[Skip to Content]
[Skip to Content Landing]
Figure.
Kaplan-Meier Survival of Patients With Uveal Melanomas
Kaplan-Meier Survival of Patients With Uveal Melanomas

Patients are stratified by largest basal tumor diameter (LBD; <12 and ≥12 mm) and by results of the 15-gene expression profile test melanoma class (class 1 indicates low risk for metastasis; class 2, high risk for metastasis). Results are given for the 339 patients in the primary cohort (A-D) and the 241 patients in the validation cohort (E and F). P values are calculated using the log-rank test. OS indicates overall survival; PFS, progression-free survival.

Table 1.  
Summary of Clinicopathologic and Molecular Features in 339 Patients With Uveal Melanomaa
Summary of Clinicopathologic and Molecular Features in 339 Patients With Uveal Melanomaa
Table 2.  
Cox Proportional Hazards Analysis of Clinical, Pathologic, and Molecular Prognostic Variables
Cox Proportional Hazards Analysis of Clinical, Pathologic, and Molecular Prognostic Variables
1.
Ramaiya  KJ, Harbour  JW.  Current management of uveal melanoma.  Expert Rev Ophthalmol. 2007;2(6):939-946.Google ScholarCrossref
2.
AJCC Ophthalmic Oncology Task Force.  International Validation of the American Joint Committee on Cancer’s 7th Edition Classification of Uveal Melanoma.  JAMA Ophthalmol. 2015;133(4):376-383.PubMedGoogle ScholarCrossref
3.
Onken  MD, Worley  LA, Ehlers  JP, Harbour  JW.  Gene expression profiling in uveal melanoma reveals two molecular classes and predicts metastatic death.  Cancer Res. 2004;64(20):7205-7209.PubMedGoogle ScholarCrossref
4.
Worley  LA, Onken  MD, Person  E,  et al.  Transcriptomic versus chromosomal prognostic markers and clinical outcome in uveal melanoma.  Clin Cancer Res. 2007;13(5):1466-1471.PubMedGoogle ScholarCrossref
5.
Petrausch  U, Martus  P, Tönnies  H,  et al.  Significance of gene expression analysis in uveal melanoma in comparison to standard risk factors for risk assessment of subsequent metastases.  Eye (Lond). 2008;22(8):997-1007.PubMedGoogle ScholarCrossref
6.
van Gils  W, Lodder  EM, Mensink  HW,  et al.  Gene expression profiling in uveal melanoma: two regions on 3p related to prognosis.  Invest Ophthalmol Vis Sci. 2008;49(10):4254-4262.PubMedGoogle ScholarCrossref
7.
Singh  AD, Sisley  K, Xu  Y,  et al.  Reduced expression of autotaxin predicts survival in uveal melanoma.  Br J Ophthalmol. 2007;91(10):1385-1392.PubMedGoogle ScholarCrossref
8.
Onken  MD, Worley  LA, Tuscan  MD, Harbour  JW.  An accurate, clinically feasible multi-gene expression assay for predicting metastasis in uveal melanoma.  J Mol Diagn. 2010;12(4):461-468.PubMedGoogle ScholarCrossref
9.
Onken  MD, Worley  LA, Char  DH,  et al.  Collaborative Ocular Oncology Group report number 1: prospective validation of a multi-gene prognostic assay in uveal melanoma.  Ophthalmology. 2012;119(8):1596-1603.PubMedGoogle ScholarCrossref
10.
Harbour  JW, Chen  R.  The DecisionDx-UM gene expression profile test provides risk stratification and individualized patient care in uveal melanoma [published online April 9, 2013].  PLoS Curr. doi:10.1371/currents.eogt.af8ba80fc776c8f1ce8f5dc485d4a618.PubMedGoogle Scholar
11.
Correa  ZM, Augsburger  JJ.  Independent prognostic significance of gene expression profile class and largest basal diameter of posterior uveal melanomas.  Am J Ophthalmol. 2016;162:20-27.e2.PubMedGoogle ScholarCrossref
12.
McLean  MJ, Foster  WD, Zimmerman  LE.  Prognostic factors in small malignant melanomas of choroid and ciliary body.  Arch Ophthalmol. 1977;95(1):48-58.PubMedGoogle ScholarCrossref
13.
Shammas  HF, Blodi  FC.  Prognostic factors in choroidal and ciliary body melanomas.  Arch Ophthalmol. 1977;95(1):63-69.PubMedGoogle ScholarCrossref
14.
Augsburger  JJ, Gamel  JW.  Clinical prognostic factors in patients with posterior uveal malignant melanoma.  Cancer. 1990;66(7):1596-1600.PubMedGoogle ScholarCrossref
15.
Kujala  E, Mäkitie  T, Kivelä  T.  Very long-term prognosis of patients with malignant uveal melanoma.  Invest Ophthalmol Vis Sci. 2003;44(11):4651-4659.PubMedGoogle ScholarCrossref
16.
Cisneros  LH, Newman  TJ.  Quantifying metastatic inefficiency: rare genotypes versus rare dynamics.  Phys Biol. 2014;11(4):046003.PubMedGoogle ScholarCrossref
17.
Vanharanta  S, Massagué  J.  Origins of metastatic traits.  Cancer Cell. 2013;24(4):410-421.PubMedGoogle ScholarCrossref
18.
Augsburger  JJ, Gamel  JW, Bailey  RS  Jr, Donoso  LA, Gonder  JR, Shields  JA.  Accuracy of clinical estimates of tumor dimensions: a clinical-pathologic correlation study of posterior uveal melanomas.  Retina. 1985;5(1):26-29.PubMedGoogle ScholarCrossref
Original Investigation
July 2016

