|Home | About | Journals | Submit | Contact Us | Français|
To identify novel genetic markers predictive of clinical benefit from epidermal growth factor receptor-directed antibody therapy in patients with metastatic colorectal cancer (mCRC).
Seventy-six consecutive patients who received cetuximab or panitumumab, either alone or in combination with chemotherapy, with available tumor tissue were included. Tumor tissue was tested for mutations at known hotspots in the KRAS, BRAF, PIK3CA, PIK3R1, AKT1, and PTEN genes by pyrosequencing. PTEN promoter methylation status was analyzed by methylation-specific PCR, and expression determined by immunohistochemistry (IHC). Forty-four patients had ≥ 4 weeks of therapy and were considered for clinical correlates.
Consistent with previous studies, KRAS gene mutations were associated with a shorter progression free (PFS) and overall survival (OS). Among the KRAS wild type patients, preservation of PTEN expression and PIK3CA WT status was associated with improved OS (median OS, 80.4 vs 32.5 weeks, HR: 0.33, p=0.0008) and a trend towards improved PFS (median PFS, 24.8 vs 15.2 weeks, HR: 0.51, p=0.06), compared to PTEN negative or PIK3CA mutant tumors. PTEN methylation was more common in the metastases than the primary (p=0.02). Simultaneous presence of methylation and mutation in the PTEN gene was associated with IHC negativity (p=0.026).
In addition to KRAS mutation, loss of PTEN expression (by IHC) and PIK3CA mutation is likely to be predictive of lack of benefit to anti-EGFR therapy in mCRC. PTEN promoter methylation and mutation status was predictive of PTEN expression, and may be utilized as an alternative means of predicting response to EGFR-targeted therapy.
Colorectal cancer (CRC) is the second leading cause of cancer mortality in the US with over 52,000 deaths in 2010 . The median survival of patients with metastatic CRC has improved to 24–30 months [2, 3], largely due to the availability of newer therapeutic options, including the epidermal growth factor receptor (EGFR) targeted monoclonal antibodies (mAb), cetuximab (Erbitux®, Eli Lilly, Indianapolis, IN) and panitumumab (Vectibix®, Amgen Inc., Thousand Oaks, CA) [4, 5]. Response rates (RR) to anti-EGFR mAb range from 10–22% when used alone or in combination with irinotecan [6, 7]. Activation of the EGFR pathway occurs after ligand binding, leading to EGFR phosphorylation and oligodimerization at the cell membrane. This in turn triggers downstream signaling events including activation of the Ras/Raf/mitogen-activated protein kinase (MAPK), and the phosphoinositide-3-kinase-(PI3K)-PTEN-AKT-mTOR pathways [8, 9]. Anti-EGFR mAb inhibit these signaling pathways by inhibiting ligand binding.
The ability to predict the likelihood of clinical benefit to these agents is an area of intense research. It is well established that mutations in the KRAS gene predict for lack of response to anti-EGFR mAb [10–13]. BRAF gene mutation is considered a prognostic marker, however its predictive nature is uncertain [14–18]. Among the KRAS wild type (WT) patients, multiple studies evaluating EGFR based therapy have documented highly variable RR ranging from 17%–60% [10–12, 19–23]. This heterogeneity suggests that there may be other predictive variables, besides KRAS, that determine responsiveness to EGFR antibodies.
Our group was one of the first to demonstrate in vitro, that mutations in the PI3K signaling pathway also predict response to cetuximab . We observed that in CRC cell lines, activating PIK3CA mutations or loss of PTEN gene expression predicted resistance to cetuximab. Since then, several clinical reports have followed, with conflicting results; with some reports suggesting a predictive role of this pathway and others refuting this finding [24–30]. Notably, a recent report suggested that mutations in exon 20 of PIK3CA specifically predicted for resistance to cetuximab .
