Machine learning models demonstrate that clinicopathologic variables are comparable to gene expression prognostic signature in predicting survival in uveal melanoma

Published:September 03, 2022DOI:


      • Molecular assays are not accessible to all uveal melanoma patients.
      • We investigate machine learning models on clinicopathologic variables for risk stratification.
      • Machine learning models included random survival forest and survival gradient boosting.
      • They performed similarly or better than gene expression prognostic signature.
      • Readily accessible clinicopathologic variables can provide adequate prognostic information.



      Since molecular assays are not accessible to all uveal melanoma patients, we aim to identify cost-effective prognostic tool in risk stratification using machine learning models based on routine histologic and clinical variables.

      Experimental design

      We identified important prognostic parameters in a discovery cohort of 164 enucleated primary uveal melanomas from 164 patients without prior therapies. We then validated the prognostic prediction of top important parameters identified in the discovery cohort using 80 uveal melanomas from the Tumor Cancer Genome Atlas database with available gene expression prognostic signature (GEPS). The performance of three different survival analysis models (Cox proportional hazards (CPH), random survival forest (RSF), and survival gradient boosting (SGB)) was compared against GEPS using receiver operating curves (ROC).


      In all three selection methods, BAP1 status, nucleoli size, age, mitotic rate per 1 mm2, and ciliary body infiltration were identified as significant overall survival (OS) predictors; and BAP1 status, nucleoli size, largest basal tumor diameter, tumor-infiltrating lymphocyte density, and tumor-associated macrophage density were identified as significant progression-free survival (PFS) predictors. ROC plots for the median survival time point showed that significant parameters in SGB studied model can predict OS better than GEPS. For PFS, SGB model performed similarly to GEPS. The time-dependent area under the curve (AUC) showed SGB model performing better than GEPS in predicting OS and metastatic risk.


      Our study shows that routine histologic and clinical variables are adequate for patient risk stratification in comparison with not readily accessible GEPS.

      Graphical abstract


      To read this article in full you will need to make a payment

      Purchase one-time access:

      Academic & Personal: 24 hour online accessCorporate R&D Professionals: 24 hour online access
      One-time access price info
      • For academic or personal research use, select 'Academic and Personal'
      • For corporate R&D use, select 'Corporate R&D Professionals'


      Subscribe to European Journal of Cancer
      Already a print subscriber? Claim online access
      Already an online subscriber? Sign in
      Institutional Access: Sign in to ScienceDirect


