Objective We aimed to better clarify the role of germline variants of the FCG2 receptor, FCGR2A-H131R and FCGR3A-V158F, on the therapeutic efficacy of cetuximab in metastatic colorectal cancer (mCRC). A large cohort with sufficient statistical power was assembled.
Design To show a HR advantage of 0.6 in progression-free survival (PFS) for FCGR2A-HH versus the rest and FCGR3A-VV versus the rest, with an 80% power, 80 Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS) wild-type (KRAS-WT) and 52 KRAS-WT patients are required, respectively. This leads to a total sample size of 952 and 619 patients, respectively. Samples were collected from 1123 mCRC patients from 15 European centres treated with cetuximab alone or in combination with chemotherapy. Fc gamma receptor (FCGR) status was centrally genotyped. Two additional externally genotyped series were included.
Results Incidences of FCGR2A-HH and FCGR3A-VV in KRAS-WT patients were 220/660 (33%) and 109/676 (16.1%) respectively. There was no difference in median PFS (mPFS) for KRAS-WT patients with FCGR2A-HH (22.0 weeks; 95% CI18.8 to 25.2) versus non-HH (22.0 weeks; 95% CI 19.4 to 24.6) or for FCGR3A-VV (16.4 weeks; 95% CI 13.0 to 19.8) versus non-VV (23 weeks; 95% CI 21.1 to 24.9) (p=0.06). Median overall survival, response rate and disease control rate assessments showed no benefit for either HH or VV.
Conclusions No differences in mPFS were found between the FCGR polymorphisms HH and the others and VV versus the others in KRAS-WT mCRC patients refractory to irinotecan, oxaliplatin and 5-fluorouracil treated with cetuximab. We cannot confirm the effects of other IgG1 antibodies, which may be weaker than previously suggested. Other markers may be needed to study the actual host antibody response to cetuximab.
- Antibody Targeted Therapy
- Colorectal Cancer
- Genetic Polymorphisms
- Immune Response
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Germline variants in the FCG receptor (FCGR), FCGR2A-H131R and FCGR3A-V158F, have been suggested to influence the efficacy of rituximab and trastuzumab in lymphoma and breast cancer.
The role of FCGR polymorphisms in chemorefractory metastatic colorectal cancer patients treated with cetuximab is still unclear. Several studies have been published but with conflicting results.
The main reason for the conflicting and confounding results were the small sample sizes, the non-cross-validation between techniques and the lack of correction for multiple testing.
What are the new findings?
No differences in median progression-free survival (mPFS) for chemorefractory KRAS wild-type (KRAS-WT) metastatic colorectal cancer (mCRC) patients treated with cetuximab were observed for FCGR2A-HH versus non-HH or for FCGR3A-VV versus non-VV.
No benefit in terms of response rate (RR), disease control rate (DCR), and median overall survival (mOS) was observed for FCGR2A-HH versus non-HH or for FCGR3A-VV versus non-VV.
An advantage in DCR and mOS for the FCGR3A-FF subgroup compared to the others was observed in KRAS-mutant patients, but there was no advantage in RR or in mPFS. The inconsistencies in the comparisons of response, progression-free survival and OS for this genotype in KRAS-mutant patients indicate that these findings are exploratory at best, without a current clinical implication.
How might it impact on clinical practice in the foreseeable future?
FCGR polymorphisms do not predict response to cetuximab in chemorefractory KRAS-WT mCRC patients.
The antibody-dependent cell-mediated cytotoxicity (ADCC) does not seem to play a role in cetuximab response.
Other markers may be needed to better study the effect of the immune system to cetuximab activity and response in chemorefractory KRAS-WT mCRC.
