Screening for Metastasis From Choroidal Melanoma: The Collaborative Ocular Melanoma Study Group Report 23

Metastasis is common following primary diagnosis of choroidal melanoma. Patterns of metastatic spread have been described. Frequently the site of metastasis is the liver, with the proportion of cases with liver metastasis ranging from 56% to 100%.^1-8 Diagnosis of metastasis is often followed by rapid physical decline of the patient. Currently, there are few life-prolonging treatments for the majority of patients with liver metastasis or other distant sites of metastasis. The median time from diagnosis of liver metastasis to death is approximately 6 months.⁹ Early detection might identify individuals at a time when intervention may decrease the rate of metastatic spread or extend survival. Early detection also may help the patient and family in planning future medical and personal care. After diagnosis with choroidal melanoma, patients typically are screened for metastasis on a semiannual or annual basis using liver function tests (LFTs) and imaging studies. Tests that have been used include AST, ALT, alkaline phosphatase (APH), gamma-glutamyl transpeptidase, lactate dehydrogenase (LDH), or bilirubin (BIL). An abnormal or elevated LFT finding usually triggers the administration of a subsequent diagnostic test such as biopsy, magnetic resonance imaging (MRI) or computed tomography (CT) scans.^10-13 More recently, routine screening by ultrasonography of the liver and a panel of LFTs have been recommended.^14,15

Retrospective evaluations of screening tests for metastases in patients with choroidal melanoma have been reported. Hicks et al¹⁴ reviewed baseline LFTs of 235 patients and classified them as either normal or abnormal based on laboratory normal ranges. Sensitivity per test ranged from 1.9% for bilirubin to 25% for alkaline phosphatase; specificity ranged from 86% to 99%, respectively. Imaging tests were found to be highly specific but lacked sensitivity. For example, chest x-ray and liver ultrasound had 100% specificity but associated sensitivity of 1.8% and 14%, respectively. Eskelin et al¹⁵ had similar findings when examining the diagnostic utility of LFTs, chest x-ray and abdominal ultrasound before and at the time of diagnosis of metastasis in 46 patients with metastatic choroidal melanoma.

The Collaborative Ocular Melanoma Study (COMS) was designed to evaluate forms of radiotherapy for choroidal melanoma with respect to overall survival and metastasis-free survival.^16,17 In the COMS, the linear measurements of apical height and largest basal diameter were used to classify a tumor as small (1 to 2.5 mm in apical height and > 5 mm in largest basal diameter) medium (> 2.5 to 10 mm in apical height and ≤ 16 mm in largest basal diameter), or large (≥ 2 mm in apical height and > 16 mm in largest basal diameter or > 10 mm in apical height).¹⁶ Pre-enucleation radiation therapy was compared with enucleation among 1,003 patients with large tumors enrolled by December 15, 1994;¹⁸ patients randomly assigned in the large-tumor trial were followed at COMS clinical centers through July 2000, with some patients followed for a total of 13 years. I-125 brachytherapy was compared with enucleation among 1,317 patients with medium tumors enrolled by July 31, 1998;¹⁹ patients randomly assigned in the medium-tumor trial were followed at COMS clinical centers through July 2003, with some patients followed for a total of 15 years. The COMS provides the largest series to date of patients with choroidal melanoma without detectable metastasis or other cancers present at time of treatment and who were assessed prospectively, enrolled, treated, and followed in a standard fashion. Our aim is to describe screening procedures and diagnosis of metastasis within the COMS over a 15-year period in these two randomized trials.

The COMS design and methods have been published.¹⁶ The COMS Manual of Procedures¹² and the COMS Forms Book¹⁷ are available. Before patient enrollment, the institutional review boards of all participating institutions reviewed and approved the COMS protocol and patient consent forms. All patients gave written consent before enrollment and random assignment to treatment. Descriptions of patients who enrolled onto the randomized trials^18,19 and initial mortality findings by treatment arm^20,21 also have been published.

The COMS procedures for initial and follow-up systemic medical evaluations were developed and approved in 1986 by the COMS Executive Committee and approved by the COMS Data and Safety Monitoring Committee; they remained in effect through the end of the study. A medical evaluation of each patient, regardless of tumor size, was performed before enrollment in the COMS to determine eligibility and rule out metastatic disease or another malignancy. A systemic evaluation was performed that included a complete cancer-oriented general physical examination and LFTs. Required laboratory tests to assess liver function were AST, ALT, APH, and BIL. A test result was considered abnormal under the following conditions: AST > 2 × the upper limit of normal for the reference laboratory; ALT > 2 × the upper limit of normal for the reference laboratory; APH > 1.5 × the upper limit of normal for the reference laboratory; BIL ≥ 2.0 mg/100 mL.

Abnormal LFT findings required repeat laboratory testing for verification. Subsequent to verification, biopsy of suspicious lesions, CT or MRI scan of the liver, ultrasonography, or other diagnostic procedures were required to confirm or rule out metastatic disease.

