Open AccessA Prognostic Gene Signature Expressed in Primary Cutaneous Melanoma: Synergism With Conventional Staging
Author(s) -
Georg Brunner,
Achim Heinecke,
Thomas Falk,
Beyhan Ertas,
Norbert BlödornSchlicht,
HansJoachim Schulze,
L. Suter,
Jens Atzpodien,
Carola Berking
Publication year - 2018
Publication title -
jnci cancer spectrum
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.345
H-Index - 10
ISSN - 2515-5091
DOI - 10.1093/jncics/pky032
Subject(s) - medicine , melanoma , receiver operating characteristic , oncology , cohort , cancer staging , gene signature , stage (stratigraphy) , ajcc staging system , logistic regression , framingham risk score , survival analysis , cancer , pathology , gene expression , cancer research , gene , staging system , biology , biochemistry , disease , paleontology
Background Current clinico-pathological American Joint Committee on Cancer (AJCC) staging of primary cutaneous melanoma is limited in its ability to determine clinical outcome, and complementary biomarkers are not available for routine prognostic assessment. We therefore adapted a gene signature, previously identified in fresh-frozen (FF) melanomas and adjacent stroma, to formalin-fixed paraffin-embedded (FFPE) melanomas. The aim was to develop a gene expression profiling (GEP) score to define patient survival probability at the time of first diagnosis. Methods Expression of 11 FF melanoma signature genes was quantified by reverse transcription polymerase chain reaction in an FFPE melanoma training cohort (n = 125), corresponding to the combined FF melanoma training and validation cohorts. The resulting GEP score was validated technically and clinically in an independent FFPE melanoma cohort (n = 211). All statistical tests were two-sided. Results We identified a prognostic eight-gene signature in the tumor area (tumor and adjacent tissue) of AJCC stage I–III melanomas. A signature-based GEP score correlated with melanoma-specific survival (MSS; Kaplan-Meier analysis: P < .0001) was independent of tumor stage (multivariable regression analysis: P = .0032) and stroma content (<5%–90%) and complemented conventional AJCC staging (receiver operating characteristic curve analysis: area under the curve = 0.91). In the clinical validation cohort, the GEP score remained statistically significant ( P = .0131) in a multivariable analysis accounting for conventional staging. The GEP score was technically robust (reproducibility: 93%; n = 84) and clinically useful, as a binary as well as a continuous score, in predicting stage-specific patient MSS. Conclusions The GEP score is a clinically significant prognostic tool, contributes additional information regarding the MSS of melanoma patients, and complements conventional staging.