Impact of Multigene Panel Testing in High‐Risk Uveal Melanoma Patients

ABSTRACT

Germline genetic testing is increasingly incorporated into the care of patients with uveal melanoma, yet optimal testing strategies remain unclear. This study aimed to evaluate the outcome of germline clinical genetic testing in uveal melanoma (UM) patients who met the National Comprehensive Cancer Network (NCCN) guidelines for genetic testing. A retrospective chart review was conducted on UM patients seen in The Ohio State University Cancer Genetics Clinic between 5/1/2021 and 9/19/2025. Seventy individuals underwent clinical genetic testing, primarily via large multigene panels. Ten UM patients, including two related individuals, had pathogenic or likely pathogenic (P/LP) variants in known cancer genes (BAP1, BRCA1, BRCA2, MBD4, MUTYH, POT1, XRCC2). Among unrelated individuals, the positive rate was 13% (8/69). Excluding carrier genes, the rate was 10.1% (7/69). Eight patients would have been missed if only tested for BAP1 per American Society of Clinical Oncology (ASCO) 2024 recommendations. There was no association between tumor size, stage, and germline P/LP variants. In summary, NCCN guidelines are useful in the prioritization of UM patients for genetic testing. Additionally, these findings support genetic counseling for multigene panel testing in high‐risk uveal melanoma patients. Further studies of the impact of germline variants on UM disease outcome are warranted.

Significance

This is the first study to assess outcomes of large panel testing in high‐risk UM patients. Through comprehensive testing, we identified patients with pathogenic variants to genes that have been previously reported in UM patients and new variants of uncertain significance, highlighting the phenotypic and locus heterogeneity found within patients with personal or family history suggestive of UM‐related tumor predisposition. Our results confirm the importance of large panel cancer genetic testing, rather than small panel or single gene, for UM patients with clinical features indicating genetic predisposition to cancer.

National Comprehensive Cancer Network (NCCN) provides clear criteria for genetic testing among uveal melanoma (UM) patients. However, the efficacy of these guidelines to detect variants in actionable cancer genes has not been evaluated recently. Additionally, comprehensive panel testing has not been broadly standardized across genetic counseling practice. Here we describe outcomes of multigene panel testing in a cohort of high‐risk UM patients. We detected pathogenic/likely pathogenic (P/LP) variants and variants of uncertain significance (VUS) in 10% and 23% of patients, respectively. Notably, only two of these patients had BAP1 P/LP variants, supporting the value of multigene panel testing for high‐risk UM patients.

Abbreviations

ASCO

The American Society Clinical Oncology

NCCN

National Comprehensive Cancer Network

P/LP

pathogenic or likely pathogenic

UM

Uveal melanoma

VUS

variant of uncertain significance

1. Introduction

Approximately 12% of unselected UM patients show evidence of a hereditary tumor predisposition syndrome, with an 8%–10% chance of detecting a pathogenic variant in a hereditary cancer gene (Abdel‐Rahman et al. ²⁰¹⁰; Abdel‐Rahman, Sample, et al. ²⁰²⁰; Huang et al. ²⁰¹⁸). Risk is higher with early‐onset disease (age under 30), multiple primary cancers, and/or family history of UM or other cancers (Abdel‐Rahman, Pilarski, et al. ²⁰²⁰; Cebulla et al. ²⁰¹⁵). Although germline pathogenic variants in BAP1 represent the most well‐established cause of hereditary uveal melanoma, emerging data suggest broader genetic heterogeneity, complicating test selection and posttest interpretation. Limited real‐world data describing outcomes of multigene panel testing in this population make it challenging to counsel patients regarding expected results, downstream management, and family implications. As a result, evidence describing germline testing outcomes in high‐risk uveal melanoma patients is needed to better inform genetic counseling discussions and support shared decision‐making. Recent genome‐wide association studies and germline sequencing analyses continue to support a complex inherited susceptibility landscape in UM, further highlighting the importance of ongoing investigation into hereditary risk factors (Mies et al. 2025).