In the COMS, patients in the large tumor trial were re-examined 6 months and 12 months after enrollment and annually thereafter until July 2000. Patients in the medium tumor trial were re-examined semiannually for the first 5 years after enrollment and annually thereafter; semiannual ophthalmic examinations for COMS data collection were optional after 5 years. Each patient enrolled onto one of the COMS trials was required to have an annual medical evaluation. Additional medical visits were scheduled to evaluate symptoms as required for good patient care. The follow-up evaluation was similar to that performed at the baseline visit and included a physical examination and LFTs. A chest x-ray was performed annually. Diagnostic imaging tests or pathology examinations were performed on the basis of elevated LFTs or whenever hepatomegaly was observed. Whenever metastatic melanoma was diagnosed on the basis of any of the diagnostic tests, a report was submitted by the clinical center coordinator to the COMS Coordinating Center (Baltimore, MD), along with copies of reports of all procedures that were performed to support the diagnosis. Biopsy or cytology slides or blocks of embedded tissue were sent to the COMS Pathology Center (Madison, WI) for histopathologic review. After the death of each COMS patient, clinic coordinators submitted any additional results and all medical and diagnostic reports regarding the patient's metastatic status at the time of death subsequently were reviewed by the COMS Mortality Coding Committee.²²

After the COMS was initiated, serum LDH was recognized as a prognostic factor for diminished survival in patients with metastatic cutaneous melanoma^23-25 and was incorporated in the revised staging of cutaneous melanoma by the American Joint Committee on Cancer.²⁶ Although LDH, albumin, and platelet levels were not collected systematically nor recorded for COMS purposes, a retrospective substudy was performed at five COMS clinical centers. For these centers, the COMS Coordinating Center provided a listing of study identifying numbers and name codes and all follow-up visits completed and dates for patients who had at least one abnormal COMS LFT reported after enrollment. The clinic coordinators at those centers, four of whom are registered nurses and one of whom is a physician's assistant, reviewed all available laboratory evaluations and extracted dates and values of these indicators, as well as the normal reference ranges for the associated laboratories, and returned the listings to the COMS Coordinating Center, where data were entered and integrated with other data on file for each patient. Test results were considered abnormal if the LDH level was above the upper limit of the normal reference range or if the albumin or platelet levels were below their respective lower limits of the normal reference ranges.

Of 2,320 patients enrolled onto the COMS randomized trials, a total of 714 patients were reported by the clinical centers as having a diagnosis of melanoma metastasis at some time during follow-up: 469 and 245 patients with large and medium choroidal melanoma at the time of enrollment, respectively. Table 2 depicts the distribution of procedures and tests used to confirm distant recurrence of melanoma. Pathologic confirmation (biopsy, cytology, autopsy) was available for 487 patients (68%). Clinical confirmation occurred frequently by scan, xray, physical examination and LFTs. Diagnosis was based on one test finding (clinical or pathologic) in only 136 patients (19%); two different test results were used in 187 patients (26%); three or more tests were available for 391 patients (55%). A total of 679 patients with melanoma metastasis reported by the clinical centers had died; the records of 675 patients were reviewed subsequently by the MCC. All but four cases were coded as either having histopathologically confirmed metastatic melanoma (79%) or clinical evidence of (suspected) metastatic melanoma (20%) at time of death. In addition, four patients of 323 who died without a prior diagnosis of metastasis were judged by the MCC to have histopathologic evidence of metastatic melanoma; for these cases, the date of diagnosis of metastasis was assumed to be the date of death. Findings were similar by tumor size group.

Overall, the 5- and 10-year cumulative metastatic melanoma rates, based on clinical diagnosis and/or mortality reviews, were 24% (95% CI, 23% to 26%) and 32% (95% CI, 30% to 34%), respectively. Rates of metastasis increased with tumor size. The 5-year cumulative rates were 38% (95% CI, 35% to 41%) and 14% (95% CI, 12% to 16%) for patients with large and medium-sized melanoma, respectively (Fig 1). Although incidence rates declined after the 2-year examination, there was no 6-month interval when no diagnosis of metastasis was made over the 12-year period. Of those diagnosed with metastasis, median survival was less than 6 months after the diagnosis. Overall, the Kaplan-Meier estimates of mortality at 1 year and 2 years after diagnosis of metastatic melanoma were 81% (95% CI, 78% to 84%) and 92% (95%CI, 90% to 94%), respectively. The sites of metastasis are shown in Table 3; the most common sites included liver (91%), lung (28%) and bone (18%) and were similar for patients enrolled with large and medium choroidal melanoma. The frequency distribution of sites was one, 52%; two, 25%; three, 13%; four, 7%; and five or more, 4%.

Table 4 displays by tumor size group the total numbers of tests of each type performed and reported during annual or semi-annual systemic evaluations. Of the combined LFTs performed, only 487 tests (< 1%) yielded abnormal findings in 221 patients. For each LFT, the median number of tests per patient was 7 (range, 0 to 28); 25% of patients had 10 or more tests.

