Variates of Survival in Metastatic Uveal Melanoma
http://www.100md.com
《临床肿瘤学》
the Departments of Medicine, Epidemiology and Biostatistics, and Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY.
ABSTRACT
PURPOSE: The course and outcome of metastatic uveal melanoma are not well described. We evaluated the survival of our patients with metastatic uveal melanoma, described factors that correlated with survival, and evaluated the influence of screening tests on time of detection and survival.
PATIENTS AND METHODS: All patients with metastatic uveal melanoma seen at Memorial Sloan-Kettering Cancer Center between 1994 and 2004 were identified from our database. We recorded date of initial diagnosis, date of metastatic disease, date of last follow-up, site of the first metastasis, how the first metastasis was discovered, treatment, and outcome of therapy.
RESULTS: The estimated median survival of the 119 patients analyzed was 12.5 months; 22% of patients were alive at 4 years. Five variates correlated independently with prolonged survival: Lung/soft tissue as only site of first metastasis, treatment with surgery or intrahepatic therapy, female sex, age younger than 60, and a longer interval from initial diagnosis to metastatic disease. Discovering metastatic disease in asymptomatic patients did not correlate with overall survival; 89% of patients had a single organ as the site of first metastasis. Although liver was the most common site, 39.5% of patients had nonliver sites, most commonly lung, as the first site of metastasis.
CONCLUSION: A substantial subset of patients with metastatic uveal melanoma survive more than 4 years with metastatic disease. Data on variates of survival and site of first metastasis may guide strategies for screening patients, although our data failed to show a survival advantage in discovering asymptomatic metastatic disease.
INTRODUCTION
Uveal melanoma arises from melanocytes located in the choroid, the layer between the sclera and the retina. Although it is the most common primary malignant primary tumor of the eye, there are only about 1,500 diagnoses per year in the United States.1 Despite optimal treatment (surgery or radiation), metastases often develop, although the true incidence of metastases from uveal melanomas has been difficult to determine. Differences in study design, diagnostic procedures, data collection, inaccuracies in biopsies of metastases, and even incorrectly diagnosed ocular tumors have led to disagreement in the literature. All studies agree that long-term prognosis is poor. Two retrospective analyses reported median survival from the time metastases was documented to be 6 months or less.2,3 Aside from surgery,4 treatment has not been shown to improve survival, and it is unclear whether early detection leads to improved survival.
Few studies have looked specifically at patients with metastatic uveal melanoma and studied the characteristics of these patients by analyzing the sites of metastasis, method of discovery of metastatic disease, treatment received by patients with metastatic disease, overall survival of patients presenting with metastatic disease, and correlates of survival. We have reviewed the Memorial Sloan-Kettering (New York, NY) experience with patients who had metastatic uveal melanoma seen at our institution over a 10-year period. We found that our patients with metastatic uveal melanoma survived substantially longer than the patients in these prior studies. We describe the natural history of uveal melanoma among our large patient cohort, quantitate survival from time of metastatic disease, describe factors that correlate with long-term survival in patients with metastatic uveal melanoma, and evaluate the value of screening blood tests, x-ray tests, and routine physical examinations on time of detection and survival.
PATIENTS AND METHODS
This is a retrospective analysis of stage IV uveal melanoma patients seen at Memorial Sloan-Kettering Cancer Center. All patients with stage IV uveal melanoma were included and a total of 119 patients (67 women and 52 men) who developed stage IV melanoma between 1994 and 2004 were identified from the melanoma database. The patients initially had been diagnosed with uveal melanoma between 1972 and 2003. Date of initial diagnosis, date of stage IV diagnosis, and date of last follow-up or death were recorded. We also recorded the site of the first metastasis, how the first metastasis was discovered, treatment of stage IV disease, and outcome of therapy. This retrospective analysis was performed with Institutional Review Board approval, which determined that this was exempt research under 45 Code of Federal Regulations 46.101.b.
Statistical Analysis
Time to stage IV was defined as the time from the date of initial diagnosis to the date of stage IV diagnosis. Survival from stage IV diagnosis was defined as the time from date of stage IV diagnosis to the date of death or last follow-up. Overall survival distributions from stage IV diagnosis were estimated using Kaplan-Meier methodology and comparisons between categoric variables were assessed using the log-rank test. A Cox proportional hazards model was used for multivariate analysis.
To determine whether tumor response was associated with improved survival, a landmark analysis was performed in which survival time was defined as the time from 2 months after treatment start (approximately the time of first response assessment) to the date of death or last follow-up. This was done to correct for patients who died from melanoma before they had an opportunity to be assessed for tumor response.5 This eliminated nine patients from the analysis. Without this correction, the comparison of survival between responders and nonresponders could be biased toward showing a benefit in the responder group.