The National Comprehensive Cancer Network (NCCN) guidelines recommend referral to genetics and genetic testing in UM patients with (1) Early age of diagnosis (< 30 years of age); (2) History of other primary cancers in the patient; or a (3) Family or personal history of other cancers known to be associated with a hereditary syndrome (NCCN Guidelines, Melanoma: Uveal ²⁰²⁵). Only BAP1, BRCA1, BRCA2, PALB2, POT1, and MBD4 genes are listed as potential candidate genes in the NCCN guidelines; however, the guidelines do not specify the optimal germline testing approach or panel composition. The American Society Clinical Oncology (ASCO) guidelines currently recommend single BAP1 germline testing for patients with UM at risk for hereditary syndrome (Ricker et al. 2024). Additional international guidance has also evolved regarding hereditary uveal melanoma evaluation. Recent European hereditary cancer guidance has also referenced both BAP1 and MBD4 in familial or hereditary uveal melanoma cases (Lalloo et al. 2023), while the UK NHS Genomic Medicine Service recommends consideration of BAP1 and additional renal cancer predisposition genes in patients with both uveal melanoma and renal cell carcinoma (Lucas 2024). These evolving recommendations further support consideration of broader multigene germline testing approaches in selected high‐risk patients.

About 1%–2% of UM cases have a hereditary BAP1 pathogenic variant, the most common cause of hereditary UM (Abdel‐Rahman, Sample, et al. ²⁰²⁰; Gupta et al. ²⁰¹⁵). Individuals with a BAP1 pathogenic variant have an increased risk for UM, mesothelioma, renal cell carcinoma, melanoma of the skin, basal cell carcinoma, and other cancers (Pilarski et al. 1993). The current study seeks to evaluate the efficiency of the genetic testing recommendations from ASCO as well as the current NCCN criteria for referral for genetics in UM patients. We evaluated criteria and outcomes of genetic panel testing of UM patients referred to The Ohio State University cancer genetics clinic between 05/1/2021 and 9/19/2025.

2. Methods

2.1. Retrospective Chart Review

A retrospective chart review was conducted for UM patients seen by a genetic counselor that meet NCCN criteria in The Ohio State University Cancer Genetics Clinic between 5/1/2021 and 9/19/2025 due to a high risk for hereditary predisposition based on a personal or family history of cancer. This study was conducted under protocols approved by an IRB. During genetic counseling, individuals were offered large panel testing; genes in panel testing changed over the years due to clinical testing company preference (Table S1). Two additional patients had prior genetic testing with smaller panels (Table S1). Patients had the option to decline clinical genetic testing. All patients were assigned a study ID code; clinical data, including personal and family cancer history, were obtained from patient charts. Individuals were assessed via ASCO guidelines for a positive in a gene outside of BAP1.

Variant classification was determined from the Ambry Genetics Classifi Variant Classification Scheme, a points‐based framework which incorporates clinical evidence, molecular impact, and functional assays to determine variant pathogenicity.

2.2. Statistical Analysis

Patients were grouped into three categories: those with P/LP variants, those with VUS, and those who were negative for testing. Significance of difference between various demographic and clinical features between the three groups was assessed using Student's t‐test and Fisher's exact test (Table 1). Significance of difference between the UM cohort frequency and the GnomAD population frequency was determined using Fisher's exact test two‐sided mid‐p value, as previously described (Abdel‐Rahman, Sample, et al. ²⁰²⁰).

TABLE 1.

Summary of the clinical phenotype of Uveal Melanoma patients.

