Advertisement
Mayo Clinic Proceedings Home

Gene Expression Profiling in Cutaneous Melanoma: Caveats for Clinicians

      To the Editor:
      Despite intensive research, differentiating the lethal cutaneous melanomas from the vast majority of biologically indolent lesions remains challenging. Gene expression profiling (GEP) has recently gained momentum, aiming for more accurate staging, guiding the need for adjuvant therapy, and possibly replacing sentinel lymph node biopsies in the future. Several GEP-based tests have been developed,
      • Gerami P.
      • Cook R.W.
      • Wilkinson J.
      • et al.
      Development of a prognostic genetic signature to predict the metastatic risk associated with cutaneous melanoma.
      • Clarke L.E.
      • Warf M.B.
      • Flake II, D.D.
      • et al.
      Clinical validation of a gene expression signature that differentiates benign nevi from malignant melanoma.
      • Meves A.
      • Nikolova E.
      • Heim J.B.
      • et al.
      Tumor cell adhesion as a risk factor for sentinel lymph node metastasis in primary cutaneous melanoma.
      • Brunner G.
      • Reitz M.
      • Heinecke A.
      • et al.
      A nine-gene signature predicting clinical outcome in cutaneous melanoma.
      some sold commercially as potential tools for clinical decision making.
      • Gerami P.
      • Cook R.W.
      • Wilkinson J.
      • et al.
      Development of a prognostic genetic signature to predict the metastatic risk associated with cutaneous melanoma.
      • Clarke L.E.
      • Warf M.B.
      • Flake II, D.D.
      • et al.
      Clinical validation of a gene expression signature that differentiates benign nevi from malignant melanoma.
      On the basis of our own experience with melanoma GEP,
      • Meves A.
      • Nikolova E.
      • Heim J.B.
      • et al.
      Tumor cell adhesion as a risk factor for sentinel lymph node metastasis in primary cutaneous melanoma.
      there is a clear need to highlight important aspects of GEP-based test development and share our own experience with technical aspects of GEP. Hopefully, this communication will facilitate a critical review and discussion of published data by clinicians and researchers.
      Whereas a developmental biomarker study often uses retrospectively collected samples, it should nevertheless be designed to simulate a prospectively controlled study,
      • Subramanian J.
      • Simon R.
      What should physicians look for in evaluating prognostic gene-expression signatures?.
      ie, the composition of the development cohort should resemble a real-life prospective cohort. For example, if a GEP-based test is intended to identify patients at risk for development of metastasis, it should not be developed from data that includes in situ melanomas. A test to predict metastasis will not be applicable to in situ melanomas because there is no clinical need. Developing tests from sample types that will not be tested in clinically relevant circumstances makes it uncertain that these tests will perform as predicted for future patients at risk. It is important to define the type of melanoma for which a test is developed because melanoma gene expression is not just a function of malignant potential. It is affected by many other factors, including histologic type, anatomic structure, spatial context, patient immune status, and age, among many additional variables. Technical limitations also interfere with GEP. For example, one typically retrieves less genetic material from thin vs thick melanomas. Sample cross-contamination by keratinocytes and other cell types is often greater when processing thin melanomas.
      Problems in GEP-based test development also originate from the inclusion of partial tumor samples. Melanoma is known to have intertumor and intratumor heterogeneity. Clones of cells within a tumor can behave differently, expressing different genetic profiles. Keeping this in mind, a test should be developed from tissue samples that contain the majority of the tumor. Performing tests on partial samples, ie, as obtained by wide reexcision surgical procedures, is inappropriate for multiple reasons. First, an unknown amount of tumor will have been removed by the preceding biopsy, including an unknown number of high-risk tumor cells (Figure). Second, the diagnostic biopsy induces a wound healing reaction that can lead to changes in gene expression akin to what is observed in cancer (“tumors are wounds that do not heal”).
      Figure thumbnail gr1
      FigureFoci of CD61 (β3 integrin)–positive cells (arrow) in diagnostic biopsy samples of a thin (left) and a thick (right) melanoma. CD61 is believed to facilitate cell adhesion, thereby promoting metastasis. MLANA = Melan-A, a melanocyte marker.
      Finally, the RNA that is used for GEP is invariably degraded and fragmented because it is derived from formalin-fixed, paraffin-embedded tissue. Poor-quality RNA may go unrecognized and lead to inaccurate results. We therefore believe that treatment decisions should not be based on GEP alone. Rather, more robust multivariate models should be developed that incorporate molecular data in addition to histopathologic findings, regional lymph node tumor status, and clinical data.

      References

        • Gerami P.
        • Cook R.W.
        • Wilkinson J.
        • et al.
        Development of a prognostic genetic signature to predict the metastatic risk associated with cutaneous melanoma.
        Clin Cancer Res. 2015; 21: 175-183
        • Clarke L.E.
        • Warf M.B.
        • Flake II, D.D.
        • et al.
        Clinical validation of a gene expression signature that differentiates benign nevi from malignant melanoma.
        J Cutan Pathol. 2015; 42: 244-252
        • Meves A.
        • Nikolova E.
        • Heim J.B.
        • et al.
        Tumor cell adhesion as a risk factor for sentinel lymph node metastasis in primary cutaneous melanoma.
        J Clin Oncol. 2015; 33: 2509-2515
        • Brunner G.
        • Reitz M.
        • Heinecke A.
        • et al.
        A nine-gene signature predicting clinical outcome in cutaneous melanoma.
        J Cancer Res Clin Oncol. 2013; 139: 249-258
        • Subramanian J.
        • Simon R.
        What should physicians look for in evaluating prognostic gene-expression signatures?.
        Nat Rev Clin Oncol. 2010; 7: 327-334