        • Jager M.J.
        • Shields C.L.
        • Cebulla C.M.
        • Abdel-Rahman M.H.
        • Grossniklaus H.E.
        • Stern M.H.
        • et al.
        Uveal melanoma.
        Nat Rev Dis Prim. 2022; 8: 4
        • Kivelä T.T.
        • Finger P.T.
        Prognostic implications of the largest basal tumor diameter versus the TNM staging system in association with the gene expression profile for uveal melanoma.
        JAMA Ophthalmol. 2017; 135: 175
        • Arnljots T.S.
        • Al-Sharbaty Z.
        • Lardner E.
        • All-Eriksson C.
        • Seregard S.
        • Stålhammar G.
        Tumour thickness, diameter, area or volume? The prognostic significance of conventional versus digital image analysis-based size estimation methods in uveal melanoma.
        Acta Ophthalmol. 2018; 96: 510-518
        • Kujala E.
        • Mäkitie T.
        • Kivelä T.
        Very long-term prognosis of patients with malignant uveal melanoma.
        Invest Ophthalmol Vis Sci. 2003; 44: 4651-4659
        • Figueiredo C.R.
        • Kalirai H.
        • Sacco J.J.
        • Azevedo R.A.
        • Duckworth A.
        • Slupsky J.R.
        • et al.
        Loss of BAP1 expression is associated with an immunosuppressive microenvironment in uveal melanoma, with implications for immunotherapy development.
        J Pathol. 2020; 250: 420-439
        • Tschentscher F.
        • Husing J.
        • Holter T.
        • Kruse E.
        • Dresen I.G.
        • Jockel K.H.
        • et al.
        Tumor classification based on gene expression profiling shows that uveal melanomas with and without monosomy 3 represent two distinct entities.
        Cancer Res. 2003; 63: 2578-2584
        • Zuidervaart W.
        • van der Velden P.A.
        • Hurks M.H.
        • van Nieuwpoort F.A.
        • Out-Luiting C.J.J.
        • Singh A.D.
        • et al.
        Gene expression profiling identifies tumour markers potentially playing a role in uveal melanoma development.
        Br J Cancer. 2003; 89: 1914-1919
        • Onken M.D.
        • Worley L.A.
        • Char D.H.
        • Augsburger J.J.
        • Correa Z.M.
        • Nudleman E.
        • et al.
        Collaborative Ocular Oncology Group report number 1: prospective validation of a multi-gene prognostic assay in uveal melanoma.
        Ophthalmology. 2012; 119: 1596-1603
        • Cox D.R.
        Regression models and life-tables.
        J R Stat Soc Ser B (Methodol). 1972; 34: 187-202
        • Wang P.
        • Li Y.
        • Reddy C.K.
        Machine learning for survival analysis: a survey.
        ACM Comput Surv. 2019; 51: 1-36
        • Grossniklaus H.E.
        • Finger P.T.
        • Habour J.W.
        • Kivela T.
        Protocol for the examination of specimens from patients with uveal melanoma.
        2017 (
        • Bankhead P.
        • Loughrey M.B.
        • Fernandez J.A.
        • Dombrowski Y.
        • McArt D.G.
        • Dunne P.D.
        • et al.
        QuPath: open source software for digital pathology image analysis.
        Sci Rep. 2017; 716878
        • Viros A.
        • Fridlyand J.
        • Bauer J.
        • Lasithiotakis K.
        • Garbe C.
        • Pinkel D.
        • Bastian B.C.
        Improving melanoma classification by integration genetic and morphologic features.
        PLoS Med. 2008; 5e120
        • Robertson A.G.
        • Shih J.
        • Yau C.
        • Gibb E.A.
        • Oba J.
        • Mungall K.L.
        • et al.
        Integrative analysis identifies four molecular and clinical subsets in uveal melanoma.
        Cancer Cell. 2017; 32 (e15. 204-220
        • R Core Team
        R: a language and environment for statistical computing.
        R Foundation for Statistical Computing, Vienna, Austria2021
        Date accessed: February 26, 2022
        • Van Rossum G.
        • Drake F.L.
        Python 3 reference manual.
        CA: CreateSpace, Scotts Valley2009
        • Friedman J.H.
        Stochastic gradient boosting.
        Comput Stat Data Anal. 2002; 38: 367-378
        • Ishwaran H.
        • Gerds T.A.
        • Kogalur U.B.
        • Moore R.D.
        • Gange S.J.
        • Lau B.M.
        Random survival forests for competing risks.
        Biostatistics. 2014; 15: 757-773
        • Huberty C.J.
        Problems with stepwise methods-better alternatives.
        Adv Social Sci Methodology. 1989; 1: 43-70
        • Simon N.