RESULTS
The median age at diagnosis of metastatic melanoma for this cohort of 119 patients was 63.1 years (range, 25.7 to 86.4 years). As of last follow-up, 31 of 119 patients were still alive; the median follow-up of survivors was 17 months (range, 0 to 96 months). Estimated median time from initial diagnosis of uveal melanoma until detection of metastatic disease (time to stage IV) for all patients was 53 months (range, 0 to 359 months). Because our information was taken mainly from chart notes written at the time of diagnosis of stage IV, we did not have detailed information about size or location of the primary tumor.
Sites of Metastases
Given that information on the site of first metastasis is critical for determining a rational screening procedure and we could find no previous studies reporting the location of the first metastasis, we were interested in identifying the site of first metastasis in our patient population (Table 1). We found that 89% of patients had a single organ involved as the first metastatic site and that in 68% of these patients (60.5% of all patients), the liver was the first metastatic site. Surprisingly, however, almost 40% of all patients presented with nonliver sites as first metastasis; the most common site was the lung. This suggests that screening procedures that examine only the liver would be expected to miss a substantial proportion of patients.
Over the course of their metastatic disease, a total of 77.3% of patients eventually developed liver involvement and 42.9% never had disease beyond the liver. Consistent with autopsy studies,2 we found that a small percentage of patients developed metastases to bone (17%) and brain (10%). Although our data confirm that liver is the most common site of metastasis in uveal melanoma, a significant proportion of patients had nonliver organs as the site of first metastasis.
Methods of Discovery of Metastatic Disease
In our attempt to evaluate the value of screening procedures, we were interested in discovering how the first metastases were discovered in these patients. This information was available in 113 patients (Table 2). More than half of the patients were found to have metastatic disease based on symptom presentation. These patients presented with metastatic disease a median of 79 months from the time of initial diagnosis of primary uveal melanoma. A substantial proportion (37.2%) of patients were discovered to have metastatic disease based on radiographic imaging while they were asymptomatic. This cohort of patients was found to have metastatic disease much earlier (median, 30 months from the time of initial diagnosis) than patients who presented with symptoms (P = .0016). Blood tests and routine physical examinations rarely picked up the first metastasis in asymptomatic patients. These findings suggest that routine radiographic screening led to early discovery of metastatic uveal melanoma, but that physical examinations and blood tests in asymptomatic patients were of little value.
Treatment Received for Metastatic Uveal Melanoma
Data on treatment were available for 93 patients; outcome data were available for 88 of 119 patients. Of 19 patients (20%) who underwent surgical resection, 10 patients had complete surgical resections. Of 18 patients (19%) who received intrahepatic chemotherapy, one patient had a complete response to the therapy. Systemic therapy was given to 41% of patients (chemotherapy, immunotherapy, or other investigational treatments) and 19% received no treatment for their metastatic disease. Few of the patients were treated within the context of a clinical trial, and although partial responses and stable disease were reported by some of the treating physicians, we could not formally assess these responses. Complete responses, although rare, were only achieved with local therapy, such as surgery or intrahepatic chemotherapy.
Overall Survival
The survival curve showed an estimated median overall survival of 12.5 months (95% CI, 9.5 to 15.7 months; Fig 1). A significant minority of patients appears to have had a more indolent course, indicated by the observation that after 18 months, the slope of the survival curve flattens out. As a result, an estimated 22% of patients were alive at 4 years. This suggests that metastatic uveal melanoma consists of at least two distinct subpopulations with different tumor biology.
Univariate Analysis of Survival
To identify characteristics that correlated with survival, we tested a set of variates in a univariate analysis (Table 3). We found that age younger than 60 years, female sex, first metastases involving lung or soft tissue, a longer time interval from initial diagnosis to development of metastatic disease, and a history of having undergone surgery or intrahepatic chemotherapy all correlated with prolonged survival. Patients who received surgery or intrahepatic chemotherapy (local treatment) had a particularly improved survival (median, 32.4 months), compared with patients who received systemic chemotherapy (median survival, 9.5 months). This is presumably related to patient selection; patients who were candidates for surgery or intrahepatic chemotherapy would have had more localized metastatic disease.
We also noted that patients who had had a complete or partial response to chemotherapy or surgery had a much longer estimated median survival (49.1 months) compared with treated patients who did not have a response (9.5 months). Although this difference was highly statistically significant, we did not include tumor response in the multivariate analysis because tumor responses were not assessed formally in this patient set.
We were surprised to note that neither the method of discovery of metastatic uveal melanoma (symptoms v routine screening) nor the person who discovered the metastasis (patient v physician) affected survival significantly.