	P/LP	VUS	Neg	Total (P/LP, VUS, Neg)	p a
					P/LP vs. Neg	P/LP vs. VUS	VUS vs. Neg
Median age (range)	63.5 (45–81)	53 (22–76)	55.5 (17–87)	56 (17–87)	0.1013	0.1574	0.7332
Sex
M	2	7	19	28	0.2838	0.3989	1
F	8	9	25	42
Tumor location
Choroidal	8	15	34	57	0.4346	0.4769	0.6999
Ciliochoroidal	1	1	6	8
Iridociliary	0	0	1	1
Iridociliochoroidal	1	0	0	1
Iris	0	0	3	3
AJCC T category
1	2	5	17	24	0.476	0.2855	0.2031
2	5	3	15	23
3	1	4	8	13
4	2	1	3	6
Unknown	0	3	1	4
AJCC stage
I	2	5	17	24	0.05714	0.5577	0.1981
II	5	7	24	36
III	3	2	1	6
IV	0	0	1	1
Unknown	0	2	1	3
GEP class
1A	3	6	13	22	0.2345	0.0783	0.6753
1B	0	2	4	6
2	5	1	9	15
N/A	2	7	18	27
Primary treatment
Brachytherapy	8	11	33	52	0.7857	0.8103	0.3312
Other surgical	0	0	4	4
Other nonsurgical	0	2	1	3
Enucleation	2	3	5	10
No treatment	0	0	1	1
Metastasis b
Yes	0	6	2	8	1	0.0609	0.0019
No	7	8	40	55
Personal cancer history
Yes	6	9	20	35	0.4945	1	0.5634
No	4	7	24	35
Family cancer history
Yes	9	16	44	69	0.1852	0.3846	1
No	1	0	0	1

Abbreviations: AJCC, American Join Committee on Cancer; GEP, Gene expression profiling; Neg, Negative; P/LP, pathogenic/Likely pathogenic; VUS, Variant of Uncertain Significance.

^a p‐value determined by Student's t‐test (age) and Fisher's exact test (sex, clinical characteristics, cancer history).

^b Assessed for patients with at least 12‐month follow‐up in Ophthalmology or who had metastasis present at the time of diagnosis.

3. Results

Sixty‐eight patients were seen for genetic counseling and had large clinical genetic testing, and 2 patients had prior genetic testing with smaller panels (Table S1). Of the patients who received testing, 28 were men (40%), 42 (60%) were women, with an age of diagnosis ranging from 17 to 87 and a mean age of 56 years (Table 1). Most patients had choroidal tumors (81.4%), although tumors with ciliary and iris involvement were also present in this cohort. Most UM patients (74.3%) received primary brachytherapy. Among the 63 patients with at least 12 months of follow‐up, 8 (12.7%) developed metastatic disease. Notably, there was a significant difference between the development of metastasis between patients with VUSs in comparison to patients with negative results. Out of 14 VUS patients, 6 had metastatic disease progression. Given the small sample size and lack of functional data, interpretations about the underlying pathogenicity of the VUS's could not be made. Many UM patients had a prior cancer diagnosis (50%), and all but one had family cancer history (98.6%).

Ten individuals tested positive for pathogenic or likely pathogenic (P/LP) alterations in known cancer genes (BAP1, BRCA1, BRCA2, MBD4, MUTYH, POT1, XRCC2) (Table 2, Table S2). (8/10) of the positives would have been missed if testing was only done for BAP1. Two patients were in the same family and positive for the XRCC2 variant, thus our positive rate in unrelated patients was 9/69 (13%) and that family will be counted as one individual going forward. Importantly, the positive rate for pathogenic variants in an actionable, dominant cancer gene (2 BAP1, BRCA1, BRCA2, MBD4, 2 POT1) was 7/69 (10.1%) Two patients 2/69 (2.9%) were noted to be carriers of hereditary cancer syndrome genes, having one copy of a cancer gene with known recessive inheritance (MUTYH and 2 XRCC2). Forty‐four patients (63.8%, 44/69) tested negative for any germline cancer gene alteration. Five (7.2%) patients with a P/LP variant also had at least one VUS, and 16 (23.2%) additional patients had least one VUS (Table 1, Tables S2 and S3). MBD4 was added to the 77‐gene panel in November 2024, so 65 patients were not tested for MBD4.

TABLE 2.

Summary of the clinical phenotype of patients with pathogenic/likely pathogenic variants.