Prognostic Implications of Tumor Diameter in Association With Gene Expression Profile for Uveal Melanoma

Author Affiliations
  • 1Ocular Oncology Service, Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida
  • 2Sylvester Comprehensive Cancer Center, University of Miami Miller School of Medicine, Miami, Florida
  • 3Tumori Foundation, California Pacific Medical Center, San Francisco
 

Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.

JAMA Ophthalmol. 2016;134(7):734-740. doi:10.1001/jamaophthalmol.2016.0913
Abstract

Importance  Uveal melanoma (UM) can be divided into prognostically significant subgroups based on a prospectively validated and widely used 15-gene expression profile (GEP) test. Class 1 UMs have a low risk and class 2 UMs have a high risk for metastasis.

Objective  To determine whether any clinicopathologic factors provide independent prognostic information that may enhance the accuracy of the GEP classification.

Design, Setting, and Participants  This retrospective observational study performed at 2 ocular oncology referral centers included 339 patients in a primary cohort and 241 patients in a validation cohort. Both cohorts had a diagnosis of UM arising from the ciliary body and/or choroid. All patients underwent tumor biopsy for GEP prognostic testing. Clinicopathologic variables included patient age and sex, tumor thickness, largest basal tumor diameter (LBD), ciliary body involvement, and pathologic cell type. Patients from the primary cohort were enrolled from November 1, 1998, to March 16, 2012; from the validation cohort, from November 4, 1996, to November 7, 2013. Follow-up for the primary cohort was completed on August 18, 2013; for the validation cohort, December 10, 2013. Data were analyzed from November 12, 2013, to November 25, 2015.

Main Outcome and Measures  Progression-free survival (PFS). The secondary outcome was overall survival.

Results  The primary cohort included 339 patients (175 women [51.6%]; mean [SD] age, 61.8 [13.6] years). The most significant prognostic factor was GEP classification (exp[b], 10.33; 95% CI, 4.30-24.84; P < .001). The only other variable that provided independent prognostic information was LBD (exp[b], 1.13; 95% CI, 1.02-1.26; P = .02). Among class 2 UMs, LBD showed a modest but significant association with PFS (exp[b], 1.13; 95% CI, 1.04-1.24; P = .005). The 5-year actuarial metastasis-free survival estimates (SE) were 97% (3%) for class 1 UMs with LBD of less than 12 mm, 90% (4%) for class 1 UMs with LBD of at least 12 mm, 90% (9%) for class 2 UMs with LBD of less than 12 mm, and 30% (7%) for class 2 UMs with LBDs of at least 12 mm. The independent prognostic value of LBD and the 12-mm LBD cutoff were corroborated in the independent validation 241-patient cohort.

