The Food and Drug Administration (FDA) approved cetuximab in 2004 for treatment of mCRC in combination with irinotecan, as well as in monotherapy in patients intolerant to irinotecan. However, only a minority of patients respond to cetuximab-based therapy, and there are currently no molecular markers able to identify patients who will benefit from this therapeutic approach. A better understanding of molecular mechanisms that may predict resistance or response to cetuximab is therefore urgently needed.
Similarly to previous reports (Cunningham et al, 2004
; Saltz et al, 2004
), we observed no correlation between the intensity of EGFR expression as detected by IHC and response to cetuximab-based therapy. In the present cohort, eusomic patients who did not respond to cetuximab were actually classified either score 1+ (66%) or score 3+ (33%). These results are in agreement with those from two small trials, showing that EGFR evaluation using IHC is misleading in predicting response to the MoAb. Moreover, in a retrospective series of 16 chemo-refractory patients not expressing EGFR, cetuximab was shown to produce an RR of 25% (Chung et al, 2005
). In addition, two PR were seen in nine patients with EGFR-negative tumours enrolled in a phase II study of single agent cetuximab (Lenz et al, 2005
). It has been shown that the choice of fixative and storage time of tumour tissue, (Atkins et al, 2004
) the choice of primary antibody and scoring system (Kersting et al, 2006
), and the lack of standardised criteria for evaluation (Langner et al, 2004
) all represent potential pitfalls and have a substantial impact on determination of EGFR immunoreactivity. It is therefore highly questionable whether mCRC patients should be selected for cetuximab-based therapy only on EGFR reactivity by IHC.
Our results indicate that both high polysomy on chromosome 7 and EGFR
gene amplification appear to be a pre-requisite for response to cetuximab (observed in 25 and 75% of cases, respectively). Supporting this hypothesis, three eusomic patients had no benefit from the MoAb. Our findings are consistent with those reported (Moroni et al, 2005
; Lievre et al, 2005
). Of note, in one of these, patients with trisomy were defined as polysomic and benefited from the drug in 89% of cases (Moroni et al, 2005
). In our series, however, 12 cases with high polysomy on chromosome 7 and two cases with EGFR
gene amplification did not benefit from cetuximab. Consequently, the evaluation of the EGFR
gene status appears to be insufficient to predict response to the MoAb.
In the present cohort, the majority (90%) of patients with mutated K-Ras
did not benefit from cetuximab. Similar results have been reported in two additional trials, whereas in another study a not significant trend was observed. If we pool our data with those of these three studies (Moroni et al, 2005
; Lievre et al
, 2006; Di Fiore et al, 2007
), the difference between K-Ras
mutated and K-Ras
wild-type sequence patients as regards to cetuximab treatment is highly significant, indicating that the assessment of K-Ras
mutations in mCRC plays a fundamental role in predicting cetuximab efficacy. As regards to the type of K-Ras
mutation, the mostly observed alterations occurred on codon 12 (70%), against 30% on codon 13. These findings are consistent with previous reports (Frattini et al, 2004
). Interestingly, besides the expected occurrence of GaT
mutation, no alterations of GtT
(the other most frequently observed K-Ras
mutation in sporadic CRC) (Frattini et al, 2004
) were found on codon 12. The present data reinforce, therefore, the knowledge that the GtT
mutation typically correlates with an indolent clinical course, and seldom occurs in mCRC (Sarli et al, 2004
Other factors besides the EGFR
gene status and K-Ras
mutations are likely to be involved in mechanisms of resistance to cetuximab. The EGFR signal activation leads not only to downstream effects on Ras-MAP kinase pathway, but also regulates the PTEN-PI3K-Akt cascade. The loss of expression of PTEN protein has been observed in 30% of sporadic CRC (Thomas and Grandis, 2004
). No data on PTEN protein expression and correlation with response to cetuximab in mCRC have been reported yet. We demonstrate that loss of PTEN protein expression may be a useful marker in predicting response to cetuximab. In fact, none of the 11 patients with loss of PTEN expression did benefit from the treatment with MoAb, while a response was observed in 10 out of 16 patients with intact PTEN expression. The six patients with intact PTEN expression who did not benefit from cetuximab had K-Ras
mutation in four cases, eusomy in one case and both in one case: all these factors have been shown to predict resistance to cetuximab.
The effect of PTEN expression on cetuximab response is similar to the one observed in trastuzumab-treated breast cancer patients (Pandolfi, 2004
), supporting the concept that PTEN expression plays a fundamental role in predicting the response to drugs against HER family members.
The population of the present study is comparable with those of previous related studies in terms of included patients, and it is homogeneous since it encompasses unselected patients native from an isolated geographic area, all evaluated and treated in one institution. None of the patients were found to exhibit MSI, which suggests that tumour development in our patients followed the same pathway (Fearon and Vogelstein, 1990
). Actually, the rate of PR (37%) is surprisingly high when compared with published data of pivotal trials including cetuximab (Cunningham et al, 2004
). On the other hand, our cohort is characterised by high frequency of polysomy and EGFR
gene amplification. We can therefore speculate that environmental and lifestyle factors might lead both to frequent cell division deregulation (as deductible by the observed high rate of polysomy) and to EGFR
gene amplification. Although the clinical course might be more aggressive, this particular constellation makes perhaps a targeted approach more effective than expected.
The possibility that the present findings are related to the response to the previous chemotherapy regimen rather than cetuximab sensitivity or resistance may raise questions on the validity of our as well as previous results. This hypothesis, however, is unlikely since all patients included in this study were refractory to previous chemotherapeutic treatment, and drugs, such as fluoropirimidines, oxaliplatin and irinotecan, act against thymidylate synthase and topoisomerase I and not against the EGFR signalling pathway.
Overall, our findings allow to propose an algorithm in order to possibly select patient for cetuximab therapy (). Those presenting with eusomy on chromosome 7 are more likely to be refractory to the MoAb, while patients with high polysomy or EGFR
gene amplification should be considered for cetuximab therapy. A benefit from the MoAb may be expected in patients presenting with wild-type K-Ras
and intact PTEN expression. Only one patient escaped to this algorithm, in that he responded to cetuximab in presence of K-Ras
mutation. A possible explanation could be that in our series all molecular analysis have been performed on primary CRC, and the gene profile on primary tumour and metastasis might differ, as previously reported (Scartozzi et al, 2004
Algorithm in predicting response to cetuximab according to the EGFR and K-Ras status, and PTEN protein expression.
In conclusion, our results indicate that different downstream proteins of the EGFR cascade have a deep effect on response to cetuximab. In particular, this is the first report on the predictive role of the expression of PTEN protein in mCRC. These data, which need to be validated in large prospective clinical trials, might represent a valid platform for oncologists in selecting patients for cetuximab-based therapy, with evident clinical and economical consequences.