1116P - Uveal melanoma tumor size correlated with gene expression profiling: the MD Anderson Experience

Date 28 September 2014
Event ESMO 2014
Session Poster Display session
Topics Skin Cancers
Pathology/Molecular Biology
Translational Research
Basic Scientific Principles
Basic Principles in the Management and Treatment (of cancer)
Presenter Sapna Patel
Citation Annals of Oncology (2014) 25 (suppl_4): iv374-iv393. 10.1093/annonc/mdu344
Authors S. Patel1, E. Reinhardt2, P. Fox3, D. Gombos4
  • 1Melanoma Medical Oncology, MD Anderson Cancer Center, 77030 - Houston/US
  • 2Melanoma Medical Oncology, MD Anderson Cancer Center, 77030-4009 - Houston/US
  • 3Biostatistics, MD Anderson Cancer Center, 77030-4009 - Houston/US
  • 4Head And Neck Surgery, MD Anderson Cancer Center, 77030-4009 - Houston/US



Uveal melanoma is the most common eye cancer in adults affecting approximately 6,000 patients worldwide each year. Classically, prognostication for this disease involved histopathologic parameters such as tumor size, thickness, histologic cell type, mitoses, or extraocular extension. Recently, gene expression profiling (GEP) has armed clinicians with a powerful tool for risk assessment that more accurately stratifies patients according to metastatic risk than previous histopathologic categories. This test, commercially available in the U.S. as DecisionDx-UM, stratifies patients into low-risk for distant metastasis (DM), Class 1A and 1B, and high-risk for DM, Class 2. The relationship between tumor size and GEP remains unclear. Here we aim to analyze and clarify that relationship.


Between December 2009 and April 2014, consecutive uveal melanoma patients at MD Anderson Cancer Center who received prognostication with DecisionDx-UM were captured. Data regarding original tumor size was obtained from echography measurements or pathology specimens, where available. We correlated tumor size and GEP in this dataset.


A total of 115 uveal melanoma patients with GEP and size measurements were included for analysis. A Kruskal-Wallis test and ordinal logistic regression were conducted to assess the relationship between largest tumor dimension (basal and apical) and GEP.

Largest tumor dimension was not significantly different by GEP class, nor was it a significant predictor of GEP class (p = 0.6390). Class 1 tumors had similar sizes compared to Class 2 tumors.

GEP and Largest Tumor Dimension

GEP Largest dimension (mm) p-value
n Median Minimum Maximum
Class 1A 31 13.00 5.90 25.00 0.6390
Class 1B 29 12.00 5.50 22.00
Class 2 55 13.00 5.20 22.00


In our single institution case series, uveal melanoma tumor size did not significantly correlate with GEP. Therefore size alone does not predict a patient's more powerful prognosticator, GEP. Clinicians should not rely on tumor size to counsel patients regarding metastatic risk.


All authors have declared no conflicts of interest.