Importantly, multiple components of the PI3K signaling pathway have been shown to be mutated in colon cancer, including mutations in PIK3CA, PTEN, PIK3R1 (regulatory subunit of the PIK3CA gene, p85α ), and AKT1. Furthermore, loss of PTEN expression in CRC has been shown to be mediated by promoter hypermethylation [32, 33]. In this study we performed a comprehensive investigation of all components of the PI3K signaling pathway previously shown to be mutated or methylated in CRC, and linked perturbations in the pathway with anti-EGFR mAb response. Mutation status of KRAS and BRAF were also performed as a means of validating our findings with previous reports.
All patients with available archived tissue who had received cetuximab or panitumumab for the treatment of mCRC, at Montefiore Medical Center through April 2009, were identified by database searches and included in the study. Formalin fixed paraffin embedded (FFPE) tissue blocks containing at least 60% tumor tissue were identified by a clinical pathologist and used for all analyses. An IRB approved protocol was followed for the collection of patient samples and clinical information.
DNA was isolated from FFPE blocks using the Qiamp FFPE DNA kit (Qiagen, Valencia, CA) and analyzed for mutation at fifteen hotspots (Supplementary Table (ST)-I) by pyrosequencing using the allele quantification (AQ) mode . Pyrosequencing assays were designed using PSQ Assay Design Software v1 and performed on the PSQ HS 96 platform from Qiagen. DNA was amplified using polymerase chain reaction (PCR) with biotinylated primers. The primer sequences, nucleotide dispensation order and PCR cycle settings are listed in ST-2. A threshold value for the pyrosequencing (AQ mode) was determined using a panel of 22 CRC cell lines, for which the mutation status of these loci had previously been determined by conventional Sanger sequencing. Validated pyrosequencing assays have been previously described for KRAS [35, 36] and BRAF [35, 37, 38] genes.
PTEN expression was determined by immunohistochemistry using 5-µm thick sections from FFPE blocks. Slides were deparaffinized, rehydrated, and blocked in 3% H2O2 for 10 minutes followed by antigen retrieval (steaming for 30 minutes) using Dako Target Retrieval Solution (pH 9.0). Staining was performed in an automatic slide stainer (Dako LV-1 Autostainer) using the Dako universal staining system. The primary rabbit monoclonal anti-PTEN antibody (D4.3) XP (Cell Signaling) (dilution 1:50) was applied for 60 minutes at room temperature followed by secondary antibody (Dako Cytomation Envision+ system-HRP Labeled Polymer, anti rabbit) for 30 minutes and followed by DAB reagent (Cell Marque). Sections were scored semi-quantitatively by light microscopy by two pathologists, using a four-tier scoring system: 0-no staining; 1-weak staining; 2-moderate staining, and 3-strong staining (Figure II); and, additionally, the percentage of staining was given a score of 1(<10%), 2(10%–50%), or 3(>50%). The sections were subsequently classified as negative (<10%, weak or moderate staining), positive (>50%, moderate or strong staining) or indeterminate (<10%, strong; 10–50%, any staining; >50%, weak staining).
Genomic DNA was bisulfite treated using EZ DNA Methylation kit from Zymo Research (Orange, CA). Methylation specific PCR (MSP) was performed at four independent CpG sites within the PTEN promoter previously shown to be prone to methylation in CRC (ST-3) [33, 39–41], using ZymoTaq Premix (Zymo Research) . Reactions were carried out on a Peltier thermocycler in a 25 µL volume consisting of 12.5 µL of ZymoTaq premix solution, 2 µL each of 10 µM forward and reverse primers, 2 µL of bisulfite treated DNA template with approximately 25ng of DNA and 6.5 µL of DNAse, RNAse free water. Bisulfite converted genomic methylated DNA Standard (Zymo Research) and untreated genomic DNA were used as positive controls for the methylated and unmethylated PCR reactions, respectively, and water as a negative control. Presence of a MSP product was verified by ethidium bromide staining.
Medical records, including cancer diagnosis, treatment history, drug toxicities, and response to treatment of all patients were reviewed independently by two medical oncologists. Objective RR was determined based on established RECIST criteria . Patients who had received at least four weeks of therapy with cetuximab or panitumumab (with or without cytotoxic chemotherapy) were considered for clinical correlates. All data was tabulated using Epiinfo v 3.5 software (Centers for Disease Control, Atlanta, GA).