Conclusions and Relevance  Class 2 UMs had better prognosis when the LBD was less than 12 mm at the time of treatment. These findings could have important implications for patient counseling, primary tumor treatment, clinical trial enrollment, metastatic surveillance, and adjuvant therapy.

Introduction

Uveal melanoma (UM) is the most common primary intraocular cancer and frequently gives rise to metastasis, especially to the liver.1 Numerous clinical and pathologic features in UM have been linked to metastatic risk, including patient age, largest basal tumor diameter (LBD), tumor thickness, ciliary body involvement, epithelioid tumor cell morphology, and extraocular tumor extension.2

With the use of gene expression profiling (GEP), primary UMs can be classified into 1 of 2 prognostically significant molecular subgroups. Class 1 UMs have a low risk and class 2 UMs have a high risk for metastasis.3 Previous investigations4-7 have shown that GEP has greater prognostic accuracy than clinical, pathologic, and chromosomal features in UM. Hence, we developed a quantitative polymerase chain reaction–based 12-gene expression array performed on a microfluidics platform, which was validated in a National Cancer Center–funded prospective, multicenter clinical trial conducted by the Collaborative Ocular Oncology Group (COOG).8,9

In the initial COOG report, metastasis was detected in only 3 of 276 class 1 UMs (1.1%) compared with 44 of 170 class 2 UMs (25.9%) (log-rank test, P < 10−14).9 No clinicopathologic feature or chromosome 3 status provided prognostic information that was independent of GEP. In the present study, we analyzed our single-center experience after longer follow-up and inclusion of more patients with small UMs to reinvestigate whether any clinicopathologic factor may provide prognostic information that is independent of GEP. Our findings were confirmed in an independent patient cohort from another ocular oncology referral center.

Box Section Ref ID

Key Points

  • Question Do any clinicopathologic factors provide prognostic information that is independent of the gene expression profile test?

  • Findings This cohort validation study showed class 2 uveal melanomas had better prognosis when their largest basal diameter was less than 12 mm at the time of treatment.

  • Meaning Gene expression profile classification remains the strongest indicator of metastatic risk in uveal melanoma, but a smaller tumor diameter provides additional prognostic information that could have important implications for patient counseling, primary tumor treatment, clinical trial enrollment, metastatic surveillance, and adjuvant therapy.

Methods
Clinical Data Collection

Clinical data were collected from the ocular oncology centers directed by two of us (D.H.C. and J.W.H.). The primary cohort was treated by one of us (J.W.H.) at Washington University in St. Louis from November 1, 1998, to March 16, 2012, and the validation cohort was treated by the other (D.H.C.) at the Tumori Foundation at California Pacific Medical Center from November 4, 1996, to November 7, 2013. The study included patients with primary UMs arising in the choroid and/or ciliary body. Patients with purely iris melanomas were excluded. This study was approved by the institutional review boards of the University of Miami School of Medicine, Washington University in St Louis, and California Pacific Medical Center. Written informed consent was obtained from all patients. All patient records were accessed in a Health Insurance Portability and Accountability Act–compliant fashion in accordance with the Declaration of Helsinki.

The following clinical data were recorded: patient age at diagnosis, sex, pretreatment LBD and tumor thickness, ciliary body involvement, histopathologic cell type (spindle, epithelioid, or mixed), primary treatment modality, date of earliest detected metastasis, date and cause of death, and date of the last follow-up. We measured LBD using ultrasonography and indirect ophthalmoscopy, and the larger value was used. Molecular prognostic class (class 1 or class 2) was determined using a 12-gene expression profile measured by real-time polymerase chain reaction on a microfluidics platform as previously described.10 When this test was migrated from our research laboratory to a Clinical Laboratory Improvement Amendments–certified clinical facility (Castle Biosciences), extensive validation was performed to confirm that the test results were comparable between the 2 laboratories.

Statistical Analysis

Final follow-up for the primary cohort was completed on August 18, 2013 and for the validation cohort on December 10, 2013. Data were analyzed from November 12, 2013 to November 25, 2015. We performed Cox proportional hazards regression to estimate the effects of clinicopathologic covariates on survival. The primary outcome measure was progression-free survival (PFS), defined as the interval from UM diagnosis to the detection of metastatic disease. Data were censored at last follow-up. The secondary outcome measure was overall survival, defined as the interval from UM diagnosis to death due to any cause. Survival functions were illustrated with nonparametric Kaplan-Meier survival curves. A stepwise log-rank test was used to identify the LBD threshold that best separated classes 1 and 2 cases based on PFS. Statistical analysis was performed using MedCalc (version 15.8; MedCalc Software) and the SPSS statistical package (version 22; IBM Software). P < .05 was considered significant.

