Choroidal melanoma is the most common primary intraocular malignancy of adults, and complications arising from metastatic disease are the most common cause of death. The Collaborative Ocular Melanoma Study (COMS) trials were a series of multicenter, randomized clinical studies examining treatments for uveal melanoma. In the COMS, the presence of metastatic disease at the time of diagnosis was less than 1%. However, the melanoma-specific mortality rates at 5 years were 10% for medium and 30% for large tumors. Since rates of local tumor control, by either radiotherapy or enucleation, approximate 100%, the assumption made by most researchers is that the majority of patients that died from metastatic ocular melanoma in the COMS trials had micrometastases at the time of diagnosis that were undetectable by the methods used in the COMS.
In the face of the low metastatic detection rate in the COMS, many clinicians agree that preoperative screening for metastatic disease in patients with choroidal melanoma is justified for 3 main reasons, each with significant controversy:
Patients with metastatic disease have a limited survival not affected by local therapy and thus the presence of metastases at diagnosis may obviate, or alter the approach to, local therapy. In contrast to this theory, some patients with metastatic disease do have prolonged survival. For example, Rietschel and associates reported that 22% of 119 patients were still alive 4 years after diagnosis of metastases. Thus, it may be reasonable for patients to undergo local globe-conserving therapy despite the presence of concurrent metastatic disease in order to preserve vision and prevent ocular complications that may arise from unchecked tumor growth.
Patients with metastatic disease may respond better to therapy if the metastases are detected earlier. Recently, a meta-analysis of data from prospective studies analyzing survival in patients with metastatic uveal melanoma revealed a median survival of 5 to 24 months after diagnosis of metastases, with a reported prolongation of survival in selected patient groups subjected to aggressive invasive interventions. This apparent prolongation of survival, however, was likely explained by surveillance and lead-time biases and not directly by the screening itself or any subsequent treatments. Therefore, the detection of metastatic disease at an earlier stage may only appear to prolong survival, but not actually extend it.
Many patients desire prognostic information to better help with life planning. Cook and associates reported that 97% of patients who underwent cytogenetic testing desired prognostic information about the possibility of metastatic disease even if it would not prolong their survival. However, imaging studies may result in false-positive results, subjecting patients to unnecessary biopsies and additional procedures, and studies of patient preferences with respect to screening have not addressed the consequences of incidental findings.
Current Screening Strategies
No prospective trials have compared different preoperative screening regimens. In the COMS, preoperative metastatic screening included liver function tests (LFTs) and chest radiographic imaging.
Liver Function Tests
The liver is the initial site of metastasis in 46% of patients, and up to 93% of patients with metastatic disease eventually develop hepatic metastases. Because LFTs are a relatively insensitive marker for mild liver dysfunction, these tests are generally normal, even in patients with known metastatic disease. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) associated with at least 1 abnormal LFT before diagnosis of metastasis in patients in the COMS after local treatment were only 14.7%, 92.3%, 45.7%, and 71.0% respectively.
The laboratories ordered in an LFT panel typically include aspartame and alanine aminotransferase, alkaline phosphatase (Alk Phos), and lactate dehydrogenase (LDH). Among these tests, LDH may have the highest sensitivity and PPV. Because both LDH and Alk Phos may reflect tumor burden, these tests may have prognostic significance as well. Other serum markers such as osteopontin, S-100beta, and melanoma-inhibitory activity have been studied and may play roles in screening for metastatic disease.
The sensitivity and specificity of chest radiography (CXR) in a series of 235 choroidal melanoma patients undergoing preoperative testing were reported to be 1.8% and 100% respectively. In the COMS, even when restricting the analysis to CXRs obtained within the 90-day period before diagnosis of metastasis, sensitivity, specificity, PPV, and NPV were 35%, 98%, 65%, and 93%.
The COMS and other studies have recommended preoperative CXR to rule out a primary lung tumor, but no routine follow-up CXR. However, since the PPV of the test prior to local treatment is even lower than when used for follow-up, the usefulness of this test at diagnosis or follow-up is questionable. Patients who are current or former smokers or those who have pulmonary symptoms may be particularly good candidates to undergo this test both at diagnosis and on follow-up.
Abdominal Computed Tomography Scan
Based on the low sensitivity of LFTs, clinicians have explored the use of abdominal imaging for metastatic detection. Feinstein and associates reviewed the records of 91 patients who underwent computed tomography (CT) scanning within 1 month of uveal melanoma diagnosis, and analysis of their data indicated a sensitivity, specificity, PPV, and NPV of 100%, 91%, 27%, and 100%, respectively. The low PPV was attributable to a variety of benign hepatic lesions detected with CT.
Abdominal Ultrasound (US)
Eskelin and associates reported on 46 uveal melanoma patients who developed metastases after undergoing baseline screening examinations followed by yearly examinations utilizing LFTs, CXR, and abdominal ultrasound (US). Abdominal US revealed unequivocal hepatic metastases in 36 patients (78%), of whom 12 (33%) had normal LFTs. US was suggestive of metastases in 5 additional patients (11%), all of whom were confirmed to have hepatic metastases by fine-needle aspiration biopsy, CT, or both. US was negative in 2 patients (4%), both of whom had liver metastases and at least 1 abnormal LFT. Of 30 metastases identified in 4 patients at diagnosis, US detected 37%, magnetic resonance imaging (MRI) detected 67%, and CT detected 77%; however, the authors recommended US given its lower cost and easier accessibility. The authors concluded that screening with abdominal US, CXR, and LFTs at diagnosis and subsequent semiannual LFTs and abdominal US would result in detection of 98% of metastatic lesions prior to patients’ reporting of symptoms.
Positron Emission Tomography/Computed Tomography
Positron emission tomography/computed tomography (PET/CT) combines high-resolution anatomic and functional imaging by using radioactive glucose. Although early case reports and small series were encouraging in that they identified patients with metastatic disease whose LFTs were normal, the prevalence of metastatic disease in the largest published study was only 3.8%. Therefore, while PET/CT is an improvement over prior screening techniques, with current resolution capabilities a significant proportion of metastases are likely still being overlooked. Additionally, compared with the previous imaging modalities discussed, PET/CT is expensive and, like other imaging modalities, false-positive findings may lead to unnecessary imaging and biopsies.
Abdominal Magnetic Resonance Imaging
While no studies have been published to date assessing the sensitivity of MRI for uveal melanoma micrometastases, several centers perform this test routinely. Advantages include high-resolution images, relative accessibility compared to PET/CT, and lack of radiation exposure. The main disadvantages are high cost and potential false-positive findings.
In attempts at genetic prognostication, various approaches have been employed that involve acquisition of tumor tissue at the time of initial treatment. Initially, certain chromosomal alternations were recognized to portend poor prognoses, including monosomy 3, loss of 1p and 8p, and gain of 8q. More recently, patterns of tumor RNA expression have been shown to be more accurate predictors of which patients are more likely to develop metastatic disease. However, such genetic testing is not currently absolutely predictive of metastases and collection of samples for analysis remains controversial. This is not preoperative testing per se, but has become an integral part of the diagnostic algorithm at many centers.