Multivariate Analysis of Survival
We tested these variates in a multivariate analysis and found five variates that correlated independently with prolonged survival (Table 4). Lung or soft tissue as only site of first metastasis correlated with an improved survival, as did treatment with surgery or intrahepatic therapy, female sex, age younger than 60, and a longer time interval from diagnosis to metastatic uveal melanoma. Whether metastases were first discovered by screening tests (computed tomography scans, blood tests, or routine physical examination) or as a result of symptoms did not correlate with the likelihood of having surgery or with overall survival.
DISCUSSION
Metastatic uveal melanoma is a substantial problem, but relatively little has been published characterizing the prognosis and variates that correlate with survival. Except for a small single-institution study,6 no one has looked carefully at correlates of survival in metastatic uveal melanoma. We have described our single-institution experience in 119 patients with metastatic uveal melanoma. We observed that the estimated median overall survival was 12 months, but that a substantial proportion of patients demonstrated a more indolent course, resulting in a 4-year survival of 22% from the time metastatic disease is diagnosed. Multivariate analysis showed that survival correlated with female sex, age younger than 60 years, metastases to lung or soft tissue sites only, prolonged interval between initial diagnosis and subsequent metastatic disease, and surgical or intrahepatic treatment of metastatic disease. Two previous studies have reported survival data for metastatic uveal melanoma and both have reported far poorer survival than we observed. Gragoudas et al3 reported 145 patients from the Massachusetts Eye and Ear Infirmary up through 1988. They found a median survival of only 3.7 months with a 1-year survival of only 13%. The poor prognosis of these patients was underscored by the fact that 8% of the patients died within 2 weeks of diagnosis of metastatic uveal melanoma. They too found that survival correlated with a younger age. However, they did not see a correlation between survival and the other variates we found to be significant.
The other report describing the prognosis of metastatic uveal melanoma was from the Collaborative Ocular Melanoma Study (COMS) group, in which 714 patients with metastatic uveal melanoma were described.7 Here again, a poor prognosis was observed, with a 1-year survival of 19% and a two year survival of only 8%. They did not study correlates of survival. Neither of these two prior reports included a Kaplan-Meier analysis of overall survival in the entire cohort, so it is impossible to know if a biphasic survival curve was observed as we report here.
It is unclear why our patients with metastatic uveal melanoma experienced a longer survival than the patients in the previous reports. One potential source of bias in our report is referral bias. Our patients may have been subtly selected simply by the fact that they sought treatment at a tertiary care cancer center. For the two prior reports, many of the patients were treated by community physicians. Another possibility is that in our patient population, the median time from primary uveal melanoma to metastasis was 53 months compared with 29 months in the data set of Gragoudas et al3 and approximately 32 to 42 months in the Collaborative Ocular Melanoma Study data set.7 The relatively short interval between diagnosis and metastatic disease in the data set of Gragoudas et al may explain why they did not find that the interval to metastatic disease was a significant prognostic variable, as we did.
A unique aspect of our report is the description of the site of first metastases. This is important in considering a rational screening strategy. In our cohort of patients, almost 40% presented with a first metastasis outside of the liver, meaning that screening strategies that focus only on the liver are likely to fail. Perhaps this contributed to our inability to detect any survival benefit associated with detection of asymptomatic metastases. Patients who presented with symptoms had similar survival as patients whose metastases were detected before they were symptomatic. We also noted that physical examinations and blood tests were extremely inefficient in detecting asymptomatic metastases. One of the reasons metastatic disease should be detected early would be to increase the chance of detecting surgically resectable disease. Indeed, treatment with surgery did correlate with improved survival in our data set. However, metastatic disease detected before it became symptomatic was not any more likely to be treated with surgery or intrahepatic chemotherapy than metastatic disease discovered because of symptoms.
The survival benefit associated with local treatment of liver disease in the metastatic setting as well as the survival benefit gained by the absence of metastasis to the liver might argue for the use of adjuvant hepatic treatment. Intrahepatic melphalan in the metastatic setting has shown a response rate of 62%,8,9 and could be considered as an adjuvant treatment strategy in patients with sufficiently high risks of developing metastatic disease.
Overall, from our data, there was no survival advantage associated with screening for metastatic disease. This observation is in contrast to the only other prior report examining the association of screening for metastatic disease with survival. Gragoudas et al3 described 145 patients who developed metastatic disease among 1,116 patients treated for primary uveal melanoma who had been screened annually with chest x-ray and liver function tests. Despite screening, a majority of patients (64%) were diagnosed with metastatic disease on the bases of symptoms, a proportion similar to our observations (59%). In contrast to our findings, Gragoudas et al noted that asymptomatic patients detected by routine screening had a significantly longer survival than patients presenting with symptoms, although metastases detected by screening were not discovered significantly sooner than metastases detected due to symptoms. This suggests that screening did not detect metastases earlier but may have simply identified a cohort with biologically less aggressive behavior. This is in contrast to our data in which patients who presented with symptoms as their first sign of metastatic disease had a much longer disease-free interval from the time of initial diagnosis compared with asymptomatic patients discovered to have metastatic disease on routine radiographic screening tests. This suggests that screening did pick up metastatic disease earlier. However, once metastatic disease was noted, survival was not different in these two groups in our data set. This may be because overall survival from metastatic disease was more prolonged in our patients than in the patients described by Gragoudas et al.