        • Friedman J.
        • Hastie T.
        • Tibshirani R.
        Regularization paths for Cox's proportional hazards model via coordinate descent.
        J Stat Software. 2011; 39: 1
        • Kursa M.B.
        • Rudnicki W.R.
        Feature selection with the Boruta package.
        J Stat Software. 2010; 36
        • Lundberg S.M.
        • Lee S.I.
        A unified approach to interpreting model predictions.
        Adv Neural Inf Process Syst. 2017; 30
        • Pölsterl S.
        SCIKIT-survival: a library for time-to-event analysis built on top of SCIKIT-learn.
        J Mach Learn Res. 2020; 21: 1-6
        • Berus T.
        • Markiewicz A.
        • Biecek P.
        • Orlowska-Heitzman J.
        • Halon A.
        • Romanowska-Dixon B.
        • Donizy P.
        Clinical significance of nucleoli cytomorphology assessment in patients with uveal melanoma.
        Anticancer Res. 2020; 40: 3503-3512
        • Scholes A.G.M.
        • Damato B.E.
        • Nunn J.
        • Hiscott
        • Grierson I.
        • Field J.K.
        Monosomy 3 in uveal melanoma: correlation with clinical and histologic predictors of survival.
        Invest Ophthalmol Vis Sci. 2003; 44: 1008-1011
        • Koopmans A.
        • Verdijk R.
        • Brouwer R.W.
        • van den Bosch T.P.
        • van den Berg M.M.
        • Vaarwater J.
        • et al.
        Clinical significance of immunohistochemistry for detection of BAP1 mutations in uveal melanoma.
        Mod Pathol. 2014; 27: 1321-1330
        • van Essen T.H.
        • van Pelt S.I.
        • Versluis M.
        • Bronkhorst I.H.G.
        • van Duinen S.G.
        • Marinkovic M.
        • et al.
        Prognostic parameters in uveal melanoma and their association with BAP1 expression.
        Br J Ophthalmol. 2014; 98: 1738-1743
        • Derenzini M.
        • Trerè D.
        • Pession A.
        • Montanaro L.
        • Sirri V.
        • Ochs R.L.
        Nucleolar function and size in cancer cells.
        Am J Pathol. 1998; 152: 1291-1297
        • Mäkitie T.
        • Summanen P.
        • Tarkkanen A.
        • Kivela T.
        Tumor-infiltrating macrophages (CD68(+) cells) and prognosis in malignant uveal melanoma.
        Invest Ophthalmol Vis Sci. 2001; 42: 1414-1421
        • Bronkhorst I.H.G.
        • Vu T.H.K.
        • Jordanova E.S.
        • Luyten G.P.M.
        • van der Burg S.H.
        • Jager M.J.
        Different subsets of tumor-infiltrating lymphocytes correlate with macrophage influx and monosomy 3 in uveal melanoma.
        Invest Ophthalmol Vis Sci. 2012; 53: 5370-5378
        • Oliva M.
        • Rullan A.J.
        • Piulats J.M.
        Uveal melanoma as a target for immune therapy.
        Ann Transl Med. 2016; 4: 172
        • Bercovici N.
        • Guerin M.V.
        • Trautmann A.
        • Donnadieu E.
        The remarkable plasticity of macrophages: a chance to fight cancer.
        Front Immunol. 2019; 10: 1563
        • Herrspiegel C.
        • See T.R.O.
        • Mendoza P.R.
        • Grossniklaus H.E.
        • Stålhammar G.
        Digital morphometry of tumor nuclei correlates to BAP-1 status, monosomy 3, gene expression class and survival in uveal melanoma.
        Exp Eye Res. 2020; 193107987
        • Chen Y.N.
        • Wang Y.N.
        • Chen M.X.
        • Zhang K.
        • Chen R.T.
        • Fang R.
        • et al.
        Machine learning models for outcome prediction of Chinese uveal melanoma patients: a 15-year follow-up study.
        Cancer Commun (Lond). 2022 Jan 9;
        • Eleuteri A.
        • Damato B.
        • Coupland S.E.
        • Taktak A.
        Enhancing survival prognostication in patients with choroidal melanoma by integrating pathologic clinical and genetic predictors of metastases.
        Int J Biomed Eng Technol. 2012; 8: 18-35
        • Vaquero-Garcia J.
        • Lalonde E.
        • Ewens K.G.
        • Ebrahimzadeh J.
        • Richard-Yutz J.
        • Shields C.L.
        • et al.
        PRiMeUM: a model for predicting risk of metastasis in uveal melanoma.
        Invest Opthalmol Vis Sci. 2017; 58: 4096-4105
        • DeParis S.W.
        • Taktak A.
        • Eleuteri A.
        • Enanoria W.
        • Heimann H.
        • Coupland S.E.
        • Damato B.
        External validation of the Liverpool uveal melanoma prognosticator online.
        Invest Opthalmol Vis Sci. 2016; 57: 6116-6122