Cetuximab, a chimeric immunoglobulin 1 (IgG1) monoclonal antibody (moAb) that targets the extracellular domain of the epidermal growth factor receptor (EGFR), has shown efficacy in metastatic colorectal cancer (mCRC) patients in combination with chemotherapy as first and second lines,1–4 and in monotherapy as third line.5 Although initially developed for all patients with mCRC, it became clear from the analysis of several studies in 2008 that the benefit of this drug is restricted to the KRAS wild-type (WT) population.6 Although the mechanisms of action of cetuximab are not completely known, it is likely that the competition in binding the EGFR with some of its natural ligands can be responsible for the inhibition of cell proliferation, angiogenesis and metastasis.7
Antibody-dependent cellular cytotoxicity (ADCC) has been identified as an important mechanism of tumour cell death. It is mediated by the bifunctional structure of IgG molecules. As a chimeric IgG1 monoclonal antibody (moAb), cetuximab has an antigen-binding fragment (Fab) that engages the tumour cell antigen and a crystalline fragment (Fc) that binds a Fc gamma receptor (FCGR) on an effector cell, such as natural killer (NK) cells, monocytes, or macrophages. The ADCC is initiated when the Fab and the Fc portions of the moAb bind the tumour cell antigen and the FCGR. FCGRs are composed of three distinct classes: FCGR1, FCGR2 and FCGR3. The second two can be further divided into FCGR2A/FCGR2B and FCGR3A/FCGR3B. FCGR1 has high affinity for IgG and can bind to monomeric IgG, FCGR2 and FCGR3, which have weaker affinity for the monomeric form, need multimeric immuno complexes to effectively interact.8
Single nucleotide polymorphisms (SNP) that affect affinity to human IgG have been demonstrated for FCGR2A and FCGR3A, with a histidine(H)/arginine(R) polymorphism at position 131 for FCGR2A and a valine(V)/phenylalanine(F) polymorphism for FCGR3A at position 158.9
The strongest evidence for a role of FCGR polymorphisms in the efficacy of therapeutic IgG moAbs comes from studies on follicular lymphoma patients treated with rituximab in combination with the cyclophosphamide, hydroxydaunorubicin, oncovin and prednisone (CHOP) regimen as first-line treatment. Several papers suggested an advantage for FCGR3A VV over the other SNPs.10–13 Other publications suggested a similar effect from FCGR3A VV or FCGR2A HH with trastuzumab in breast cancer.14 ,15 However, these results are less consistent, and recent data dispute the validity of this effect.16 ,17 Following these conflicting results, FCGR has not been adopted as a predictive biomarker by the medical community for the treatment of breast and lymphoma.
Several studies attempted to explore the role of FCGR SNPs in the treatment of cetuximab in mCRC.
The studies performed are diverse in treatment lines and selection of explored genotypes, thus resulting in conflicting and confusing associations. Most were small with an insufficient statistical power, did not correct for multiple testing and did not consider the KRAS status. Moreover, the use of different genotyping techniques without cross-validation contributed to these confounding results.
We aimed to follow the associations found in lymphoma and breast cancer for FCGR3A-VV and FCGR2A-HH and test their relevance in mCRC patients treated with cetuximab. A large database with sufficient statistical power was generated using the data from a European consortium18 to which additional cohorts from Japan and the USA were added. Our large sample size together with the cross-validation of techniques and the multiple testing contributed to the obtaining of conclusive results.
Materials and methods
The current study investigated chemorefractory mCRC patients treated with cetuximab alone or in combination with chemotherapy. The primary endpoint was median progression-free survival (mPFS) in the FCGR3A-VV KRAS WT subgroup versus the non-VV subgroup, and the FCGR2A-HH subgroup versus the non-HH subgroup. Sample size calculation assumptions included the mPFS in KRAS-WT patients to be 24 weeks and the incidence of KRAS mutation to be 40%. Based on Zhang et al,19 the incidence of a favourable FCGR2A-HH genotype was expected to be 14%, and that of FCGR3A-VV genotype to be 14% as well. To obtain an improvement in mPFS in the favourable subgroups versus the rest, with a HR of 0.6, a total of 80 KRAS-WT FCGR2A-HH and 52 KRAS-WT FCGR3A-VV patients were needed. This yielded a calculated total of 952 and 619 patients, respectively.
Fifteen European centres (a European consortium) and two additional cohorts from Japan and the USA contributed to the final analysis. Clinical data on treatment were collected from a total of 1123 patients (899 belonging to the consortium, 91 from Japan and 133 from the USA), including 645 men (57.4%) and 476 women (42.4%) with an age range of 22–85 years. The patients’ demographics are described in table 1.