Patient ID	Sex	Age	Gene panel a	P/LP variant	VUS	Other personal cancer Hx	Family Hx of cancer	AJCCT category	AJCC stage	Treatment type
1	F	69	77a	BAP1, c.1526C>A		Basal cell carcinoma Endometrial‐62	Mesothelioma, UM, liver	T2a	IIA	Brachytherapy
2	F	62	71	BAP1, c.1717delC	MLH1, c.1874A>G MSH3, c.1249C>T	No	Kidney × 3 (26, 62, 50), Melanoma‐62, lung, prostate, brain, colorectal, cervical, breast, stomach	T4b	IIIB	Enucleation
3	F	68	77a	BRCA1, c.5109T>G		Ovarian‐54	Ovarian, breast	T2a	IIA	Brachytherapy
4	F	43	77a	BRCA2, c.5946delT		Breast‐44	Cutaneous melanoma, lung	T1a	I	Brachytherapy with FNAB
5	F	63	85	MBD4, c.1670T>A	BRIP1, c.1489G>C	No	Lung × 2, melanoma	T2a	IIA	Brachytherapy with FNAB
6	F	64	77a	POT1, c.1851_1852delTA	POT1, c.670G>A	Lung‐79	Lung × 3, tonsillar	T2a	IIA	Brachytherapy
7	F	77	71	POT1, c.122dupC		Breast‐73	Unknown	T2d	IIIA	Brachytherapy, enucleation
8	F	81	77a	MUTYH, c.734G>A	BRIP1, c.2379 + 4G>A	No	Bladder, melanoma, prostate × 3, basal cell	T3a	IIB	Enucleation
9	M	54	77a	XRCC2 b , c.651_652delTG	CDKN2A, c.‐2A>T	No	Ovarian‐45 Bladder‐92 Leukemia‐70 Skin	T1a	I	Brachytherapy w/o FNAB
10	M	45	77a	XRCC2 b , c.651_652delTG	CDKN2A, c.‐2A>T	Thyroid‐40 Basal cell‐76		T4a	IIIA	Brachytherapy, laser, enucleation

Abbreviations: Age, Age at diagnosis; FNAB, fine needle aspiration biopsy; P/LP, pathogenic/Likely pathogenic; VUS, Variant of Uncertain Significance.

^a Gene panels 77a, 71 did not include MBD4; Gene panels 71, 85 did not include XRCC2 (see Table S1).

^b These individuals are related.

We compared the frequency of P/LP variants in our cohort to that of a control population (gnomAD v4.0, non‐Finnish European group) (Table 3). Three genes (BAP1, POT1, XRCC2) showed a significantly higher frequency of P/LP variants in our cohort relative to the control population. Notably, a P/LP variant in one gene (XRCC2) has not been previously reported in UM patients.

TABLE 3.

Frequencies of P/LP variants in UM cohort and public database.

Gene	UM cohort P/LP carrier frequency	Population P/LP variant frequency a	Odds ratio (95% CI)	p b	Published UM cases with P/LP variants
BAP1	2/70 (2.9%)	77/590031 (0.013%)	218.9 (35.5–781.1)	4.509e−05	> 50 reported cases
BRCA1	1/69 (1.4%)	733/590031 (0.124%)	11.7 (0.6–59.1)	0.08697	42,10–13
BRCA2	1/70 (1.3%)	1913/590031 (0.324%)	4.4 (0.2–22.2)	0.22850	813–16
MBD4	1/5 (20%)	949/590031 (0.161%)	N/A	N/A	2412,16–21
MUTYH	1/68 (1.5%)	574/590031 (0.097%)	15.1 (0.7–76.7)	0.06719	N/A
POT1	1/69 (1.4%)	269/590031 (0.046%)	63.5 (10.4–219.8)	0.00051	33,22
XRCC2	1/54 (1.9%) c	397/590031 (0.067%)	27.5 (1.3–140.8)	0.03709	N/A

Abbreviations: CI, Confidence interval; N/A, No answer; P/LP, pathogenic/likely pathogenic; UM, Uveal Melanoma.

^a The gnomAD version 4.1.0 non‐Finnish European cohort frequency was estimated by the number of variant alleles reported as P/LP variants and null variants not reported as benign or of uncertain significance.

^b p‐value determined by Fisher's exact test with two‐sided mid‐p value.

^c Frequency of XRCC2 mutations excluded related individuals.

