PerspectiveIn the Journals

Biomarker expression may predict metastasis in uveal melanoma

Show Citation

March 1, 2016

High levels of PRAME mRNA appeared associated with increased metastasis risk among patients with class 1 uveal melanoma, according to study results published in Clinical Cancer Research.

Between 2,000 and 3,000 cases of uveal melanoma are diagnosed in the U.S. each year. Gene expression profiling is used to classify these tumors as class 1 or class 2, with class 1 tumors considered much less likely to metastasize, according to study background.

William J. Harbour

J. William Harbour, MD

“The class 2 tumors have a certain genetic profile that appears to point to metastasis,” J. William Harbour, MD, associate director for basic research and leader of the Eye Cancer Site Disease Group at Sylvester Comprehensive Cancer Center at University of Miami Miller School of Medicine, as well as professor, vice chairman for translational research, and director of the ocular oncology service at Bascom Palmer Eye Institute, told HemOnc Today. “These tumors often have mutations in the tumor-suppressor gene BAP1 and copy losses to chromosome 3. However, with longer follow-up, we found that about 10% of class 1 patients will eventually develop metastatic disease, and these patients did not have the same genetic risk factors strongly associated with metastasis.”

Thus, Harbour and colleagues sought to determine potential biomarkers for the identification of those patients with class 1 tumors at an increased risk for metastasis.

The researchers had access to tumor samples from 389 consecutive patients with uveal melanoma (class 1, n = 216; class 2, n = 173). They performed global gene expression profiling and used single nucleotide polymorphism microarrays to analyze potential biomarkers in these tumors.

Sixty-four class 1 tumors were analyzed for PRAME mRNA expression.

Sixty-three patients with class 2 tumors developed metastatic disease, compared with 12 patients with class 1 tumors (log rank P < .0001).

The liver served as the most frequent metastatic site among patients with class 1 (50%; n = 6) and class 2 (90%; n = 36) tumors (P = .02). Other metastatic sites among class 1 patients included the lungs (n = 4), bone (n = 3) and stomach (n = 2).

The researchers identified PRAME mRNA as a significant predictor of metastatic disease in patients with class 1 tumors (P = .0006).

After a median follow-up of 8.2 months (interquartile range, 3.3-57.1), patients with PRAME-positive class 1 tumors demonstrated a 5-year actuarial metastasis rate of 38%, compared with 0% for patients with PRAME-negative class 1 tumors.

The 5-year metastasis rate observed among all patients with class 2 tumors was 71%.

Still, patients with PRAME-positive class 1 tumors achieved a significantly longer metastasis-free survival than patients with class 2 tumors (median, 88 months vs. 32 months; P = .04).

The researchers validated these findings in three independent datasets, including one dataset that used disomy 3 to identify patients with low-risk uveal melanoma.

Copy number alterations associated with PRAME positivity in class 1 tumors included gains of 1q, 6p, 8q and 9q, as well as the loss of 6q and 11q.

Further, PRAME expression appeared associated with larger tumor diameter (P = .005) and SF3B1 mutations (P = .003) in class 1 patients.

The researchers acknowledged the small number of patients with PRAME-positive class 1 tumors as a study limitation. They further noted that certain patients with PRAME-negative class 1 tumors may have developed metastatic disease after the end of follow-up.

“Metastatic uveal melanoma is notorious for not responding to medications that cutaneous melanoma will often respond to — in particular, the immunotherapies that have revolutionized the treatment of metastatic cutaneous melanoma,” Harbour said. “However, approximately 5% of patients with uveal melanoma will respond to immunotherapy, and it is known from other cancers that PRAME expression often predicts good responses to immunotherapies. Our next step is to determine a correlation between uveal melanomas that are responsive to immunotherapy and PRAME expression, because then we would have not only a prognostic marker but a predictive marker for patients.” – by Cameron Kelsall

For more information:

J. William Harbour, MD, can be reached at

Disclosure: The study was funded in part by the NIH, NCI, Research to Prevent Blindness, Melanoma Research Alliance, Melanoma Research Foundation, 2015 AACR Ocular Melanoma Foundation Fellowship and Department of Defense. Harbour reports inventing intellectual property used in the conduct of this study, as well as royalties associated with its commercialization. He further reports a consultant role with Castle Biosciences, which licensed the intellectual property used in this study. The other researchers report no relevant financial disclosures.

itj+ Perspective

Sapna P. Patel, MD

Sapna P. Patel, MD

Uveal melanoma (UM) is the second most common type of melanoma after cutaneous-origin melanoma. However, molecularly and biologically speaking, these two cancers are innately dissimilar. Gene expression profiling (GEP) can categorize UMs into those with low-metastatic potential (Class 1) and those with high-metastatic potential (Class 2). Class 1 UM tumors are not, however, devoid of metastatic risk, and there has not been a reliable biomarker to predict which of these tumors would ultimately result in metastatic disease.

Field and colleagues suggest a novel biomarker known as PRAME may very well predict those patients whose Class 1 tumors will metastasize. PRAME is the preferentially expressed antigen in melanoma, a cancer-testis antigen, found on chromosome 22. Transcriptomic profiling using mRNA expression of discriminating GEP genes clustered Class 1 tumors into two groups, those with and without metastatic disease. Genome-wide mRNA analysis of a small cohort of Class 1 metastatic disease-positive and Class1 metastatic disease-negative tumors with at least 1 year of follow-up revealed PRAME as the single most overexpressed gene in the metastatic cohort. PRAME expression was then confirmed in Class 1 UMs via quantitative polymerase chain reaction and immunohistochemistry. Another dataset of Class 1 UMs with long-term follow-up (median, 68 months) predicted the 5-year actuarial risk of metastasis at 5% for Class1 PRAME-negative UMs and 31% for Class1 PRAME-positive UMs. 

PRAME correlation with the five currently well-described UM driver mutations (GNAQ, GNA11, BAP1, EIF1AX and SF3B1) is not yet known. In a small cohort, the authors note SF3B1 mutations — known along with EIF1AX mutations to be associated with Class 1 Ums — to be the most common of the “good” mutations found in Class 1 metastatic disease-positive tumors. EIF1AX mutations were absent in Class1 metastatic disease-positive tumors. Additionally, PRAME association with the current GEP subcategories Class 1A and Class 1B is unclear as is the rate of metastatic disease in Disomy3 PRAME-positive UMs due to a small cohort of these latter tumors in the dataset.

The researchers make bold assertions as to clinicopathologic differences between metastasis in Class 1 and Class 2 tumors, most notably the younger age and extrahepatic first site of metastasis more common in Class 1 tumors, as well as the diverse metastasis-free survival (Class 1, 88 vs. Class 2, 32 months). But it is not clear that overt clinicopathologic differences exist between Class 1 metastatic disease-positive and Class 1 metastatic disease-negative tumors. Importantly, there are inconsistent data presented regarding longest basal tumor diameter and association with PRAME expression or Class 1 metastatic disease-positive vs. Class 1 metastatic disease-negative UMs. 

In summary, this work demonstrates increased PRAME expression in Class 1 UMs may predict those tumors with metastatic potential. Taken with GEP, this information arms patients and clinicians with the opportunity for enhanced surveillance measures, particularly in alternate organ sites in addition to the liver.

Sapna P. Patel, MD
The University of Texas MD Anderson Cancer Center

Disclosure: Patel reporters an investigator role on translational research studies investigating uveal melanoma.