The European consortium was comprised of 15 European centres that were contacted to provide predefined clinical parameters, normal tissue samples, or DNA extracted from blood, and at least two slides of formalin-fixed paraffin embedded (FFPE) tumour tissue, and one consecutive haematoxylin-eosin-stained slide. Nine of these centres had already participated in a previous European consortium and six new members were added.20 The specimens of two additional cohorts (91 patients from Japan and 133 patients from the USA) were analysed separately, but their clinical data were combined.
Tumour response was evaluated by the investigators according to the Response Evaluation Criteria in Solid Tumours21 or WHO criteria, and classified as complete response (CR), partial response (PR), stable disease (SD) or progressive disease. The response rate (RR) was classified as CR plus PR. Disease control rate (DCR) was classified as RR plus SD. Progression-free survival (PFS) was reported by the centres in weeks, and measured as the time from the first day of cetuximab administration until the observation of radiological or clinical progression, or death from any cause.
This study was conducted after obtaining approval by the local ethics review boards.
Tumour and normal DNA were extracted from FFPE slides after pathological review and macrodissection using the QIAamp DNA FFPE Tissue Kit. Germline DNA was extracted from blood samples of 349 patients by the DNeasy Blood and Tissue Kit (Qiagen, Belgium).
KRAS mutational analysis
The Sequenom platform was used to assess KRAS mutations. KRAS mutations in exons 12 (G12D, G12A, G12V, G12S, G12R and G12C), 13 (G13D), 59 (A59T), and 61 (Q61H, Q61P, Q61R and Q61L) were analysed as previously described.18
FCGR2A-H131R and FCGR3A-V158F genotyping and technical validation
The mutations, histidine/arginine-131 of CD32A (FCGR2A-131 H/R; rs1801274) and valine/phenylalanine-158 of CD16A (FCGR3A-158V/F; rs396991), are also known as FCGR2A 535A>G and FCGR3A 818A>C polymorphisms. To avoid confounding nomenclature, online supplementary figures S1 and S2 include illustrations from the Mutation Discovery database that show the sequence and flanking regions of the two SNPs that we analysed.
An allele-specific qPCR (Taqman) was performed by means of specific predesigned probes for the FCGR2A-131H/R and FCGR3A-158V/F polymorphisms (part No. C__9077561_20 and C__25815666_10, respectively, Applied Biosystems, Foster City, California, USA) and run on 7500 real-time PCR system (Applied Biosystems). The 25 µL reaction solution was incubated in a 96-well plate at 95°C for 10 min, followed by 50 cycles of 92°C for 15 s and 60°C for 1 min. An internal validation of the assays was performed by directly sequencing a set of 12 samples using primers (available upon request) that were able to detect the SNPs of interest and distinguish them from pseudogenes (see online supplementary figures S3 and S4). All the Taqman results were confirmed without exception.
External validation consisted of sending 18 samples to an independent centre (Leids Universitair Medisch Centrum (LUMC), The Netherlands)22 to confirm our results. Genotyping was performed on a Taqman 7500 (Applied Biosystems) with the same predesigned assays used in our analysis. A 100% compatibility was found between our centre and the LUMC, thus providing an independent technical replicate. The samples from Japan and those from the USC Norris Comprehensive Cancer Center were locally genotyped and were not available as DNA to us. The Japanese group used a method that combines the multiplex PCR and sequence-specific oligonucleotide probes with microsphere-based suspension array technology (Luminex, Austin, Texas, USA).23–25 The samples from the USA were analysed for FCGR2A H131R polymorphism by PCR followed by allele-specific restriction enzyme digestion,24 and for the FCGR3A V158F polymorphism by an allele-specific PCR reaction.25
The incidence of the genotypes in the European consortium as well as those in the Japanese and US series is reported in table 2.
The median PFS was our primary end point. RR, DCR and median overall survival (mOS) were our secondary outcome variables. Additional genotype grouping analyses were performed but regarded as exploratory.
