Circulating tumor DNA accelerates diagnosis and treatment guidance for metastatic uveal melanoma with hepatic lesions not amenable to biopsy

Abstract

Uveal melanoma is a common intraocular tumor in which up to 50% of patients develop metastasis. Uveal melanoma often presents asymptomatically and is usually diagnosed by clinical examination and imaging, as tissue biopsy is not feasible. Recently, analysis of circulating tumor DNA has enabled diagnosis and mutation analysis in such tumors where tissue sampling cannot be performed. In our case study report, we demonstrate the use of the Caris Assure Assay for the diagnosis of metastatic uveal melanoma in a patient where tissue biopsy was not feasible. Plasma analysis identified a pathogenic SPEN variant and human leukocyte antigen (HLA) genotype (HLA-A*02:17 and HLA-A*02:01 alleles), conferring therapeutic eligibility for tebentafusp, and molecular evidence supportive of metastatic uveal melanoma, enabling initiation of appropriate systemic therapy before additional tissue sampling. Subsequent tissue biopsy from the liver (56 days after liquid biopsy) confirmed canonical GNAQ and SF3B1 driver mutations, reinforcing the diagnosis and clarifying the molecular underpinnings of the metastatic lesion itself. While the liquid biopsy and ablation of one of the lesions confirmed the diagnosis, the liquid biopsy was a key first step that provided a comprehensive diagnostic and prognostic picture without the need for an invasive procedure. This case highlights how integrating liquid biopsy into the diagnostic pathway for uveal melanoma can lead to earlier, more informed treatment decisions.

Introduction

Uveal melanoma, the most common primary intraocular malignancy in adults, is distinguished from cutaneous melanoma by its unique genetic profile and metastatic pattern. Approximately 50% of uveal melanoma cases develop metastases, with the liver as the first and often sole site in up to 90% of cases, a pattern that contributes to poor prognosis. Detecting metastatic disease early is critical, as it can guide therapy and improve outcomes [1]. Diagnosing metastatic disease conventionally requires tissue biopsy of suspicious lesions. However, biopsy of uveal lesions poses the risk of complications including hemorrhage, infection, and sampling error, and it is particularly challenging when hepatic lesions are less than 1 cm in size or located in anatomically inaccessible areas [2]. Early uveal melanoma metastases may also remain clinically silent and appear indeterminate on imaging [3], further complicating timely diagnosis.

Liquid biopsy offers a minimally invasive, repeatable alternative by detecting circulating tumor DNA (ctDNA) shed from the metastasis and other tumor-derived material in the blood. Analysis of ctDNA can identify hallmark uveal melanoma driver mutations such as GNAQ and GNA11, as well as prognostic markers like BAP1 mutations and monosomy 3, thereby providing both diagnostic and prognostic information [4,5]. Importantly, longitudinal monitoring of ctDNA levels often parallel tumor burden and may reveal progression earlier than imaging [5].

This report highlights a scenario in which liquid biopsy using next-generation sequencing (NGS) technology established a diagnosis of metastatic uveal melanoma in a patient with three liver lesions less than 1 cm in diameter, where tissue biopsy was not initially feasible. We aim to communicate the clinical utility of ctDNA analysis as a noninvasive alternative to tissue biopsy in establishing metastatic disease, and the necessity for targeted therapy and timely treatment.

Case report

Our patient is a 57-year-old female with a history of stage IIB uveal melanoma, initially diagnosed 9 years ago. The tumor was discovered during an eye exam for new glasses, where a lightly pigmented choroidal melanoma was found on the back of her left eye, measuring 14 mm in base and 6.3 mm in thickness. The patient had an unremarkable physical exam at the time with an Eastern cooperative oncology group of 0, retained liver and renal function, and no noticeable visual deficits. The tumor resulted in retinal detachment, but there was no extraocular extension. She was successfully treated with plaque radiotherapy, resulting in a regressed scar of 3.3 mm. Family history was notable for her mother’s breast cancer, diagnosed at age 40, and hematological cancer in her father, diagnosed in his 30s.