Overall, the sensitivity, specificity, positive predictive value and negative predictive value associated with at least one abnormal LFT before first diagnosis of metastasis at any site was 14.7%, 92.3%, 45.7% and 71.0%, respectively; these values increased to 24.1%, 98.2%, 75.3% and 85.4% when restricting the time frame for test results to those obtained within 90 days before the reported diagnosis of metastasis. Restricting the analysis to test results obtained within the 90-day period before the diagnosis of metastasis resulted in increased diagnostic attributes for most tests. Results did not change appreciably for the analysis based on the last test results versus the analysis based on the test results obtained within the 90-day period. Table 5 depicts the diagnostic attributes of each liver function and diagnostic tests (CT, MRI, or other imaging scans) as well as findings of hepatomegaly and abnormal chest x-rays performed within 90 days of reported diagnosis of metastasis to any site, by tumor size group. For both tumor size groups, an individual LFT provided high specificity of 98% to 99%. However, sensitivity ranged from 0% to 19% and was higher for patients with large melanoma. PPV ranged from 67% to 92% for patients with large tumors and was lower for patients with medium-sized melanoma. Conversely, the NPV was approximately 70% versus 90% for patients with large versus medium tumors, respectively. Of the LFTs, alkaline phosphatase possessed the highest diagnostic attributes. As expected, diagnostic tests were associated with the highest sensitivity (88% to 94%), specificity (87% to 97%), PPV (94%) and NPV (87% to 93%), since these tests were mainly triggered by abnormal LFT findings. Similar results were obtained when metastasis was restricted to the site of the liver (data not shown).

For the subset of 43 patients at five COMS clinical centers where LDH, platelet and albumin levels were abstracted retrospectively, there were reports of 25 patients (58%) with LDH values above the normal reference range; six patients (14%) and 11 patients (26%) had low platelets or low albumin levels, respectively, on at least one test before diagnosis of metastasis. These patients were selected on the basis of time between first abnormal LFT to diagnosis of metastasis; time between first elevated LDH to diagnosis of metastasis revealed no systematic differences, suggesting no advantage of LDH in detecting metastasis before the routine COMS-required LFTs.

The COMS provided the opportunity to assess a large series of patients with choroidal melanoma who were followed prospectively and evaluated systematically using a standard protocol that included semi-annual LFTs and annual chest x-rays to screen for metastatic spread. The characteristics of our LFT results are similar to those reported by Hicks et al¹⁴ and Eskelin et al.¹⁵ In our series, both sensitivity and specificity improved with the use of multiple tests. Our choice of cutoff level for abnormal findings of LFTs was higher than reported in both earlier studies, which may have resulted in decreased sensitivity of COMS evaluations. Hicks et al concluded that hematologic tests were not useful for ruling out metastatic disease at presentation with uveal melanoma, due to their low positive predictive value. They recommended liver ultrasonography and chest x-ray at baseline only, given their findings of 100% specificity and low sensitivity for both tests. We recognize that LFT values are highly variable and may be influenced by other factors that elevate liver function levels, such as the use of cholesterol-lowering agents. The benefit of annual chest x-ray is questionable based on our results, in which less than 3% (297 of 11,948 x-rays performed) were abnormal. We concur with Eskelin et al¹⁵ that annual screening for metastatic uveal melanoma with LFTs is warranted and that chest x-rays should be required at baseline or whenever warranted by other test results.^29,30 Our analyses suggest that alkaline phosphatase may possess the best test attributes of the LFTs we examined but that multiple tests improve sensitivity. Also, our findings of lower diagnostic attributes associated with testing of patients with medium-sized tumors suggests that different screening strategies may be warranted based on tumor size at time of diagnosis and treatment.

We agree with others^14,15 that periodic screening for metastatic uveal melanoma is primarily beneficial when effective treatment is available or when patients are eligible to be candidates for clinical trials of promising treatment. Such therapies are needed given the potentially high prevalence of micrometastasis at time of diagnosis of choroidal melanoma and the high incidence of metastasis detected clinically after diagnosis of the primary tumor. Among 7,541 patients screened for COMS trials, only 70 (< 1%) had metastatic melanoma detected at the time of screening. In our randomized trials, the majority of deaths were associated with histopathologically confirmed or suspected metastatic melanoma. However, because of the short time from detection of metastasis until death, early detection of metastases may be valuable to the patient and family to permit personal planning and arrangements for future medical care. Although treatments for metastatic melanoma have demonstrated equivocal results with respect to survival, development and evaluation of new combined or systemic therapies^24,31-33 and adjuvant therapies³⁴ must continue to be pursued. In the interim, we recommend continued routine surveillance of LFTs in conjunction with diagnostic imaging studies such as ultrasonography, MRI, or other scans.

The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Received more than $2,000 a year from a company for either of the last 2 years: Wui-Jin Koh, Ortho Biotech.