A p value below 0.05 was considered to indicate statistical significance. The PFS and OS in the various subgroups were summarised with the use of Kaplan–Meier curves, and Kaplan–Meier estimates of OS and PFS probabilities are shown where applicable. The difference between these groups was compared with the use of the log rank test, with the HR and its 95% CI from a Cox regression model with a single covariate. The association between biological and clinical features was investigated using Pearson's χ2 test or Fisher's exact tests.
Data on 1123 patients were collected, exceeding the required target population. Patients’ characteristics and oncological outcomes are presented in table 1. The general FCGR status within the groups and according to KRAS status is provided in table 2. FCGR2A distribution did not concur with the expected distribution presented by Zhang et al19 (14% for FCGR2A HH). Polymorphism spread in the combined analysis of the three groups was 45.3% for HR, 32.1% for HH and 22.6% for RR. FCGR3A polymorphism for VF, VV and FF was 45.5%, 16.3% and 38.2%, respectively. Moreover, as reported in table 2, the genotype incidence between the Japanese series and the European consortium is significantly different for all genotypes, while the comparison between our consortium and the US series reveals a significantly more frequent incidence of the VV genotype compared to the non-VV genotype in the US series. As reported below, however, these genotype frequency differences are not associated with differences in biological effects (see online supplementary table S1). No clear association was seen between KRAS status and FCGR distribution. The only significant association was found for the FCGR3A-FF versus the non-FF (Fisher’s Exact test 1-sided p=0.034).
Outcome of FCGR genotypes in KRAS WT population
No difference in mPFS was observed for FCGR2A HH versus non-HH, with HH reporting 22 weeks (95% CI 18.7 to 25.2 weeks) and non-HH reporting 22 weeks (95% CI 19.4 to 24.5 weeks).
A difference was observed for FCGR3A-VV versus non-VV. FCGR3A-VV resulted in a mPFS of 16.4 weeks (95% CI 13.0 to 19.7 weeks) versus 23 weeks for non-VV (95% CI 21.1 to 24.8 weeks). Nevertheless, no statistical significance was reached. (p=0.06). Comparisons of the median PFS in the European consortium series with the Japanese and the US ones (see online supplementary table S1) revealed the same lack of effect in all of them, despite differences in genotype incidences, as expected. As reported in tables 3 and 4, our secondary end-points analysis for mOS, RR and DCR failed to show any benefit for FCGR2A-HH versus non-HH or for FCGR3A-VV versus non-VV.
Exploratory analysis: genotype combinations or different groupings
The outcomes of FCGR polymorphism subgroups in KRAS mutant patients are described in table 5.
Exploratory analysis showed an advantage in DCR for the FCGR3A-FF subgroup compared with the others (Pearson χ2 test, p=0.042). A similar advantage in mOS was observed for the FF subgroup compared with the others (Pearson χ2 test, p=0.03) but not for RR and mPFS. This finding is exploratory and not part of the main design. We also highlight the inconsistency between the response, PFS and OS comparisons for this genotype.
The current study represents the largest retrospective analysis for evaluating the association between FCGR2A and FCGR3A genotypes and clinical outcome in mCRC patients treated with cetuximab. Validated central lab analysis was used in order to generate robust results. Our study confirms the findings from our previous European Consortium publication18 by showing that KRAS mutations are associated with a lack of response to cetuximab in patients with mCRC refractory to 5-FU, irinotecan and oxaliplatin.
The study results failed to support the primary study hypothesis, and no correlation was seen between mPFS and FCGR2A-HH compared with the others, and FCGR3A-VV compared with the others. No other clinically significant correlation was observed between FCGR2A and FCGR3A SNPs and cetuximab efficacy, thus leading to the conclusion that these polymorphisms cannot be considered as predictive markers of response to cetuximab in chemorefractory mCRC KRAS-WT patients. There was a numerical difference in the mPFS in the FCGR3A-VV group compared to the non-VV group among the KRAS-WT patients suggesting a possible detrimental effect. This, however, reached only a borderline (p=0.06) significance and cannot be considered as clinically relevant.
