Cetuximab has shown efficacy in patients with metastatic colorectal cancer in several phase II trials leading, in 2004, to FDA approval for the treatment of irinotecan-refractory metastatic colorectal cancer. Several retrospective and prospective studies have clearly demonstrated that KRAS mutation confers resistance to these patients [6–10, 23, 24] but the complete mechanism of cetuximab sensitivity remains only partially understood. The present study was conducted in patients receiving cetuximab before KRas-mutation testing was introduced as a requirement. As expected, no response rate was observed in mutated KRAS patients vs 15.2% in wt tumors, even though the difference did not reach significance. We presently analyzed 8 gene polymorphisms involving 5 relevant candidate genes potentially related to the pharmacodynamics of cetuximab, namely EGFR, EGF, CCND1, FCGR2A and FCGR3A, on 58 CRC patients receiving cetuximab-based therapy. Statistical analyses were conducted in the whole set of patients, as well as in the sub-group of 34 patients with wt KRas tumors, so as to reflect the current cetuximab-treated population.
Numerous studies have reported a relationship between favorable outcome of cetuximab-treated patients and related skin toxicity [2, 3, 25]. Accordingly, present data show a higher response rate in patients developing grade 2-3 cutaneous toxicity as compared to patients with grade 0-1 (14% vs 5%, respectively), even though not significant. Present results also show a tendency for an association between intron 1 EGFR polymorphism and cetuximab-related skin toxicity: the incidence of grade 2-3 toxicity was 1.5-fold greater in patients bearing short CA-repeats in intron 1 of EGFR gene (CA sum ≤ 35) as compared to others (p = 0.058, Figure 1). This observation concords well with previous studies by Amador et al.  and Graziani et al.  reporting that patients developing cutaneous rash after anti-EGFR therapies presented shorter CA-repeats in intron 1 of EGFR gene as compared to patients who did not develop rash. Experimental studies have reported an inverse correlation between the number of CA-repeats in the intron 1 of the EGFR gene and EGFR gene transcription [28–30]. It can thus be hypothesized that elevated ubiquitous EGFR expression (including skin and tumor) renders the cells more susceptible to anti-EGFR effects.
In addition to the influence of intron 1 polymorphism on EGFR gene transcription, EGFR gene presents two functional polymorphisms in the promoter region: the -216G/T polymorphism located in a Sp1 binding site [31, 32], and the -191C/A polymorphism located 4 bp upstream of a transcription initiation site . These two SNPs may thus have an impact on EGFR gene regulation. Present data obtained on patients with wt KRas tumors show a significantly longer TTP in homozygous EGFR -191CC patients relative to other patients (p = 0.030, univariate analysis). However, this genotype was not retained in a multivariate analysis.
Cyclin D1 is a downstream effector of EGFR signaling that regulates cell cycle. The CCND1 A870G gene polymorphism affects the splice donor site at the exon 4/intron 4 boundary, resulting in two different mRNA transcripts (a and b) . Both the A allele and the G allele can encode these two transcripts. However, the A allele preferentially encodes transcript b, which results in a longer half-life cyclin D1 protein . The impact of CCND1 A870G polymorphism on cancer progression has been studied in head and neck cancer patients, with conflicting results [34, 35]. In our study, patients homozygous for the CCDN1 870AA genotype had a significantly greater response rate than AG or GG patients, both in the whole population and in patients with a wt KRas tumor (75.0% vs 7.1%, respectively in wt KRas patients). In addition, patients with the CCND1 870AA genotype had a significantly longer median TTP than GG patients, with AG patients having an intermediary TTP, both in the whole population and in patients with a wt KRas tumor (median TTP were 7.9, 3.0 and 2.6 months, in AA, AG and GG wt KRas patients, respectively). Of note, in patients with wt KRas tumors, CCND1 polymorphism also influenced specific survival, with a significantly shorter survival in GG patients. The positive influences of CCND1 870A allele are thus consistent with one another, even though they do not concord with the sole published study having analyzed the impact of CCND1 A870G polymorphism on the outcome of advanced colorectal cancer patients receiving cetuximab therapy . In this latter study, conducted on a limited sample of 39 patients, the 870 G allele had a favorable impact on survival .
In addition to direct anti-EGFR effect, IgG1 mAbs such as cetuximab mediate anti-tumor effects by the ADCC mechanism. Fragment C of the mAb binds to the Fc receptors (FcR) carried by immune cells, thus triggering tumor cell lysis. Functional polymorphisms on two FcR genes (FCGR2A, FCGR3A) affecting the affinity of FcR for fragment C have been identified [37, 38]. These polymorphisms may thus influence ADCC efficiency [39, 40]. Even though some studies have reported significant associations between these polymorphisms and clinical efficacy of rituximab , trastuzumab  or cetuximab [43, 44], data conflict regarding which alleles are linked to favorable patient outcome. In the present study, we report a significant influence of FCGR3A F158V polymorphism on survival both in the whole population and in patients with a wt KRas tumor, with VV patients presenting a dramatically shorter survival. The favorable influence of the FCGR3A 158F allele was also reported in a study by Zhang et al.  and a study by Pander et al. , but not in the study from Bibeau et al. . These discrepancies related to the impact of FCGR3A 158F/V polymorphism on cetuximab efficacy are difficult to account for but could be due to the relatively limited sample size of these studies.
We observed that none of the 13 patients pre-treated with bevacizumab had a response to cetuximab and that this subgroup had a significantly decreased specific survival as compared with non-pretreated patients (9.8 months vs 4.9 months, p = 0.018). This difference remains statistically significant in a multivariate analysis adjusted for age, sex, PS status and KRas status (data not shown). This negative influence of bevacizumab pre-treatment cannot be imputed to patient characteristics which were not significantly different between bevacizumab pretreated patients and non-pretreated patients, although it must be noted that 42% of bevacizumab pretreated patients carried KRAS mutated tumors vs 29% in non-pretreated patients (p = 0.48). Importantly, a multivariate analysis including bevacizumab pretreatment revealed that CCND1 A870G and FCGR3A F158V polymorphisms both remained significant independent predictors of patient survival (whole population and KRas wt tumors).
The retrospective design of this study, conducted on a relatively small number of patients, may place intrinsic limitations on the present original data. However, results obtained in the sub-group of wt KRas patients, corresponding to the current cetuximab-treated population, clearly suggest that CCND1 A870G polymorphism may be used as an additional marker for predicting cetuximab efficacy, TTP and overall survival. Of note, FCGR3A F158V polymorphism and CCND1 A870G polymorphism were significant independent predictors of overall survival in patients with wt KRas tumors. Such promising observations deserve further confirmation in a prospective study conducted on a larger population of CRC patients receiving cetuximab-based therapy.