Protein and epigenetic biomarkers for variability in epidermal growth factor receptor inhibition in cancer treatment and characterization of the inhibitor-induced skin toxicity in an <em>in vitro</em> model

Abstract

Epidermal growth factor receptor (EGFR) is often over-expressed or over-activated in cells of solid tumors and can be targeted by specific inhibitors (EGFRIs), like erlotinib, gefitinib, cetuximab and panitumumab. Unfortunately, EGFRIs only seem to work in a subset of patients suffering from EGFR-expressing tumors. Efficient biomarkers to reliably select patients who will benefit from an EGFRI therapy and determine their optimal dosage are still needed. One of the most common adverse drug reactions observed during therapy with all approved EGFRIs is the development of a typical skin rash. It has been shown in various studies that occurrence and severity of this EGFRI-induced rash are positively correlated with patient outcome (prolonged overall and progression-free survival). However, the rash usually takes several weeks to manifest and is usually treated with topical or systemic medication. Hence, its severity may be suppressed, rendering it unsuitable as a clinical predictive marker. Therefore, the aim of this study was to identify rapidly determinable predictive biomarkers for the severity of EGFRI-induced skin rash, which would be helpful to allow early preventive treatment of the rash while at the same time still allowing prediction of EGFRI efficacy early on and facilitate optimization of personalized cancer therapy for individual patients. They can also help to prevent exposure of patients to substances which are ineffective but still cause discomforting side effects.<br /> Since previous work revealed evidence for genetic, pharmacokinetic and cytokine markers, the intention of this work was to identify functional biomarkers that are able to measure EGFR inhibition variability in the periphery, especially in blood plasma and skin cells. Therefore, molecules that interact with, regulate or functionally modulate EGFR inhibition were selected and their levels determined in plasma samples from patients treated with an EGFRI by enzyme-linked immunosorbent assay (ELISA) and it was checked for correlations with the development of EGFRI-induced skin toxicity. The EGFRI cetuximab, the EGFR ligand amphiregulin (AREG) and the growth factors hepatocyte growth factor (HGF) and 25-OH-vitamin D were selected as candidates following these criteria.<br /> Since miRNAs have recently been shown to be very important in the response to cancer therapy, investigating the suitability of specific mature or precursor miRNA molecules as functional, epigenetic biomarkers for the development of EGFRI-induced skin toxicity was of special interest. Therefore, next generation sequencing (NGS) was performed in keratinocyte and fibroblast cultures from healthy donors to compare miRNA profiles between cells previously identified as rather erlotinib-sensitive and those identified as rather erlotinib-insensitive. <br /> Results show that plasma levels of the EGFRI cetuximab were not significantly associated with development of EGFRI-induced rash. <br /> While the investigated candidates AREG and 25-OH-vitamin D also failed to show a significant correlation with the rash, the plasma concentration of the MET ligand HGF was significantly inversely correlated with severity of EGFRI-induced skin rash. This inverse correlation was also seen with overall survival (OS) in patients who developed EGFRI-induced rash but not in patients with no rash. HGF seems to be a promising biomarker, whose predictive and/or prognostic value should definitely be validated in a larger and well-controlled patient cohort.<br /> In this study primary human dermal keratinocytes as well as fibroblasts have proven to be suitable as in vitro models for studying EGFRI-induced skin rash and associated potential biomarkers. Comparison of miRNA profiles between erlotinib-sensitive and rather -insensitive cells combined with a literature review to identify the miRNAs with the most supporting data, led to identification of the precursor miRNAs mir-146a, mir-31, mir-221, mir-520e and mir-944 as most promising predictive biomarkers in keratinocytes, with mir-146a, mir-520e and mir-944 having known targets and a plausible suggested mechanism of action and mir-31 and mir-221 having already been shown to be associated with sensitivity to an EGFRI in at least one previously published study. In fibroblasts the precursor miRNA mir-34a was identified as the by far most promising predictive biomarker with MET reported as direct target and an associated plausible mechanism of action and two confirmatory previous studies. Further identified promising miRNAs in fibroblasts were mir-382, mir-494, mir-520e and mir-7-1, also with plausible suggested targets. <br /> Overall, the final conclusion of this study is that it might be possible to develop a kind of "predictive profile" consisting of several different biomarkers which taken together may have a predictive value on individual extent of EGFR inhibition in patients. The results of this study suggest that parameters like the plasma concentration of HGF, the expression of mir-146a and mir-31 in dermal keratinocytes and the expression of mir-34a in dermal fibroblasts might be promising parameters to include in such a predictive profile. However, all suggested parameters need to be further confirmed and validated in clinical settings. Eventually, such a predictive score might be used in the future to predict the requirement for changes in dosage of a specific EGFRI, closer monitoring of therapy efficacy by more frequent tumor imaging or maybe even change in therapy away from using an EGFRI or at least adding an adjuvant drug to the EGFRI for better efficacy.