Intratumoral Molecular Heterogeneity in a BRAF-Mutant, BRAF Inhibitor-Resistant Melanoma: A Case Illustrating the Challenges for Personalized Medicine

Results and Discussion

We sought to determine whether the patient's initial vemurafenib resistance developed through the reactivation of the MAPK pathway due to additional mutations to the NRAS or BRAF genes. Sanger sequencing was conducted on the patient's vemurafenib-resistant metastasis to detect somatic mutations in exon 1 and exon 2 of the NRAS gene and in exon 11 and exon 15 of the BRAF gene. This revealed a G13R NRAS mutation in addition to the original BRAF V600E mutation within the single vemurafenib-resistant metastasis (Fig. 1B–D). We used immunocytochemistry to analyze the expression levels of downstream signaling proteins within the progression biopsy to identify any subclones of tumor cells within the lesion. Immunohistochemistry was conducted on the progressed metastasis using antibodies against Ki-67, p-ERK1/2, p-AKT, cyclin D1, MITF, p27^Kip1, p16^INK4a, p53, BCL-2, Mcl-1, PTEN, MAP3K8 (COT), IGF1Rβ, and PDGFRβ. The staining for p-ERK1/2 expression clearly identified 2 areas of tumor with differential staining (Fig. 2A). The percentage of immunopositive tumor cells was assessed for each subclone using a Dako ACIS III Automated Image Analysis System. One subclone had high p-ERK1/2 expression (95% tumor cells positive, labeled subclone A; Fig. 2D), and was adjacent to melanoma cells with low expression of p-ERK1/2 (3% tumor cells positive, labeled subclone B; Fig. 2E). The proliferative rate represented by Ki-67 expression was higher in the subclone A (10% tumor cells positive) than in subclone B (3% tumor cells positive; Fig. 2F and G). Conversely, the percentage of cyclin D-positive tumor cells was lower in subclone A than in subclone B (46% and 74%, respectively; Fig. 2H and I). The expression of the remaining proteins did not seem to differ between the subclones.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

A, expression of p-ERK1/2, which is high in subclone A and relatively negative in subclone B (scale bar, 1 mm). B and C, allele-specific RT-PCR showing the presence of NRAS wild-type in both subclones A and B, but NRAS G13R in only subclone B. D, high magnification image of p-ERK1/2 expression in subclone A (scale bar, 100 μm). E, high magnification image of p-ERK1/2 expression in subclone B (scale bar, 100 μm). F, Ki-67 expression in subclone A (scale bar, 100 μm). G, Ki-67 expression in subclone B (scale bar, 100 μm). H, cyclin D1 expression in subclone A (scale bar, 100 μm). I, cyclin D1 expression in subclone B (scale bar, 100 μm).

We then carried out mutation testing separately on each of the subclones A and B, which were identified by their p-ERK1/2 expression. Each subclone was macrodissected and genomic DNA was extracted from each individual subclone. BRAF and NRAS mutation status of the subclones was determined by allele-specific specific real-time PCR (RT-PCR). In addition, subclone genomic DNA was analyzed by Mass Spectrometric SNP genotyping with OncoCarta v0.1 and the MelaCarta panel v0.1 to detect any additional somatic mutations. The genomic DNA mutational status of subclones A and B was different; subclone A was NRAS wild-type and BRAFV600E mutant (Fig. 2B), whereas subclone B was NRAS G13R mutant and BRAFV600E mutant (Fig. 2C). The sensitivity of this allele-specific RT-PCR allowed the detection of as little as 2% mutant NRAS and BRAF allele (data not shown; ref. 10). For this reason, it is unlikely that the differential mutational status of the subclones is due to the assay sensitivity. The OncoCarta and MelaCarta did not reveal any additional mutations.

Mutational analysis using mutant-specific RT-PCR of the antecedent primary melanoma and lymph node metastases that occurred 5 and 3 years before the diagnosis of widely disseminated metastatic disease, revealed the presence of V600E BRAF mutation, but not the NRAS G13R mutation (Fig. 1A). Similarly, analysis of 7 metastatic tumors resected after cessation of treatment with vemurafenib and during treatment with single-agent trametinib or the combination of trametinib and dabrafenib detected a BRAF V600E mutation without a NRAS G13R mutation in all biopsies (Fig. 1A).

Here, we show intratumoral molecular heterogeneity in a progressing V600E BRAF-mutant melanoma metastasis from a patient treated for 7 months with the BRAF inhibitor vemurafenib. Within the same individual progressing and vemurafenib-resistant metastasis, 2 distinct subpopulation were observed, both V600E BRAF-mutant, and only one G13R NRAS-mutant. The acquisition of an additional NRAS mutation causing MAPK pathway reactivation has been proposed as a mechanism of BRAF inhibitor resistance in BRAF mutant melanoma patients (5). In this study, the NRAS G13R mutation may be one but not the sole driver of disease progression. This issue of tumor heterogeneity is further complicated by incomplete tumor shrinkage, at the macroscopic and radiologic levels, induced by BRAF inhibitors in the majority of tumors and patients. Thus, a biopsy of tumor progression on any BRAF inhibitor may frequently yield a mixture of tumor cells with distinct kinetic growth properties. That an NRAS mutation was not detected in the subsequent tumor biopsies raises tantalizing questions as to how the MEK inhibitor treatment alone (first sequential therapy after failure on vemurafenib alone) might have influenced the selective pressure, for or against, the double BRAF/NRAS mutant melanoma population. Our tissue biopsy series here illustrate the dynamic nature of tumor evolution on multiple distinct targeted regimens.

In this patient, we have shown that intralesional genetic heterogeneity is present within BRAF inhibitor-resistant metastatic melanoma and highlighted the fact that sampling error is a potential pitfall of fine-needle aspiration or punch biopsy techniques. Advances in genome analysis methods such as deep sequencing (11–13) and sequencing of circulating tumor cells (14) may provide superior information regarding the heterogeneity of a patient's tumors. These methods remain exploratory due in part to the complex analyses required. Knowledge of heterogeneity may help predict and improve a patient's response to combination therapies based on the level and signature of mutational heterogeneity. These technologies may also help confront another likely issue arising from the tumor heterogeneity: multiple mechanisms of acquired drug resistance in the same progressing tumor or multiple progressing tumors in the same patient.

However, the current method of assessing heterogeneity in a subset of sampled lesions is unlikely to adequately predict tumor heterogeneity in vivo, nor reflect the ongoing genetic changes that occur during treatment (13, 15). Thus, the strategy of biopsying metastatic disease to decide the next systemic therapy after progression on a targeted therapy will be complicated by heterogeneity. We need to anticipate multiple mechanisms of resistance, predict the critical downstream effector pathways, and treat with the most effective broad-ranging combination of therapies at first diagnosis of metastatic disease. These therapies may include the targeting of oncogenes along with immunotherapy or chemotherapy (16–18).