Cohort studies on breast cancer and lymphoma patients supported the hypothesis that FCGR polymorphisms affect the efficacy of trastuzumab and rituximab, with the most consistent results showing an advantage for FCGR3A-VV compared to VF or FF.11 ,13–15 The mCRC studies that examined the correlation between cetuximab and FCGR status mostly included small patient populations. Zhang et al,19 with a total of 39 mCRC patients analysed, they suggested advantage in RR and PFS for HH and FF versus RR or VV (p=0.003 and p=0.004, respectively). Increasing the cohort to 130 patients who participated in the phase II open-label multicentre clinical trial (IMC 0144), however, failed to replicate those results.26
Graziano et al20 analysed 110 chemorefractory patients treated with irinotecan and cetuximab. Those authors investigated several possible associations between genetic variants and clinical outcomes. Their multivariate analysis revealed associations between OS and EGFR for the intron-1 S/S and EGF 61 G/G genotypes, but no association was detected with FCGR status. Paez et al27analysed a diverse group of 104 mCRC patients of whom 92 were treated with cetuximab and chemotherapy, the majority with irinotecan-based chemotherapy, and six with oxaliplatin regimens. Only 56% of those patients were chemorefractory and had failed at least two previous lines. That study failed to show a significant association between the different FCGR polymorphisms and response, PFS or KRAS mutation status. By contrast, Bibeau et al28 analysed 69 patients and described an advantage in PFS for VV FCGR3A patients compared to F carriers (6.9 vs 3.2 months, p=0.047). Those authors reported the association with FCGR status to be maintained in KRAS-mutant (KRAS–M) patients for whom PFS was 5.5 months for VV patients versus 2.8 months for F carriers, p=0.039, although only 27 KRAS-M patients were analysed. FCGR and KRAS were independent predictors of efficacy in their multivariate analysis.
Rodríguez et al29 recently analysed 106 mCRC patients treated with cetuximab for FCGR status, as well as for the EGFR downstream mutations KRAS, tested in codons 12 and 13, Neuroblastoma RAS Viral (V-Ras) Oncogene Homolog (NRAS), tested in codons 12, 13 and 61, PIK3CA tested in exon 20 and V-Raf Murine Sarcoma Viral Oncogene Homolog B (BRAF) tested in exon 15. As expected, DCR was correlated with KRAS-WT tumours (61% vs 39%, p=0.049). In patients with EGFR downstream mutations, FCGR2A-HH subtype demonstrated a higher DCR than the others (67% vs 33%, p=0.017). This was maintained in the multivariate analysis.
In our current study, an exploratory analysis in patients with KRAS-M demonstrated an advantage for FCGR3A-FF compared to VF and VV in DCR and OS. This finding was inconsistent with the results of RR and with PFS, that showed no advantage for the FF subgroup. As already mentioned, cetuximab has no efficacy in tumours with mutations in KRAS18 since no inhibition of the EGFR pathway is observed because of KRAS constitutive activation. Nevertheless, in those tumours, cetuximab might trigger the activation of the immune system. Ongoing trials that asses activity of cetuximab in KRAS-mutant patients, based on FCGR genotype might help addressing this questions.30
Finally, a first-line setting analysis was performed by Pander et al22 as part of the phase III CAIRO2 study with chemotherapy and bevacizumab alone, or with the addition of cetuximab in mCRC patients. That substudy assessed the association of the EGFR, EGF, CCND1, FCGR2A and FCGR3A polymorphisms with PFS in KRAS-WT mCRC patients treated with cetuximab. The authors analysed 246 patient subsets of KRAS-WT of which 127 patients were treated with cetuximab and 119 were not. The analysis reported an unexpected decreased mPFS in FCGR3A-VV or VF compared with the FF genotype (8.2 vs 12.8 months, respectively; HR-1.57 (95% CI 1.06 to 2.34), p=0.025). We were able to replicate these findings when we performed the same comparison in our chemorefractory cetuximab-treated cohort. The size of our 740-patient KRAS WT cohort is the largest to be reported thus far, and helps in obtaining a more accurate assessment of the incidences of FCGR SNPs in mCRC patients. Those incidences were different from previously reported values,19 ,22 but our study still had sufficient numbers of patients to maintain the power required for the primary analysis.
