It is now well established that the identification of predictive biomarkers of therapeutic efficacy for reliable patient stratification is a prerequisite for economical, ethical and logistic sustainability of targeted therapy approaches in the treatment of cancer. Intrinsic to the concept that each tumour has an independent genetic identity, molecular therapies targeting one specific oncoprotein are likely to be useful only in a fraction of patients. This means that administration of targeted drugs to unselected cohorts will result in costs, no clinical benefit and possibly adverse effects for the majority of the population.

These observations also apply to the case of metastatic colorectal carcinomas (mCRCs). The prototypic and most widely used targeted therapies against advanced colorectal tumours are cetuximab and panitumumab, two monoclonal antibodies that block the epidermal growth factor receptor (EGFR/HER1). Addition of these antibodies to chemotherapy in the first-line treatment of unselected mCRC patients results in a modest improvement in median progression-free survival (less than one month) and response rate (5-8% more than chemotherapy alone). When used as monotherapy, the anti EGFR antibodies are effective in only 10-15% of mCRCs. This is likely due to hyperactivation of parallel or downstream compensatory pathways, which substitute or complement EGFR-driven signals in these tumors.

This notion suggests that a significant improvement of efficacy for mCRC targeted therapies can hinge on efforts committed to the optimization of drug combination protocols and the discovery of associated predictive biomarkers. For example, we (and others) have recently identified HER2 amplification as a predictor of resistance to anti-EGFR antibodies in mCRC, demonstrating that it can also predict response to combinatorial therapies against HER2 and EGFR (Bertotti et al., 2011).

Within this project, we decided to concentrate our attention on a significant fraction (34%) of KRAS wild-type tumors, which respond to cetuximab treatment with disease stabilization but not with tumor regression (Bertotti et al., 2011). The indolent growth of these tumors in the presence of antibodies against EGFR suggests that they are biologically sensitive to EGFR blockade, even if this is not sufficient to induce clinically relevant tumor shrinkage. If this is true, this subgroup of tumors can represent an opportunity for improved efficacy of targeted therapy in mCRC. Indeed, the partial lack of response exhibited by these tumors could have two reasonable explanations:

1) For some reasons EGFR is not effectively and completely inhibited by cetuximab.

2) Some other pathways complement or partially compensate for EGFR-dependent signals’ inhibition.

In both cases, a number of possibilities could be envisioned to improve treatment efficacy, which range from testing alternative strategies to inhibit EGFR (in case of cetuximab inefficacy) to identifying complementary pathways that, once characterized, can be challenged by combinatorial therapeutic intervention.To tackle this issue, we have established a large series of mCRC specimens (∼400, in continuous expansion) directly transplanted into mice and grown as xenografts. This setting is far more physiologically relevant than cell lines and allows for molecular profiling of tumor biomarkers together with real-time animal randomization into prospective ‘trials’ with targeted agents (‘xenopatients’). Proof-of-concept for reliability of this approach showed that samples from patients known to be responsive or resistant to the anti-EGFR antibody cetuximab were also sensitive or resistant in the corresponding xenograft cohorts and this tied in with specific validated predictive biomarkers such as KRAS status. Overall, we were able to challenge with cetuximab more than 200 cases. Based on response criteria loosely inspired to the clinical setting, we classified as SDs approximately 40% of KRAS wild-type tumors, which is concordant to that described in clinical trials. Therefore, we shought to explore whether SDs can be turned into overt responses by the means of combining cetuximab with other drugs targeting the same or complementary pathways. Lessons learned primarily from breast cancer indicate that concomitant inhibition of HER2 and other HER family members - such as HER3 and/or EGFR - proves to be more effective than inhibition of HER2 alone. Based on this knowledge, we focused on the other members of the HER family as potential subsidiaries for EGFR. To challenge this hypothesis, we treated 18 SDs with both lapatinib and cetuximab. This approach couples two potential mechanisms of cooperation between the two drugs: first: lapatinib targets EGFR, thus synergizing with cetuximab for full EGFR inhibition; second, lapatinib also inactivates HER2, hindering the enzymatic potential of the EGFR/HER2 complex and preventing compensatory activation of downstream pathways.

Indeed, results have been encouraging, pinpointing 5 tumors -(27.7% of tested cases) that respond to cetuximab monotherapy with disease control but show evident regression following combined administration of cetuximab and lapatinib. Further experiments showed that the EGFR-specific inhibitor erlotinib was as efficacious as lapatinib in inducing tumor regression when combined with cetuximab, suggesting that the main contribution of lapatinib is to complement cetuximab in inhibiting EGFR, while the role of HER2 kinase activity is likely to be marginal in this setting.

The observation that lapatinib synergizes with cetuximab in many but not all cetuximab-SD cases prompted us to screen for other means of partial desensitization to EGFR-inhibition in colorectal liver metastases. We got some unexpected help from the multidimensional molecular annotations assembled by the TCGA Consortium for 276 CRC cases. The authors reported overexpression of IGF2 (insulin-like growth factor 2) in approximately 20% of CRCs. Intrigued by the entity and frequency of this alteration, we analysed IGF2 gene expression in our dataset. We confirmed the rate of overexpression, but we also observed that IGF2 outliers were particularly enriched among cases that respond to cetuximab with stable disease and, to a lesser extent, in the category of the non-responders. Of note, we observed a similar enrichment for IGF2 overexpressors also in an independent dataset (Khambata-Ford 2007) composed by ‘real’’ patients treated with cetuximab monotherapy. Here again, IGF2 overexpression segregates with stable disease and not with tumor regression.

Remarkably, in agreement with the notion of partial independency from EGFR signaling, none of the 5 IGF2-overespressing SDs that were treated with the cetuximab+lapatinib combination did respond with massive shrinkage to treatment. In contrast, 2 IGF-overexpressing tumors out of 3 tested demonstrated sensitivity to a combination of cetuximab and BMS754897 (a specific inhibitor of the IGF2 receptor, IGF1R), which was ineffective in tumors expressing normal levels of IGF2.

These data pinpoint two previously neglected, but not minor, tumor subpopulations in mCRC that appear to display biological sensitivity, but not over clinical responsiveness, to anti-EGFR antibodies: 1) Up to 15-20% of KRAS wild type tumors could benefit from alternative EGFR-inhibition strategies, including combinations of small molecules and antibodies. Preliminary observations suggest that high autocrine levels of specific EGFR ligands could distinguish this subpopulation.2) Up to 20-25% of KRAS wild type tumors, which overexpress IGF2, could be sensitive to combined inhibition of EGFR and IGF1R.We propose these therapeutic hypotheses for clinical validation as a means to improve the clinical benefit of EGFR inhibition-based targeted therapies in mCRC.

Citation Format: Andrea Bertotti. Rational drug combinations to improve anti-EGFR targeted therapies in metastatic colorectal cancer. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr NG02. doi:10.1158/1538-7445.AM2014-NG02

©2014 American Association for Cancer Research.
2014