidence of clinical benefit than objective response especially considering biological agents, such as mAbs. The TTP analysis showed a significantly worse outcome for subjects bearing a mutated K-RAS allele in their tumors compared with those carrying wild-type (WT) K-RAS (P = 0.0443; Fig. 1A ). The same was not observed for CRC patients carrying the BRAF mutation (P = 0.5369), probably due to the limited number of tumors carrying these mutations compared with those carrying the K-RAS mutations. However, the presence of either one or other mutation was still significant (P = 0.0259; Fig. 1B). This could imply that constitutive activation of the RAS/RAF signaling pathway impairs the response to therapy based on anti-EGFR antibodies and is in good agreement with what observed in other models, such as the response of lung cancer to gefitinib or erlotinib ( 8, 18).
Oncogenic activation of the RAS signaling pathway impairs the response of CRC cells to cetuximab. The results we obtained from the genetic profiling of CRCs suggest that the presence of mutated K-RAS negatively interferes with the clinical response to anti-EGFR mAbs, such as cetuximab and panitumumab. They also indicate that the acquisition of a secondary K-RAS mutation (or an equivalent event activating the RAS/RAF pathway) might be a mechanism by which tumor cells initially responsive to anti-EGFR mAbs become resistant to this therapeutic regimen. To test this hypothesis, we used cellular models of CRCs. Specifically, we took advantage of a naturally occurring CRC cell line (DiFi) carrying a 20-fold amplification of the EGFR gene ( 2). We have shown previously that the proliferation of DiFi cells is inhibited by nanomolar concentration of cetuximab; therefore, these cells represent a suitable model to understand the molecular and cellular basis of sensitivity to anti-EGFR mAb therapy ( 2). Furthermore, our mutational analysis of K-RAS and BRAF indicated that both genes were WT in DiFi cells (data not shown), thus allowing further functional experiments.
An activated allele of K-RAS (Gly12Val) was then introduced into DiFi cells alongside with the corresponding control vector using lipid-mediated plasmid DNA transfection. The transfection efficacy was verified using a plasmid encoding for the GFP gene. To assess whether we had successfully expressed an active RAS protein into transfected DiFi cells, we measured the amount of RAS bound to GTP. To this end, we did a RAF-CRIB domain pull-down experiment on cells transfected with mutated K-RAS or the control vector (see details in Materials and Methods). The results clearly indicated that cells transfected with the K-RAS (Gly12Val) plasmid have a significantly higher amount of RAS protein bound to GTP when compared with those transfected with the control vector ( Fig. 2A ). We then treated DiFi cells expressing mutated and WT RAS with increasing concentrations of cetuximab (from 5–20 nmol/L) and measured the percentage of cell viability. We found that cells expressing the activated RAS (K-RAS Gly12Val) were less sensitive to the drug as indicated by their markedly increased survival upon treatment with cetuximab ( Fig. 2B). The experiment was repeated twice and each time we obtained comparable results and P values.
These results indicate that the presence of oncogenically active K-RAS (such as the Gly12Val mutant) impairs the therapeutic potential of cetuximab in CRC cells. They also suggest that cancers responsive to anti-EGFR mAbs could become resistant to this drug by acquiring a secondary genetic lesion that triggers the constitutive activation of the RAS/RAF/MAPK signaling pathway.