Following treatment, the patient was monitored with alternating liver ultrasounds and MRIs, in addition to full-body imaging. Subsequent imaging was negative for disease involvement in the liver, cardiopulmonary disease, cerebral extension, and mammary involvement. Re-evaluation two and a half years ago revealed a subcentimeter liver lesion that could not be safely biopsied. Liver MRIs were conducted every 3 months thereafter, showing slow growth of the one liver lesion. One year later, a biopsy was attempted on this lesion, as it was now 1 cm; however, the biopsy was unsuccessful, and imaging continued. Six months after the liver biopsy attempt, a second subcentimeter lesion developed, and more recently, 4 months ago, a third subcentimeter lesion developed. The patient now had lesions in liver segments 6, 7, and 8. Because of their size, none were amenable to biopsy.

Approximately 2 years following the initial appearance of the liver lesion, a blood-based circulating DNA/RNA test was performed, which detected a pathogenic SPEN tumor mutation (Variant allele frequency (VAF): 0.3%) in the patient’s bloodstream, as well as a PBRM1 variant of unknown significance (VAF: 0.6%), and HLA-A*02:01. SPEN is a strong indicator of uveal melanoma, providing confidence in the diagnosis of metastatic uveal melanoma [4]. The low disease burden and stability of the lesions prompted the team to refer the patient to Interventional Radiology for liver-directed therapy. A biopsy and successful ablation of the segment 6 lesion were performed. Immunohistochemistry was positive for S100, HMB45, SOX10, and Melan A, further supporting the diagnosis of metastatic uveal melanoma. The segment 7 and 8 lesions were not amenable to ablation because of their positioning behind the lung and small size (Fig. 1).

Fig. 1.

Axial imaging of one of the liver metastases in hepatic segment 6. (a–c) T1-weighted MRI demonstrates a T1 hyperintense lesion (black arrow in a) consistent with melanin-containing metastasis in segment 6. The lesion increases slightly in size from 9 to 10 mm over time: (a) 29 August 2024, (b) 6 December 2024, and (c) 9 June 2025. (d) Subtraction postcontrast T1 image from 6 December 2024 confirms arterial enhancement. (e) Intraprocedural ultrasound from 15 August 2025 shows the lesion (white arrow) during biopsy and subsequent microwave ablation. (f) Noncontrast CT from 15 August 2025 demonstrates microwave probe placement within the lesion. CT, computed tomography.

Given the patient’s HLA-A02:01 positivity and partial response to liver-directed therapy, where one of three hepatic lesions was successfully treated while the other two could not be targeted, the patient was subsequently considered for treatment with tebentafusp (Kimmtrak), which targets the HLA-A02:01/gp100 peptide. The patient will be given the first three doses during a hospital stay to allow for at least 18 h of monitoring for cytokine release syndrome. Subsequent infusions will be administered in an outpatient setting starting at week 4. The patient will undergo an MRI the following month, along with ongoing follow-up with oncology. The patient was also offered combination immune checkpoint inhibitor therapy with ipilimumab (Yervoy) and nivolumab (Opdivo), as an alternative treatment to tebentafusp.

Specimen processing

Molecular profiling was performed at Caris Life Sciences (Phoenix, Arizona, USA), a College of American Pathologists/Clinical Laboratory Improvement Amendments-certified laboratory. Hematoxylin and eosin-stained formalin-fixed, paraffin-embedded (FFPE) slides of the patient’s tumor underwent review by a board-certified pathologist or trained pathologist assistant. A minimum of 20% tumor nuclei in the area for microdissection was required, with a minimum 10 mm² dissection area. Tumor enrichment was achieved by harvesting targeted tissue using manual microdissection techniques.

MI cancer seek

Total nucleic acid (TNA) was auto-extracted using a MagMax FFPE DNA/RNA Ultra extraction kit (Thermo Fisher Scientific, Waltham, Massachusetts, USA). To perform simultaneous DNA and RNA sequencing using the MI Tumor Seek Hybrid assay, library preparation was performed on the Bravo Automated Liquid Handling Platform (Agilent, Santa Clara, California, USA) using KAPA Library Prep reagents (Roche, Indianapolis, Indiana, USA) and custom cDNA primers (IDT; Coralville, Iowa, USA; GeneLink, Elmsford, New York, USA). RNA was labeled during first strand cDNA synthesis. Custom KAPA baits panels were designed to enrich for 720 clinically relevant genes at high coverage and high read-depth, and an additional more than 20 000 genes at lower depth, along with single-nucleotide polymorphism and pathogen panels (Roche). Sequencing was performed on the NovaSeq 6000 System (Illumina, San Diego, California, USA) using recommended reagents. The average sequencing depth of this assay is 230x for 20 859 genes (whole exome), 1000x for 720 genes with known and potential clinical relevance, and 1500x for 230 reportable genes. RNA is sequenced to a minimum of 1.37 million total mapped reads. Sequencing data were extracted into split FASTQ files (RNA and DNA) for further processing using Caris’s bioinformatics pipeline. For RNA, Spliced Transcripts Alignment to a Reference (STAR) software was used for alignment, trimming, and fusion detection using the RNA FASTQ files from the TNA split pipeline [6]. Transcripts per million molecules were generated using the Salmon expression pipeline [7].