Descriptive statistics was used for patient characteristics. Analysis for progression free survival (PFS) and overall survival (OS) was carried out using Graphpad Prism v 5.00 software (GraphPad Software, San Diego, CA) and reported as Kaplan-Meier curves. The log-rank test was used to determine the p value. Categorical variables were analyzed using Fisher’s exact test (FET) for 2 variables and by Exact Contingency Table (ECT) for 3 variables. All tests were reported as two tailed. The primary hypothesis being tested was that loss of PTEN or presence of PIK3CA mutations will render resistance to the anti-EGFR agents. All other analysis was considered exploratory.
A total of 76 patients (Table I) and 122 tissue samples were identified (66 primary, 56 metastases, 30 matched pairs). All tissue samples from all the patients were used for mutation analysis. Forty-four patients had ≥ 4 weeks of therapy and were considered for clinical correlates. These patients had similar clinical characteristics as the entire set of 76 patients in terms of median age, sex ratio, racial distribution, and location of primary tumor. Among the 44 patients considered for clinical correlates, 36 (82%) had irinotecan refractory disease, and 15 (34%) were treated with single agent therapy.
Four of the 44 patients evaluable for clinical response demonstrated a RECIST defined partial response (PR, 9.1%), and 18 had stable disease (SD, 40.9%). Notably, the patients with a PR appeared to have a superior PFS and OS than those without (table IV). Among the 4 patients with PR, one had a dual mutation in the KRAS (G12D) and the PIK3CA (E545K) genes. None of the other 3 patients had any detectable mutation in the KRAS, BRAF or PIK3CA genes.
The overall prevalence of K-Ras mutation was 42.1% and of BRAF was 7.9%, consistent with prior data [19, 22] (Table II). We identified one tumor sample that harbored mutations in both BRAF (V600E) and KRAS codon 12, and another with mutations in KRAS codons 12 and 13. Consistent with prior observations, patients who were WT for KRAS had a better PFS and OS than KRAS mutant patients (Figure I, Panels A and B; Table IV). When mutations in both KRAS and BRAF were collectively considered, PFS and OS was further improved in the WT group (Table IV). With only 4 patients exhibiting a PR, we were unable to demonstrate an association of objective response with KRAS or BRAF mutation status.
The overall prevalence of PIK3CA mutation was 17.1% consistent with prior data  (Table II). We found one tumor sample that harbored mutations in both exons 9 (E545K) and 20 (H1047R) of the PIK3CA gene. As reported previously, we found a considerable overlap of KRAS mutations with PIK3CA mutations, with 25% of KRAS mutations also harboring PIK3CA mutations . Patients who were WT for PIK3CA had a better PFS than those with mutant PIK3CA; however, no difference in OS was observed (Table IV).
We next analyzed whether PTEN expression, PTEN mutation status, or PTEN promoter methylation status was predictive of response to anti-EGFR therapy.
The results of the IHC studies are displayed in Table III and Figure II. Overall, PTEN expression was considered positive by IHC in 42.6% of the samples. Among the primary tumor samples, the PTEN staining revealed positive expression in 43.9%, negative in 36.4%, and indeterminate in 19.7% of the samples. In contrast, in metastatic samples, positive expression was observed in 41.1%, negative in 28.6% and indeterminate in 30.6% of the samples. Thus, there appeared to be a small difference in negativity of PTEN between primary and metastases.
Patients with preserved PTEN IHC expression (either in the primary or the metastases) appeared to have a superior OS than patients with PTEN loss, however, we were unable to demonstrate a difference in the PFS (Table IV). To determine the utility of PTEN expression as an additional biomarker, we analyzed PTEN expression specifically in KRAS WT patients. Encouragingly, we observed a more profound effect in OS and PFS (Table IV).
The presence of truncating PTEN mutations was assessed in the PTEN gene in 5 previously described hotspots [32, 44, 45]. Of these, we observed the presence of mutations only in exon 8 (R355X), in 27.6% of patients and 19.7% of all samples (22.7% of primary samples and 16.1% of metastases). However, neither PTEN mutations in the primary nor in metastases predicted response to anti-EGFR therapy.