4. Discussion

This is the first study to assess outcomes of large panel testing in high‐risk UM patients who met NCCN criteria. Seven of the P/LP variants were in actionable/potentially actionable genes with available management guidelines (BAP1 [2], BRCA1, BRCA2, MBD4, POT1 [2]), highlighting the importance of referring UM patients with a strong personal and/or family history of cancer to genetic counseling and testing (NCCN Guidelines, Melanoma: Uveal ²⁰²⁵). Pathogenic variants in these genes have been reported previously in UM patients (BAP1 [2], BRCA1, BRCA2, MBD4, POT1 (2)) (Abdel‐Rahman, Sample, et al. ²⁰²⁰; Huang et al. ²⁰¹⁸; Repo et al. ²⁰²⁵; Mersch et al. ²⁰¹⁵; Johansson et al. ²⁰¹⁹; Cruz et al. ²⁰¹¹; Easton et al. ¹⁹⁹⁷; Moran et al. ²⁰¹²; Le Ven et al. ²⁰²⁵; Derrien et al. ²⁰²¹; Rodrigues et al. ²⁰¹⁸; Palles et al. ²⁰²²; Saint‐Ghislain et al. ²⁰²²; Villy et al. ²⁰²⁴; Nathan et al. ²⁰²¹). Currently, the NCCN UM genetic testing guidelines list five genes: BAP1, BRCA1, BRCA2, PALB2, and MBD4, to be associated with genetic predisposition to UM, while the ASCO guidelines recommend testing for only a single gene, BAP1 (NCCN Guidelines, Melanoma: Uveal ²⁰²⁵; Ricker et al. ²⁰²⁴). This study supports our earlier finding of phenotypic and locus heterogeneity in patients with evidence of UM‐related tumor predisposition, consistent with findings from other groups (Abdel‐Rahman, Sample, et al. ²⁰²⁰). A recent French population‐based study, Le Ven et al., analyzed 381 patients with UM and identified 21 pathogenic variants across BAP1 (3), MBD4 (7), BRCA2 (3), MLH1 (1), MSH2 (1), MSH6 (2), and PMS2 (4), supporting the possibility that individuals with Lynch syndrome are also at risk for UM (Le Ven et al. 2025). In addition, a Finnish germline study of 270 unselected UM patients excluding those with known pathogenic variants in BAP1 or MBD4 identified single P/LP variants in BRCA1, BLM, and MET (Repo et al. 2025). This data further broadens the spectrum of cancer predisposition genes associated with UM and reinforces the utility of large multigene panel genetic testing, rather than small panel or single gene, for UM patients with features suggestive of hereditary cancer risk.

Across our current and prior studies, germline pathogenic or likely pathogenic variants have been identified in multiple additional hereditary cancer genes reported in UM patients. Together with emerging data from other cohorts, these findings support the genetic heterogeneity of hereditary UM and provide further rationale for multigene panel testing rather than single‐gene approaches in clinically high‐risk patients (Abdel‐Rahman, Sample, et al. ²⁰²⁰; Huang et al. ²⁰¹⁸; Repo et al. ²⁰²⁵; Mersch et al. ²⁰¹⁵; Johansson et al. ²⁰¹⁹; Cruz et al. ²⁰¹¹; Easton et al. ¹⁹⁹⁷; Moran et al. ²⁰¹²; Le Ven et al. ²⁰²⁵; Derrien et al. ²⁰²¹; Rodrigues et al. ²⁰¹⁸; Palles et al. ²⁰²²; Saint‐Ghislain et al. ²⁰²²; Villy et al. ²⁰²⁴; Nathan et al. ²⁰²¹). Identification of germline P/LP variants in BRCA1, BRCA2 and in this study provide additional, although limited, evidence for the potential association of these genes with predisposition to UM. Since the majority of UM are treated with radiation and tumor tissue is rarely available, we were unable to assess biallelic inactivation (second hit) in the tumor.