Caris assure

Patient blood samples underwent liquid biopsy testing using the Caris Assure^TM assay. Blood was collected in two PAXgene ccfDNA blood collection tubes (Qiagen, Hilden, Germany), which were shipped on the same day to Caris. Buffy coat and plasma were separated. Circulating DNA and RNA (ctNA) were sequenced simultaneously on a NovaSeq 6000 instrument (Illumina). To facilitate simultaneous analysis of DNA and RNA, RNA was labeled during reverse-transcription followed by bait-based pull-down of DNA and cDNA (RNA) exonic regions during library preparation. The data were then processed via Caris’s proprietary bioinformatics pipeline to detect single-nucleotide variants (SNVs), insertion/deletions (INDELS), fusion genes, and copy number alterations. White blood cells (buffy coat) were also sequenced in parallel to plasma to identify incidental pathogenic germline variants (PGVs) in 67 hereditary cancer genes and clonal hematopoiesis–derived variants. The average sequencing depth of clinically relevant genes in plasma by Caris Assure is 8000x, with comparable sequencing depth for the buffy coat. Compared with tissue testing performed within 30 days, this test has a sensitivity of 93.8% and specificity of less than 99.9% for actionable SNVs and INDELs, and more than 99% sensitivity and specificity for incidental PGVs.

Results

Initial liquid biopsy result

A liquid biopsy was performed as tissue biopsy material was insufficient for molecular testing. Given the clinical suspicion of metastatic disease, ctDNA analysis was pursued, as a liquid biopsy can provide genomic evidence of disease dissemination in uveal melanoma. In localized uveal melanoma, tumor variants are rarely detected in plasma; however, the presence of ctDNA variants can indicate metastatic involvement [8,9].

Both plasma and matched buffy coat samples were analyzed. This approach enabled differentiation of tumor-derived variants from Clonal hematopoiesis of indeterminate potential variants (clonal hematopoiesis of indeterminate potential), thereby reducing the risk of a false-positive interpretation. Plasma analysis identified a pathogenic SPEN nonsense variant (Q2597*, c.7789C>T). No incidental germline or Clonal hematopoiesis of indeterminate potential-related alterations were detected in the buffy coat, confirming the somatic origin of this finding.

In addition, the liquid biopsy assay identified the patient’s human leukocyte antigen (HLA) genotype by NGS. The patient was found to carry HLA-A*02:17 and HLA-A*02:01 alleles, conferring therapeutic eligibility for tebentafusp in the setting of HLA-A*02:01–positive metastatic uveal melanoma.

Collectively, these liquid biopsy results provided molecular evidence supportive of metastatic uveal melanoma and enabled initiation of appropriate systemic therapy without delay, rather than waiting for additional tissue sampling.

Tissue results

A liver biopsy was obtained 56 days after the initial liquid biopsy, once the lesion became accessible. Comprehensive genomic profiling of the tissue sample identified two well-established pathogenic variants in uveal melanoma: GNAQ Q209L (c.626A>T; VAF 46%) and SF3B1 R625H (c.1874G>A, VAF 40%). These represent canonical driver alterations in uveal melanoma, frequently associated with disease initiation (GNAQ/GNA11 mutations) and progression (SF3B1 mutations, often linked to later metastatic evolution). Immuno-oncology biomarkers were negative, including PD-L1 (22C3), tumor mutational burden (2 mut/Mb), and microsatellite instability. HLA genotyping again confirmed HLA-A*02:17 and HLA-A*02:01 alleles, supporting eligibility for tebentafusp in the setting of metastatic uveal melanoma.