Finally, we determined whether PTEN promoter methylation predicted for clinical benefit from anti-EGFR therapy. PTEN promoter methylation was assessed using 4 pairs of primers (ST-3). Methylation was observed at only one of the 4 sites examined (Figure III). Overall, among the 76 patients, PTEN methylation at this site was observed in 28.8% of primary, and 44.6% of metastatic samples. Similar to PTEN mutations, no association between PTEN promoter methylation and benefit from anti-EGFR therapy was observed.
Given that PTEN loss and PIK3CA mutations are mutually exclusive , we analyzed the effect of perturbations of the PI3-kinase pathway in the KRAS WT patients by collectively considering the PTEN expression and PIK3CA mutation status. Patients who had preserved PTEN IHC expression (either in the primary or the metastases) and a WT PIK3CA gene, had a superior OS than patients who had PTEN loss (median OS, 80.4 vs 32.5 wk, HR: 0.33, p=0.0008)(Figure I, Panel C) and a very strong trend towards an improvement in PFS (median PFS, 24.8 vs 15.2 wk, HR: 0.51, p=0.06)(Figure I, Panel D).
We were fortunate to have matched pairs of samples from 30 patients enabling an in depth evaluation of the PTEN gene. There were 1–3 metastases with available tissue corresponding to each primary tumor. The median time from sampling of the primary tumor to that of metastases was 2 years [range: 0 (simultaneous sampling) to 10 years]. Among the 30 pairs, PTEN expression was preserved in 15 (50%) primary and 13 (43.3%) metastases (Table III). The concordance rate of PTEN mutations, methylation and expression between primary and metastatic sites was 73.3%, 66.7% and 46.5%, respectively (ST-4). Interestingly, methylation in the primary tumor was identified in only 10 (33.3%) patients and in metastases in 18 (60%) patients (Table III, FET 0.02)].
In transitioning from primary to metastases, 9 patients demonstrated a gain in methylation and 6 patients changed from PTEN positive to negative/indeterminate. Notably, of these 6 patients, 3 demonstrated a gain in methylation. Taken together, methylation may explain loss of PTEN staining in metastases in about 50% of the patients. One of the 6 patients had a gain in PTEN mutation during the transition.
We next focused on the simultaneous presence of a truncating mutation and promoter methylation among these 30 patients. We found 5 such samples, of which 4 stained negative and 1 indeterminate for PTEN IHC. Therefore, simultaneous presence of both mutation and methylation was responsible for PTEN loss in 80% of samples. This was in contrast to samples that lacked simultaneous dual hits to the PTEN gene, where only 38% of samples were PTEN IHC negative. This difference in PTEN staining between samples with dual mutation/methylation versus none or only mutation or only methylation was statistically significant (ECT, p=0.026).
Biomarker driven anti-cancer therapy underwent a paradigm shift when the USFDA and ASCO recommended that the use of anti EGFR mAb in mCRC be restricted to patients with a WT KRAS gene status, because of a lack of benefit in those with a mutation [47–49]. In fact, there appears to be a negative outcome when patients with a KRAS mutation are treated with the anti-EGFR drugs [50, 51]. Furthermore, there is emerging data that not all KRAS mutations are alike, and one particular mutation, G13D, may not be a valid biomarker of exclusion after all . In this retrospective study, we performed a comprehensive analysis of mutations in the PI3K/PTEN-AKT-mTOR pathway to identify further determinants of anti-EGFR treatment benefit. While we failed to identify mutations in AKT or PIK3R1 we observed that loss of expression of PTEN expression and mutations in PIK3CA, both of which hyper activate the PI3K/PTEN-AKT-mTOR pathway, are further predictors of anti-EGFR treatment response.
