Mechanisms of BRAFi-induced hyperproliferative cutaneous conditions.

It is well recognized that cutaneous conditions such as verrucal keratosis (VK) and squamous cell carcinoma (cuSCC) are one of the most frequent and significant adverse events associated with the BRAF inhibitors (BRAFi), vemurafenib and dabrafenib. These agents are first line in the treatment of V600 mutant metastatic melanoma and have been shown to induce hyperproliferation of wild-type BRAF keratinocytes. 1-3. The mechanism by which BRAFi induce hyperproliferation in keratinocytes has been postulated to be a result of paradoxical activation of the mitogen-activated protein kinase (MAPK) pathway in wtBRAF cells. Heidorn et al.2 hypothesized that this was a result of the BRAFi forming a dimer with wtBRAF which then acts as a scaffold to activate CRAF, thereby activating MEK and ERK 2. In this study, we aimed to analyse the molecular interplay involved in the formation of BRAFi-induced hyperproliferative keratinocytic lesions, cuSCC and VK, seen in patients taking BRAFi for stage IV metastatic melanoma. cuSCC and VK samples were collected from patients treated with BRAFi. Selected samples were previously tested for mutation using Oncocarta v1.0 panel (Sequenom™, Brisbane Australia). After histological diagnosis, tissue samples underwent immunohistochemistry analysis with antibodies directed against components of the MAPK pathway (p-BRAF, p-CRAF, p-MEK, p-ERK), Pi3K-AKT pathway (p-AKT3, Pi3K), cell cycle control molecule cyclin D, markers of cellular proliferation (Ki67, keratin 6), keratin 10 and oncogenic protein (p53). Normal epidermis from the biopsy was used as control. Antibodies were purchased from Abcam, Spring or Cell Marquee. p-BRAF antibody was abandoned due to inability to optimize staining. Slide staining was performed using the IPX platform (Dako Autostainer Plus). Automated steps using Novocastra products were applied in sequence, protein blocking (RE7158), primary antibody application, secondary layer (RE7260-K), DAB visualization (RE7163) and a haematoxylin counterstain. High-resolution scanned images were obtained using Leica SCN 400 Client and the expression of markers (presence, intensity and location of stain) assessed by two reviewers. Diseased skin was compared to normal skin found at the periphery of the tissue sample. Intensity was rated on a scale of 0–3 at five separate locations (basal layer, suprabasal layer, stratum spinosum, stratum granulosum and infiltrate). Scores from the two reviewers were combined and recoded in three categories: none, mild and strong stain. All results were recorded in Filemaker pro 11.0v4. Chi-square statistical analysis was performed using JMP 7.0. Twenty tissue specimens were collected from 14 patients treated with the BRAFi dabrafenib, either alone (n = 13) or in combination with the MEK inhibitor (MEKi) trametinib (n = 1) (Table S1) for stage IV metastatic melanoma. This included 10 cuSCC and 10 verrucal keratosis. Of these, 18 underwent genomic analysis using Oncocarta v1.0 panel. The results of this analysis have been published previously 4. All tissue samples underwent immunohistochemistry analysis for 10 separate antibodies. When comparing normal skin (tumor-surrounding skin) with VK and cuSCC (Fig. 1), p-AKT was downregulated and cyclin D, p-ERK and keratin 6 were upregulated in VK and cuSCC. Ki67 was upregulated in the basal and supra basal layer of cuSCC only. Ki67 (Figure S1) and keratin 6 (Figure S2) are both markers of cell proliferation. When comparing cuSCC to VK samples, p-ERK (a component of the MAPK pathway which is induced during cell proliferation) was upregulated in the supra basal and spinosum layer of cuSCCs compared to VK (Figure S3). Ki67 also was upregulated in cuSCC compared to VK, whereas p-AKT was downregulated in the cuSCC tissue compared to VK. Our study (Fig. 1) suggests a transition from normal skin to VK and cuSCC. Although the differences between cuSCC and normal skin are quite mark, VK shares some levels of protein expression with normal skin (Ki67 or p-AKT) and some with cuSCC (K6 or cyclin D). There was no notable difference in phosphorylated Pi3K expression between diseased skin (cuSCC and VK) and normal skin, but there was a decrease in p-AKT stain in the progression from normal skin to VK to cuSCC. It may be assumed that signalling has been diverted from the Pi3K-AKT pathway into the MAPK pathway. This could also be explained due to crosstalk between the two pathways5. Phosphorylated RAF (p-CRAF), MEK (p-MEK) and ERK (p-ERK) were all tested for, and we found that p-ERK had a higher expression in the supra basal and spinosum layers of the diseased skin in comparison with normal skin. This suggests that in our samples, the MAPK pathway is hyperactivated. When comparing samples proven to harbour RAS mutations via our genomic analysis, to those without, we found that samples with RAS mutation had a higher expression of p-ERK in their upper layers, while keratin 10 expression was lower in these samples (Fig. 2). This is consistent with hyperactivation of the MAPK pathway in RAS-mutated keratinocytes. In summary, our results indicated that the MAPK pathway is upregulated in patients treated with a BRAFi. These agents induced both cuSCC and VK through increased cellular proliferation at all layers of the epidermis. The immunohistological differences between normal skin, verrucal keratosis and cuSCC are suggestive of a progression from normal epidermis to verrucal keratosis to cuSCC. Authors would like to thank the Oncology clinical trials team at Westmead Hospital, Dr Firoz Iqbal (Pathology West), Dr Heather Medbury (Vascular Biology Research Centre), Cheryl Hart and Neil Catlett (Skin and Cancer Foundation Australia). MA acquired and interpreted the data and wrote the manuscript; RA designed the research study, scored the IHC and drafted the manuscript. GC assisted with data acquisition; FS scored the IHC and drafted the manuscript; PFP designed the research study, analysed the data and wrote the article. The authors declare no conflict of interests. Figure S1. Ki-67 expression in (a) normal skin, (b) VK and (c) cuSCC. Figure S2. Keratin 6 expression in (a) normal skin, (b) VK and (c) cuSCC. Figure S3. p-ERK expression of (a) VK Vs (b) cuSCC. Table S1. Patient demographics, drug type, histopathology diagnosis and mutation status for each sample tested. Table S2. Summarises statistically significant changes in expression between BRAFi induced disease skin (cuSCC and VK) compared to normal skin. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.