Effect of combinatorial therapies simultaneously targeting the EGFR and the RAS/RAF/MAPK signaling pathways in CRC cells. Our results suggest that constitutive oncogenic activation of the RAS/RAF signaling pathway significantly impairs the therapeutic potential of mAbs (such as cetuximab) aimed at targeting the EGFR in CRCs. It is therefore tempting to speculate that the concomitant blockade of both the EGFR and the RAS signaling could be effective in inhibiting the proliferation of cancer cells. Several inhibitors targeting the RAS signaling pathway have been developed ( 19, 20). Among them, the MAPK/extracellular signal-regulated kinase kinase inhibitors, such as PD0325901 and ARRY-142886, have shown good preclinical activity ( 20, 21) and have entered recently clinical trials. To assess whether concomitant inhibition of the EGFR and of the RAS signaling pathway could be effective, we took advantage of PD98059, a powerful cell-permeable inhibitor of MAPKs that acts as downstream effector of RAS ( 22). For these experiments, we selected CRC cell lines that have been found previously to carry either WT (HT29 and HCT8) or mutated (HCT116 and DLD-1) K-RAS. To confirm that HCT116 and DLD-1 cells indeed harbor mutated K-RAS alleles, we extracted the corresponding genomic DNA and sequenced the K-RAS locus. We confirmed that HCT116 and DLD-1 display the same Gly13Asp K-RAS mutation (data not shown). The cell lines were treated with increasing concentration of cetuximab and PD98059 and their viability was measured ( Fig. 3 ). These experiments indicate that, in cancer cells harboring oncogenic K-RAS, the simultaneous inhibition of both the EGFR and the RAS downstream signaling effectors can be more effective than targeting either pathway alone.
In the last 5 years, cancer therapy has undergone a major revolution characterized by the introduction of targeted drugs that inhibit specific molecules. Among those, the mAbs (cetuximab and panitumumab) targeting EGFR have shown remarkable efficacy in the treatment of mCRCs. Similar to other targeted therapies, anti-EGFR drugs are active only in a fraction of patients and most of them subsequently become resistant to the treatment. Accordingly, two major challenges need to be addressed to optimize the efficacy of anti-EGFR therapies. The first is to identify the genetic alterations associated with the clinical response to anti-EGFR mAbs. The second is the elucidation of the molecular basis for primary or acquired resistance to these drugs.
In the present study, we report that the presence of mutations (most of which are thought to be gain of function) in K-RAS and BRAF are associated with the lack of response to anti-EGFR mAb treatment in mCRCs patients. It has been reported previously that the Gly12Val K-RAS mutation is associated with increased risk of relapse in CRC ( 23). This implies that the negative association we found with the occurrence of K-RAS/BRAF mutations may not only be due to resistance or lack of response to anti-EGFR mAbs. However, the cellular models in which we inserted the Gly12Val mutation indicate that the presence of an active K-RAS allele directly affects the responsiveness of CRC cells to anti-EGFR mAb, such as cetuximab.
Our results have several implications. The first is that most patients with CRC carrying mutated K-RAS or BRAF are not likely to experience significant benefit on either cetuximab or panitumumab treatment. However, the same patients should not be excluded from anti-EGFR mAbs treatment as we found that there are few mCRCs, in which the presence of K-RAS mutations is compatible with a clinical response to this therapeutic regimen. The molecular determinants of response in this subset of patients are presently unknown, and this observation therefore warrants further investigations. The second suggestion stemming from our present work is that clinical trials designed to test multitherapies with both anti-EGFR and anti-MAPK inhibitors should be considered for mCRC patients.
Note added in proof: In the American Type Culture Collection catalog, thecolorectal cancer cell line HCT8 is also referred to as HRT-18. The COSMIC database(Catalogue of Somatic Mutations In Cancer, http://www.sanger.ac.uk/genetics/CGP/cosmic/) indicates that the HRT-18 cell line carries the KRAS G13D mutation.
Grant support: Italian Association for Cancer Research (AIRC; A. Bardelli and S. Siena), Italian Ministry of Health (A. Bardelli), Regione Piemonte (A. Bardelli), Italian Ministry of University and Research (A. Bardelli), Association for International Cancer Research UK (A. Bardelli), European Union FP6, MCSCs contract 037297 (A. Bardelli), and Oncologia Ca' Granda ONLUS (S. Siena). S. Benvenuti is supported by an AIRC fellowship.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Michele Nichelatti (Service of Clinical Epidemiology and Biostatistics, Oncology and Hematology Department, Ospedale Niguarda Ca' Granda) for criticisms and statistical analysis.
Note: S. Benvenuti and A. Sartore-Bianchi contributed equally to this work.
- Received November 9, 2006.
- Revision received January 17, 2007.
- Accepted January 18, 2007.
- ©2007 American Association for Cancer Research.