Consistent with our published pre-clinical observation , we observe that collective consideration of PIK3CA mutations and PTEN loss predict for an superior clinical outcome when treated with anti-EGFR therapy, with an improvement in OS and a strong trend in PFS, in the KRAS WT patients. A clinically meaningful approach is likely to be an evaluation for KRAS mutations, followed by PIK3CA mutation assessment and PTEN IHC staining in the KRAS WT tumors; offering anti-EGFR therapy for the KRAS WT/PIK3CA WT/PTEN positive patients. This approach could potentially wean out 60–70% of patients from therapy with these drugs, thereby markedly improving the clinical benefit rate. Thus, PTEN preservation is a potential biomarker of inclusion, while KRAS and PIK3CA are markers of exclusion, of therapy with anti-EGFR drugs. This identification of PTEN/PIK3CA status as a putative biomarker is consistent with prior reports [24, 26–29, 53–55]. Interestingly enough, one group has reported its findings on the possibility of using gene expression profiles to characterize KRAS, BRAF or PIK3CA activated like tumors, as a further refinement of clinical therapy with anti-EGFR agents .
We observed an interesting overlap of mutations within the same tumor. One sample harbored mutations in both BRAF (V600E) and KRAS codon 12, and another with mutations in KRAS codons 12 and 13. Such simultaneous mutations have been reported previously, as in BRAF and KRAS, KRAS codons 12 and 61, and in KRAS codons 12 and 13 [52, 57]. It appears that this is the first report of simultaneous mutations in codons 12 and 13. Furthermore, we found one tumor sample that harbored mutations in both exons 9 (E545K) and 20 (H1047R) of the PIK3CA gene. It is likely that as more tumor samples are analyzed for multiple somatic mutations, similar overlapping simultaneous mutations will continue to be identified. The co appearance of mutations in the MAPK and PI3K pathway has been reported extensively in the literature [24, 29, 52].
Another important observation is that simultaneous consideration of PTEN methylation and truncating mutation is significantly associated with loss of PTEN expression by IHC. Moreover, careful study of the matched pairs elucidated that there was an increase in methylation in the metastases as compared to the primary. We therefore propose a hypothesize that PTEN loss requires a dual “hit”, either by homozygous mutation, or an initial allelic loss by a truncating mutation, followed by a second allelic loss by promoter methylation, that takes place between the initial diagnosis and development of metastases. This methylation event is a potential mediator of PTEN loss in the metastases and is clearly worthy of further investigations in large scale tissue studies.
The lack of an association of PTEN methylation alone with IHC expression may be explained by the fact that we interrogated four sites by the MSP and may have missed other potential important sites of methylation. Broadening the panel of primers to interrogate more CpG islands may further improve the yield of methylation detection. A second possibility is that methylation alone is insufficient to downregulate PTEN expression. For example, in a recent publication, in patients with Cowden syndrome, methylation in the PTEN promoter did not result in transcriptional repression of gene expression . As regards PTEN mutations, we were unable to establish their association with overall outcome or response data. However, considering the ease of mutation testing, its role as a simple potential biomarker needs further exploration. Taken together, despite its limitations, PTEN IHC appears to remain the most robust assay to determine its expression.
The most significant finding in our study is clearly the confirmation that the loss of expression of PTEN (by IHC), and mutations in the PIK3CA gene are potential biomarkers of resistance to therapy with the anti-EGFR agents, particularly in patients with a WT KRAS in their cancer. These findings need to be further explored in large prospective randomized studies. Further identification of novel biomarkers will lead to small incremental steps towards truly individualizing therapy for patients with mCRC, thereby improving outcomes, reducing healthcare costs, and most importantly minimizing toxicity by the selective use of the most effective, and minimizing the use of drugs unlikely to benefit the patient.
This work is supported by a K-12 award from the National Cancer Institute of the National Institutes of Health (1K12CA132783-01A1 to SG) and an Advanced Clinical Research Award (ACRA) in colon cancer, by the ASCO (now Conquer) Cancer Foundation to SG.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Presented in part at the 2010 Gastrointestinal Cancers Symposium and the 2010 Annual Meeting of the American Society of Clinical Oncology.
CONFLICT OF INTEREST PAGE
John Mariadason and Sanjay Goel are co-applicants on a patent filed with the USPTO on the use of PTEN and PIK3CA mutations as predictive markers of efficacy of the anti EGFR agents. This patent application is currently under review at the USPTO.
A licensing agreement has been signed with Transgenomics Inc., should this patent be granted.
None of the other authors have any conflicts to declare.