Results

The primary cohort consisted of 339 patients (175 women [51.6%]; 164 men [48.4%]; mean age, 61.8 years) diagnosed as having UM arising in the ciliary body and/or choroid, 132 of whom were included in the initial COOG study.9 Clinical, pathologic, and molecular features are summarized in Table 1. The trend toward performing molecular prognostic biopsy on smaller tumors over time by the treating ocular oncologist (J.W.H.) is illustrated in the eFigure in the Supplement. The GEP prognostic test results included class 1 in 190 cases (56.0%) and class 2 in 149 cases (44.0%). Primary treatment was iodine 125–labeled plaque radiotherapy in 251 cases (74.0%), enucleation in 87 cases (25.7%), and observation in 1 elderly patient (0.3%). After a median follow-up of 24.5 (mean, 30.8; interquartile range, 12.0-42.8) months, metastasis was detected in 70 cases (20.6%), including 11 class 1 (5.8%) and 59 class 2 (39.6%) cases. Melanoma-specific mortality occurred in 51 cases (15.0%), including 7 class 1 (3.7%) and 44 class 2 (29.5%) cases. Death due to any cause occurred in 57 cases (16.8%), including 9 class 1 (4.7%) and 48 class 2 (32.2%) cases.

Independent Prognostic Factors Associated With Survival

First, we assessed the prognostic contribution of each clinical, pathologic, and molecular feature to PFS using multivariate Cox proportional hazards analysis in the primary cohort (Table 2). The most significant prognostic factor was the GEP class (exp[b] = 10.33; 95% CI, 4.30-24.84; P < .001). The only other variable that provided independent prognostic information was LBD (exp[b] = 1.13; 95% CI, 1.02-1.26; P = .02). With the use of all-cause mortality as the end point, GEP class was the only significant prognostic factor (exp[b] = 7.99; 95% CI, 3.29-19.40; P < .001).

To evaluate the independent prognostic value of LBD within each GEP class, we performed univariate Cox proportional hazards analysis with PFS as the end point (Table 2). Among class 1 UMs, the association of LBD with PFS was exp(b) = 1.16 (95% CI, 0.99-1.37; P = .07). Among class 2 UMs, LBD showed a modest but significant association with PFS (exp[b] = 1.13; 95% CI, 1.04-1.24; P = .005).

We then used stepwise log-rank testing to determine whether a threshold LBD could be identified that best separated UMs of each GEP class into groups at lower and higher risk for metastasis. For class 1 UMs, no LBD threshold provided a significant separation of tumors with respect to metastatic risk. However, 9 of 11 class 1 UMs (82%) that metastasized had an LBD of at least 12 mm (Figure, A). For class 2 UMs, a significant difference in metastatic risk was observed when cases were separated based on LBD of less than 12 mm vs at least 12 mm. The mean PFS was 68.9 (95% CI, 59.3-78.4) months for class 2 UMs with an LBD of less than 12 mm vs 42.1 (95% CI, 36.4-47.8) months for class 2 UMs with an LBD of at least 12 mm (log-rank test, P = .04) (Figure, B). The 5-year actuarial PFS estimates (SE) were 97% (3%) for class 1 UMs with an LBD of less than 12 mm, 90% (4%) for class 1 UMs with an LBD of at least 12 mm, 90% (9%) for class 2 UMs with an LBD of less than 12 mm, and 30% (7%) for class 2 UMs with an LBD of at least 12 mm. Similar results were obtained for all-cause mortality (Figure, C and D), where the 5-year actuarial overall survival estimates (SE) were 96% (4%) for class 1 UMs with an LBD of less than 12 mm, 91% (4%) for class 1 UMs with an LBD of at least 12 mm, 100% for class 2 UMs with an LBD of less than 12 mm, and 26% (7%) for class 2 UMs with an LBD of at least 12 mm.

