Melanoma incidence is rapidly increasing, particularly in young adults. While targeted therapies like BRAF V600E inhibitors (vemurafenib, dabrafenib) and MEK inhibitors (cobimetinib, trametinib) revolutionized treatment, acquired resistance remains a significant challenge. Most patients eventually relapse. Current approaches focusing on signature biomarkers or pathways might overlook tumor-specific alterations. This study proposes a patient-specific approach based on alterations in signaling protein networks. The researchers hypothesize that analyzing the unique signaling network structure of each tumor, using an information-theoretic approach, will identify the specific elements that should be targeted for optimal therapeutic effect. This approach, previously shown effective in breast cancer, aims to identify these patient-specific altered signaling signatures (PaSS) in melanoma to improve upon existing BRAF inhibitor-based combination treatments. The study examined 53 BRAF V600E and V600K mutant melanoma (SKCM) samples, supplemented with 372 thyroid carcinoma (THCA) samples (due to frequent BRAF V600E mutation in THCA, allowing study of commonalities and differences in altered signaling signatures between the two tumor types). The researchers anticipated that understanding the PaSS would help in designing more effective drug combinations to address the frequent drug resistance encountered in BRAF-mutant melanoma.

The introduction extensively reviews the rising incidence of melanoma, the development of targeted therapies focusing on the BRAF V600E mutation, and the limitations of current treatments due to the frequent development of drug resistance. It cites several studies on BRAF inhibitor resistance mechanisms and combination therapies, highlighting the limitations of approaches solely reliant on genomic or protein biomarkers. Existing research demonstrating the effectiveness of patient-specific altered signaling signature (PaSS) analysis in other cancers is also mentioned, setting the stage for applying this approach to melanoma.

The study employed a computational-experimental approach. First, signaling pathways associated with the BRAF gene were curated. A dataset of 353 SKCM and 372 THCA samples was analyzed using reverse-phase protein arrays (RPPA) for the expression of 216 proteins. Survival analysis, an information-theoretic method, was used to identify patient-specific altered signaling signatures (PaSS). This analysis identified 17 distinct unbalanced processes that repeat throughout the samples. Each tumor was characterized by a unique subset of these unbalanced processes (typically 1-3), represented as a patient-specific barcode. Based on the PaSS, personalized drug combinations were predicted, often employing FDA-approved drugs targeting key proteins in each active unbalanced process. The efficacy of the predicted combinations was then validated in vitro using BRAF-mutated melanoma cell lines (A375, G361, A2058) and in vivo using murine models. In vitro studies assessed cell viability using Methylene blue and MTT assays, and western blotting was used to examine the effect of the drug combinations on various signaling proteins. In vivo studies involved subcutaneous injection of melanoma cells into NSG mice followed by treatment with predicted drug combinations, with tumor growth monitored over time.

Analysis revealed 17 distinct unbalanced processes sufficient to describe the 725 SKCM and THCA tumors. Unbalanced processes 1 and 2 distinguished well between THCA and SKCM tumors. Each tumor exhibited a unique combination of 1-3 unbalanced processes, creating patient-specific barcodes. These barcodes were largely mutually exclusive, suggesting a high degree of individualized tumor heterogeneity. The PaSS did not always correlate with BRAF V600E mutation status, indicating that BRAF status alone may be insufficient for effective treatment. In vitro studies demonstrated that PaSS-based drug combinations were significantly more effective at killing melanoma cells compared to clinically used combinations or single agents. These combinations were highly selective: a combination effective for one cell line was often less effective for another. In vivo experiments showed that the PaSS-based drug combinations significantly reduced tumor growth in murine models, outperforming clinically used therapies. The study found that PaSS-based combinations effectively prevented or delayed the development of drug resistance in vitro. High-resolution PaSS analysis allowed for the identification of unique altered signaling signatures, even within cells harboring the same BRAF mutation (e.g., A375 and G361 cells). The in vivo data demonstrated superior efficacy and selectivity of the personalized drug combinations compared to clinically used therapies. However, some tumor growth curves were not completely flattened in vivo, possibly due to undetected subpopulations or the emergence of new unbalanced processes during treatment.

The findings underscore the significance of tumor heterogeneity and the limitations of relying solely on genomic biomarkers like BRAF mutation status for melanoma treatment. The high number of distinct barcodes (138) across the 725 tumors highlight extensive individual variability. The almost mutually exclusive nature of barcodes in SKCM and THCA samples suggests distinct evolutionary paths. The superior efficacy and selectivity of PaSS-based drug combinations in both in vitro and in vivo studies strongly support the approach. The incomplete flattening of tumor growth curves in some in vivo studies suggests the need for a more dynamic treatment strategy, possibly involving serial biopsies to detect and target emerging subpopulations or integrating immunotherapies.

This study demonstrates the potential of PaSS analysis for designing highly effective and selective personalized drug combinations for BRAF V600E melanoma. The approach offers a significant improvement over current therapies by targeting the specific altered signaling network in each tumor. Future research could focus on integrating this approach with other treatments, developing strategies to address emerging subpopulations during treatment, and exploring broader applications in other cancer types.

The study's in vivo experiments did not completely flatten tumor growth curves in all cases, potentially due to the emergence of new subpopulations or unforeseen signaling changes during treatment. The study primarily focused on BRAF-mutated melanomas, and further research is needed to determine the applicability of this approach to other melanoma subtypes or cancers. The relatively small sample size in the in vivo studies warrants further investigation to confirm the findings in larger cohorts.