The contribution of other genes, including MUTYH and XRCC2, to UM risk warrants further investigation. MUTYH and XRCC2 are associated with autosomal recessive cancer syndromes—MUTYH‐associated polyposis and Fanconi anemia, respectively—where biallelic inactivation is typically required for disease manifestation. However, single P/LP variants in both genes have been reported in patients with various malignancies, raising the possibility that single pathogenic variants may confer increased cancer risk (Kluźniak et al. 2019). XRCC2 is involved in homologous recombination DNA repair, a pathway implicated in hereditary cancer predisposition and melanoma biology. Additional studies are needed to clarify whether such variants contribute to UM predisposition.

In clinical practice, we apply criteria similar to NCCN but extend the definition of early‐onset uveal melanoma from diagnosis before age 30 to before age 40, consistent with NCCN age thresholds used in other cancers like breast and colon (Genetic/Familial High‐Risk Assessment: Breast, Ovarian, Pancreatic, and Prostate ²⁰²⁵; Genetic/Familial High‐Risk Assessment: Colorectal, Endometrial, and Gastric ²⁰²⁵). Using this broader clinical approach, we have also observed rare germline findings in additional cancer susceptibility genes in selected early‐onset UM patients, although the contribution of many of these genes to UM predisposition remains uncertain and requires further study. In this study, NCCN guidelines effectively prioritized insurance coverage for genetic testing, and expanding the early‐onset threshold may reduce the risk of missing hereditary UM cases.

This study has several limitations, including the relatively small sample size and the single‐institution nature of the cohort, which may limit generalizability to broader populations. In addition, MBD4 and other recently implicated UM predisposition genes were not included on earlier versions of the testing panels. An additional limitation of this study is the relatively limited follow‐up duration for some patients, as metastatic spread in UM may occur years after initial diagnosis.

Although the contribution of numerous genes to UM risk remains uncertain, our results expand the possible germline landscape of UM and highlight its genetic heterogeneity. Future studies integrating germline and somatic sequencing with functional assays will be critical to establish causality, refine risk estimates, and inform evidence‐based updates to genetic testing guidelines. As large‐scale genomic datasets and collaborative registries continue to grow, a clearer understanding of UM predisposition genes will emerge—advancing both precision risk assessment and mechanistic insight into this rare malignancy.

5. Conclusions

Multigene panels present new opportunities to discover pathogenic variants contributing to UM. Our results suggest that germline variants in cancer genes could independently impact disease outcome and prognosis, but validation in a larger cohort is needed. Current NCCN guidelines are appropriate for identification of high‐risk UM patients for testing, and using a large genetic panel rather than a single gene or limited genes panel is highly recommended.

Author Contributions

Kaylee Ramsey: data curation. Emma Schreiner: data curation, writing – original draft. Mohamed H. Abdel‐Rahman: conceptualization, methodology, supervision, writing – original draft, writing – review and editing. Frederick Davidorf: conceptualization, supervision. Isabella Gray: methodology, data curation, writing – original draft, writing – review and editing. Jenna Wolfe: data curation, writing – original draft. Lindsey Byrne: conceptualization, methodology, data curation, formal analysis, writing – original draft, writing – review and editing. Olivia B. Taylor: formal analysis, writing – review and editing. Joseph McElroy: formal analysis, writing – review and editing. Colleen M. Cebulla: conceptualization, methodology, supervision, writing – review and editing.

Funding

Program under Award No. W81XWH2110675 (PI: Mohamed Abdel‐Rahman). Opinions, interpretations, conclusions and recommendations are those of the author and are not necessarily endorsed by the Department of Defense. This work was also supported by the Patti Blow Research Fund in Ophthalmology, OLERF, Cancer Center Core Grant 2P30CA016058‐40 and award number 8UL1TR000090‐05 from the National Center For Advancing Translational Sciences, the OSU Vision Sciences Research Core Program P30EY032857, and the RPB New Chair Challenge Grant.

Ethics Statement

This work was done under research protocols approved by the Institutional Review Board of The Ohio State University.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Genetic testing panels.

Table S2: P/LP and VUS protein consequence & reference ID summary.

Table S3: Summary of the clinical phenotype of patients with variants of uncertain significance.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.