Application of the 2-Term Model to an Independent Data Set

To determine whether this 2-term predictive model consisting of GEP class plus LBD could be applied to other patients with UM, we analyzed an independent validation cohort of patients who were treated by one of us (D.H.C.) at the Tumori Foundation. This cohort consisted of 241 patients diagnosed with UM arising in the ciliary body and/or choroid, 132 of whom were included in the initial COOG report.9 This cohort did not differ significantly from the primary cohort with respect to patient age, sex, tumor thickness, ciliary body involvement, or pathologic cell type (eTable in the Supplement). However, the median LBD in the primary cohort was 14.6 (mean, 14.6; interquartile range, 12.0-17.0) mm compared with 11.5 (mean, 11.5; interquartile range, 9.0-13.5) mm for the validation cohort (Mann-Whitney test, P < .001).

The GEP was class 1 in 148 cases (61.4%) and class 2 in 93 cases (38.6%). Primary treatment was proton beam radiotherapy in 212 cases (88.0%), eye wall resection in 11 cases (4.6%), laser hyperthermia in 9 cases (3.7%), and enucleation in 9 cases (3.7%). Median follow-up was 18.9 (mean, 23.7; interquartile range, 9.3-35.6) months. Metastasis was detected in 33 cases (13.7%), including 4 class 1 (2.7%) and 29 class 2 (31.2%) cases. Melanoma-specific mortality occurred in 17 cases (7.1%), including 1 class 1 (0.7%) and 16 class 2 (17.2%) cases. Death due to any cause occurred in 24 cases (10.0%), including 4 class 1 (2.7%) and 20 class 2 (21.5%) cases. As with the primary cohort, GEP classification was the factor most strongly associated with PFS (exp[b], 8.25; 95% CI, 3.79-17.94; P < .001), and LBD provided independent but modest prognostic information (exp[b], 1.19; 95% CI, 1.05-1.34; P = .005). The most significant LBD partition within each GEP class with respect to metastatic risk was LBD of less than 12 mm vs at least 12 mm. The 5-year actuarial PFS (SE) was 100% for class 1 UMs with an LBD of less than 12 mm vs 74% (14%) for class 1 UMs with an LBD of at least 12 mm (log-rank test, P = .07) (Figure, E). The 5-year PFS (SE) was 69% (14%) for class 2 UMs with an LBD of less than 12 mm vs 20% (9%) for class 2 UMs with an LBD of at least 12 mm (log-rank test, P = .004) (Figure, F).

Discussion

In the initial prospective multicenter COOG validation study,9 no clinicopathologic feature was found to provide prognostic information that was independent of the GEP classification. In the present study, we reinvestigated whether any clinicopathologic feature may have independent prognostic value in a cohort treated by a single surgeon (J.W.H.) that included smaller tumors and longer follow-up times than were contributed by the same surgeon to the original COOG study. We confirmed that GEP class was by far the most accurate prognostic feature and that patient age, ciliary body involvement, tumor thickness, and tumor cell type provided no prognostic information that was independent of GEP class. However, we found that in class 2 UMs, LBD provided modest but significant prognostic information that was independent of GEP class and that the optimal threshold between lower and higher metastatic risk was an LBD of approximately 12 mm. A statistically significant association between LBD and outcome was not observed for class 1 UMs. These findings were confirmed in an independent validation cohort, and similar findings were recently reported by another independent group.11

For many years, greater LBD has been known to be a poor prognostic factor in UM, with various explanations being offered.12-14 A common explanation has been the concept of lead-time bias, which postulates that the lower metastatic rate of small UMs would approach that of large UMs with longer follow-up. However, very long follow-up studies have shown that large UMs have a higher metastatic rate after less than 5 years than do small UMs after more than 30 years.15 Thus, although lead-time bias undoubtedly accounts for some differences between small and large UMs, it is unlikely to fully explain our findings. Another explanation is that larger UMs have undergone more cell divisions than smaller UMs, increasing their chances of acquiring mutations that allow them to evade immune surveillance and achieve successful metastasis. This hypothesis is consistent with evidence from other cancer types, which suggests that metastasis is not a discrete event but, rather, the consequence of a vast number of stochastic and highly inefficient events that occur over time.16,17