Discussion

The patient’s initial presentation with a small, indeterminate choroidal melanoma highlights the insidious nature of uveal melanoma and the challenge of early detection. Uveal melanoma’s distinct metastatic pattern, primarily to the liver, further complicates management and underscores the need for effective, noninvasive diagnostic tools. Our case illustrates the clinical utility of liquid biopsy using NGS to diagnose metastatic disease when tissue biopsy is not feasible. Although liver lesions were noted on imaging, they were too small to biopsy safely. Because metastatic disease was strongly suspected, a liquid biopsy was performed. In uveal melanoma, ctDNA is typically undetectable from localized disease [9]. Therefore, the presence of tumor-derived variants in plasma would indicate metastatic spread. In this case, the liquid biopsy detected tumor-specific alterations, providing early molecular confirmation of metastatic disease. Prior reports of liquid biopsy in uveal melanoma have primarily centered on its value for prognosis and longitudinal disease monitoring, for example, correlating ctDNA levels with tumor burden or treatment response [2,5,10]. Our case extends this by expounding on the benefit of using liquid biopsy to establish metastatic disease and provide a therapeutic target before tissue biopsy was feasible, in this case, because of the small size and location.

Studies have demonstrated that ctDNA in uveal melanoma could be detected months before radiographic findings showed disease progression and additionally showed its value in complementing imaging when hepatic lesions of concern were indeterminate before invasive tissue biopsy was performed [5]. Our case fills a critical gap in the literature by demonstrating that ctDNA detection can not only anticipate disease progression but also provide clinically actionable diagnostic certainty by directly influencing eligibility for precision therapy. The detection of uveal melanoma specific mutations via liquid biopsy was significant in making the diagnosis of metastatic uveal melanoma, confirmed with tissue pathology and genomic testing more recently, as it made the patient a candidate for tebentafusp (Kimmtrak), the only Food and Drug Administration–approved systemic therapy for metastatic uveal melanoma that has been shown to prolong overall survival [11,12]. To our knowledge, very few reports have highlighted this application.

Tebentafusp is a T-cell receptor bispecific fusion protein that redirects T cells to kill melanoma cells that present the gp100 peptide in the context of the HLA-A*02:01 molecule. Since the drug recognizes the proteins of the specific HLA molecule, the HLA-A*02:01 phenotype is required for the drug’s activity, underscoring the importance of molecular profiling before treatment initiation [11]. The efficacy of tebentafusp is well-documented in the phase 3 IMCgp100-202 trial, where it demonstrated a significant overall survival benefit compared with other therapies. Patients receiving tebentafusp lived a median of 21.6 months, notably longer than the 16.9 months observed in the control arm, with a remarkable 27% of patients alive at three years compared with 18% in the control group [12].

Notably, the tissue biopsy findings differed from the earlier liquid biopsy, which had detected a pathogenic SPEN nonsense variant (Q2597*, c.7789C>T). Both biological and technical factors could explain this difference. First, the liquid biopsy was performed nearly 2 months before tissue sampling, allowing time for clonal dynamics and tumor evolution that could alter the spectrum of detectable variants. In addition, tissue biopsy provides a snapshot of the genomic landscape in a single metastatic lesion, while liquid biopsy aggregates signals across potentially multiple tumor deposits.

The liquid biopsy suggested metastatic disease and offered an early window into tumor genomics when tissue was unobtainable. The later tissue biopsy confirmed canonical GNAQ and SF3B1 driver mutations, reinforcing the diagnosis and clarifying the molecular underpinnings of the metastatic lesion itself. Together, these findings highlight the value of integrating liquid and tissue biopsy data: while each method has inherent limitations, their combined use can provide a more complete picture of tumor biology and guide timely therapeutic decisions.

In conclusion, while the biopsy and ablation of one of the lesions confirmed the diagnosis, the liquid biopsy was a key first step that provided a comprehensive diagnostic and prognostic picture without the need for an invasive procedure. This case highlights how integrating liquid biopsy into the diagnostic pathway for uveal melanoma can lead to earlier, more informed treatment decisions, particularly in cases where tissue samples are challenging to obtain.

Acknowledgements

Conflicts of interest

U.M.A. and G.H. declare employment at Caris Life Sciences. M.A.W. reports advisory boards for BMS and Regeneron. For the remaining authors, there are no conflicts of interest.