We conclude that differences in FGCR affinity to cetuximab secondary to naturally occurring genotypes were not related to different outcomes in chemorefractory mCRC patients who were treated with cetuximab. We are unable to explain why these polymorphisms cannot predict cetuximab response, but at least two main hypotheses can be considered. First, the patients included in our study had received more than one line of chemotherapy. Conventional chemotherapy suppresses the activation and the number of NKs and macrophages, thus inhibiting moAb-mediated cell death. Moreover, late stages of disease may also be associated with anergy, that is, the lack of response of the immune system to cancer antigens, independently from the effect of chemotherapy. Second, in metastatic solid tumours, the biodistribution of moAb might be limited due to the tumour burden, the low vascularity and the high interstitial pressure. In this context, the immune system environment may be ineffective due to the anergy and the lack of distribution of the moAb itself. These factors may negatively affect the immune response and underpower the role of polymorphisms. Nevertheless, cetuximab still remains a very effective drug, even in late lines of treatments, thus indicating that mechanisms other than the immune system mediate its effectiveness.
Overall, we can conclude that our negative findings suggest that ADCC does not play a role in cetuximab efficacy in mCRC. However, we cannot exclude the possibility that the immune response may be one of the mechanisms behind cetuximab activity in this setting. Further studies for investigating the impact of innate and adaptive immune response to cetuximab treatment are therefore recommended. Finally, our study supports the growing recognition that the introduction of new biomarkers into clinical practice can only be supported by large, well-powered analyses.
We thank the patients, their families and all the investigators who participated in the study, for their contributions to this report. We thank S Yuki from Hokkaido University Hospital, K Yamaguchi from Saitama Cancer Center, K Muro at Aichi Cancer Center Hospital, Eiji Shinozaki at Cancer Institute Hospital of Japanese Foundation for Cancer Research, and Shinya Kajiura at University of Toyama for their contribution. Esther Eshkol is thanked for editorial assistance.
RG and LV contributed equally.
Contributors RG, LV, HP and ST designed the work, provided and profiled the samples, acquired and analysed the data, interpreted the data, wrote the manuscript. KTK, BVJ, H-JL, TY, DP, CM, JS, FC, AC, MF, GF, JSJ, EVH, WZ, DY, KY, TN, DP, BV, TM, FL, KLGS, PP, FC provided patients’ samples, read and approved the manuscript. JDS profiled the samples.
Funding LV is supported by an ESMO fellowship for Translational Research. TY is supported by the Grant-in-Aid for Cancer Research (21 S4-5) from the Ministry of Health, Labour and Welfare of Japan. FC is sponsored by a grant from Associazione italiana per la ricerca sul cancro (AIRC) and from Ministero dell'Istruzione, dell'Università della Ricerca (MIUR) and progetti di ricerca di interesse nazionale (PRIN). DP is supported by the Instituto de Salud Carlos III (FIS/1101711). CMV is supported by PI12/00989 and AECC 11964. ST is a Senior Clinical Investigator of the Fund for Scientific Research Flanders (FWO-Vlaanderen) and has received research grants from the Belgian Federation against Cancer (Stichting tegen Kanker) and from the Belgian National Cancer Plan (Nationaal Kankerplan).
Competing interests All the conflicts of interest reported are outside of this work. AC has been giving lectures and attending advisory boards for Merck Serono and Roche. FC has been attending advisory boards for Merck Serono, Bayer, Roche and has been receiving research grants from Astra Zeneca, Merck Serono and Bayer. KY has been giving seminar presentations for Merck Serono and Bristol Meyer Squibb. H-JL has been attending advisory boards for Merck Serono and Bristol Meyer Squibb and has been receiving honoraria from Merck Serono. FL has been attending advisory boards for Amgen, Sanofi Aventis, has been giving educational presentations for Roche and Amgen and lectures for Sanofi Aventis, Roche, Bayer and has been receiving research grants from Roche and Merck Serono. TN has been receiving honoraria from Merck Serono. TY has been receiving honoraria from Chugai, Takedo, Merck Serono and research fundings from Daiichi, Sankyo, Taiko, Bayer, Eli Lilly, Pfizer, Chugai, Yokult. ST has been giving lectures and attending advisory boards for Merck Serono.
Ethics approval local ethics review boards.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.