These findings could have several important implications. The better prognosis for small class 2 UMs suggests that earlier treatment might potentially be associated with improved survival, but because most small uveal melanocytic tumors are class 1 UMs or benign nevi, additional research is needed to develop methods that are less invasive than ocular biopsy to identify the occasional small class 2 UM that might benefit from prompt treatment. Further, these findings need to be taken into consideration when designing clinical trials for adjuvant therapy in high-risk patients. For example, class 2 UMs with an LBD of less than 12 mm and those with larger LBDs perhaps should not be included in the same trials, given their significantly different metastatic risk. We are planning a prospective, multicenter study to validate these findings and to determine the optimal use of LBD in guiding primary tumor treatment, clinical trial inclusion criteria, and systemic adjuvant therapy.

Many patients in the validation cohort were treated with different modalities than patients in the primary cohort (ie, proton beam radiotherapy and eye wall resection). Therefore, our results may be applicable to patients regardless of treatment. Another strength of this study was that all patients in each cohort were treated by a single ocular oncologist (D.H.C. or J.W.H.), thereby minimizing interobserver variability in clinical assessment and management.18 Weaknesses of the study included the retrospective study design, which likely led to small differences in clinical tumor measurements, metastatic surveillance, follow-up intervals, and other factors, as well as the relatively short follow-up, which could have preferentially underestimated the rate of metastasis in class 1 tumors. Because PFS and overall survival were almost identical, we did not need to undertake a competing risks analysis. Chromosome 3 status was not collected because a multicenter prospective study showed that these data did not provide prognostic value that was superior to or independent of GEP class.9

Conclusions

We found that the LBD provided modest but significant prognostic information among class 2 UMs. Those UMs with an LBD of less than 12 mm had a significantly better prognosis than those with an LBD of at least 12 mm. The results of our retrospective analysis were verified in an independent patient cohort and corroborate the recently published findings of another independent group.11 These findings could have important implications for patient counseling, early treatment of the primary tumor, and adjuvant therapy for high-risk patients.

Back to top
Article Information

Corresponding Author: J. William Harbour, MD, Ocular Oncology Service, Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, 900 NW 17th St, Miami, FL 33136 (harbour@miami.edu).

Submitted for Publication: December 3, 2015; final revision received February 27, 2016; accepted March 13, 2016.

Published Online: April 28, 2016. doi:10.1001/jamaophthalmol.2016.0913.

Author Contributions: Mr Feuer and Dr Harbour had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Harbour.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Walter, Feuer, Char, Harbour.

Critical revision of the manuscript for important intellectual content: Walter, Chao, Feuer, Schiffman, Harbour.

Statistical analysis: Feuer, Schiffman, Harbour.

Obtained funding: Walter, Harbour.

Administrative, technical, or material support: Chao, Harbour.

Study supervision: Harbour.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Harbour reports being the inventor of intellectual property used in the study and receiving royalties from its commercialization and working as a paid consultant for Castle Biosciences, the licensee of intellectual property presented in this article. No other disclosures were reported.

Funding/Support: This study was supported by grant R01 CA125970 from the National Cancer Institute (Dr Harbour); a Senior Scientific Investigator Award from Research to Prevent Blindness, Inc (Dr Harbour); the Melanoma Research Alliance and Melanoma Research Foundation (Dr Harbour); a Paul Kayser Global Award from the Retina Research Foundation (Dr Harbour); a gift from the Mark J. Daily, MD, fund (Dr Harbour); a Heed Fellowship Award from the Heed Ophthalmic Foundation (Dr Walter); core grant P30EY014801 from the National Institutes of Health (Bascom Palmer Eye Institute); an unrestricted grant from Research to Prevent Blindness, Inc; and grant W81XWH-09-1-0675 from the US Department of Defense.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: Christina L. Decatur, BS, Ocular Oncology Service, Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, provided research coordinator support, for which she received compensation.

References
1.
Ramaiya  KJ, Harbour  JW.  Current management of uveal melanoma.  Expert Rev Ophthalmol. 2007;2(6):939-946.Google ScholarCrossref
2.
AJCC Ophthalmic Oncology Task Force.  International Validation of the American Joint Committee on Cancer’s 7th Edition Classification of Uveal Melanoma.  JAMA Ophthalmol. 2015;133(4):376-383.PubMedGoogle ScholarCrossref
3.
Onken  MD, Worley  LA, Ehlers  JP, Harbour  JW.  Gene expression profiling in uveal melanoma reveals two molecular classes and predicts metastatic death.  Cancer Res. 2004;64(20):7205-7209.PubMedGoogle ScholarCrossref
4.
Worley  LA, Onken  MD, Person  E,  et al.  Transcriptomic versus chromosomal prognostic markers and clinical outcome in uveal melanoma.  Clin Cancer Res. 2007;13(5):1466-1471.PubMedGoogle ScholarCrossref
5.
Petrausch  U, Martus  P, Tönnies  H,  et al.  Significance of gene expression analysis in uveal melanoma in comparison to standard risk factors for risk assessment of subsequent metastases.  Eye (Lond). 2008;22(8):997-1007.PubMedGoogle ScholarCrossref
6.
van Gils  W, Lodder  EM, Mensink  HW,  et al.  Gene expression profiling in uveal melanoma: two regions on 3p related to prognosis.  Invest Ophthalmol Vis Sci. 2008;49(10):4254-4262.PubMedGoogle ScholarCrossref
7.
Singh  AD, Sisley  K, Xu  Y,  et al.  Reduced expression of autotaxin predicts survival in uveal melanoma.  Br J Ophthalmol. 2007;91(10):1385-1392.PubMedGoogle ScholarCrossref
8.
Onken  MD, Worley  LA, Tuscan  MD, Harbour  JW.  An accurate, clinically feasible multi-gene expression assay for predicting metastasis in uveal melanoma.  J Mol Diagn. 2010;12(4):461-468.PubMedGoogle ScholarCrossref
9.
Onken  MD, Worley  LA, Char  DH,  et al.  Collaborative Ocular Oncology Group report number 1: prospective validation of a multi-gene prognostic assay in uveal melanoma.  Ophthalmology. 2012;119(8):1596-1603.PubMedGoogle ScholarCrossref
10.
Harbour  JW, Chen  R.  The DecisionDx-UM gene expression profile test provides risk stratification and individualized patient care in uveal melanoma [published online April 9, 2013].  PLoS Curr. doi:10.1371/currents.eogt.af8ba80fc776c8f1ce8f5dc485d4a618.PubMedGoogle Scholar
11.
Correa  ZM, Augsburger  JJ.  Independent prognostic significance of gene expression profile class and largest basal diameter of posterior uveal melanomas.  Am J Ophthalmol. 2016;162:20-27.e2.PubMedGoogle ScholarCrossref
12.
McLean  MJ, Foster  WD, Zimmerman  LE.  Prognostic factors in small malignant melanomas of choroid and ciliary body.  Arch Ophthalmol. 1977;95(1):48-58.PubMedGoogle ScholarCrossref
13.
Shammas  HF, Blodi  FC.  Prognostic factors in choroidal and ciliary body melanomas.  Arch Ophthalmol. 1977;95(1):63-69.PubMedGoogle ScholarCrossref
14.
Augsburger  JJ, Gamel  JW.  Clinical prognostic factors in patients with posterior uveal malignant melanoma.  Cancer. 1990;66(7):1596-1600.PubMedGoogle ScholarCrossref
15.
Kujala  E, Mäkitie  T, Kivelä  T.  Very long-term prognosis of patients with malignant uveal melanoma.  Invest Ophthalmol Vis Sci. 2003;44(11):4651-4659.PubMedGoogle ScholarCrossref
16.
Cisneros  LH, Newman  TJ.  Quantifying metastatic inefficiency: rare genotypes versus rare dynamics.  Phys Biol. 2014;11(4):046003.PubMedGoogle ScholarCrossref
17.
Vanharanta  S, Massagué  J.  Origins of metastatic traits.  Cancer Cell. 2013;24(4):410-421.PubMedGoogle ScholarCrossref
18.
Augsburger  JJ, Gamel  JW, Bailey  RS  Jr, Donoso  LA, Gonder  JR, Shields  JA.  Accuracy of clinical estimates of tumor dimensions: a clinical-pathologic correlation study of posterior uveal melanomas.  Retina. 1985;5(1):26-29.PubMedGoogle ScholarCrossref
×