Identification of asymptomatic patients with metastatic disease might be expected to result in apparent improvement in survival due to lead time bias. In fact, our univariate analysis showed a survival benefit for asymptomatic patients of 4.7 months, although it was not statistically significant. This suggests that in most patients, metastatic uveal melanoma progresses sufficiently quickly that lead time bias has a small effect and could not be detected with statistical significance in 119 patients.
Our data also indicate that, although liver is the most common site of first metastasis, organs other than the liver are often the site of first metastasis. This suggests that screening tests focusing only on the liver are likely to miss a substantial proportion of patients developing metastatic disease.
Our observations provide evidence that there is a subset of patients with metastatic uveal melanoma who can expect prolonged survival. With this in mind, it is of interest that Onken et al10 recently showed that primary uveal melanomas could be clustered into two distinct molecular classes based on a three-gene expression profile. This molecular classification strongly predicted metastatic death and suggested that there is a subtype of uveal melanoma that is more indolent. In the future, it would be of interest to examine gene expression profiles of metastatic uveal melanomas to see if the subset of long-term survivors can be identified by differential gene expression.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Presented in part at the 41st Annual Meeting of the American Society of Clinical Oncology, May 13-17, 2005, Orlando, FL.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Abramson D, Dunkel I, McCormick B: Neoplasms of the eye, in Bast R Jr, Kufe D, Pollock R, et al (eds): Cancer Medicine (ed 5). Hamilton, Ontario, Canada, BC Decker Inc, 2000
Assessment of metastatic disease status at death in 435 patients with large choroidal melanoma in the Collaborative Ocular Melanoma Study (COMS): COMS report 15. Arch Ophthalmol 119:670-676, 2001
Gragoudas ES, Egan KM, Seddon JM, et al: Survival of patients with metastases from uveal melanoma. Ophthalmology 98:383-390, 1991
Salmon RJ, Levy C, Plancher C, et al: Treatment of liver metastases from uveal melanoma by combined surgery-chemotherapy. Eur J Surg Oncol 24:127-130, 1998
Anderson JR, Cain KC, Gelber RD: Analysis of survival by tumor response. J Clin Oncol 1:710-719, 1983
Kath R, Hayungs J, Bornfeld N, et al: Prognosis and treatment of disseminated uveal melanoma. Cancer 72:2219-2223, 1993
Diener-West M, Reynolds SM, Agugliaro DJ, et al: Screening for metastasis from choroidal melanoma: The Collaborative Ocular Melanoma Study Group Report 23. J Clin Oncol 22:2438-2444, 2004
Alexander HR, Libutti SK, Bartlett DL, et al: A phase I-II study of isolated hepatic perfusion using melphalan with or without tumor necrosis factor for patients with ocular melanoma metastatic to liver. Clin Cancer Res 6:3062-3070, 2000
Alexander HR Jr, Libutti SK, Pingpank JF, et al: Hyperthermic isolated hepatic perfusion using melphalan for patients with ocular melanoma metastatic to liver. Clin Cancer Res 9:6343-6349, 2003
Onken MD, Worley LA, Ehlers JP, et al: Gene expression profiling in uveal melanoma reveals two molecular classes and predicts metastatic death. Cancer Res 64:7205-7209, 2004(Petra Rietschel, Katherin)
ABSTRACT
PURPOSE: The course and outcome of metastatic uveal melanoma are not well described. We evaluated the survival of our patients with metastatic uveal melanoma, described factors that correlated with survival, and evaluated the influence of screening tests on time of detection and survival.
PATIENTS AND METHODS: All patients with metastatic uveal melanoma seen at Memorial Sloan-Kettering Cancer Center between 1994 and 2004 were identified from our database. We recorded date of initial diagnosis, date of metastatic disease, date of last follow-up, site of the first metastasis, how the first metastasis was discovered, treatment, and outcome of therapy.
RESULTS: The estimated median survival of the 119 patients analyzed was 12.5 months; 22% of patients were alive at 4 years. Five variates correlated independently with prolonged survival: Lung/soft tissue as only site of first metastasis, treatment with surgery or intrahepatic therapy, female sex, age younger than 60, and a longer interval from initial diagnosis to metastatic disease. Discovering metastatic disease in asymptomatic patients did not correlate with overall survival; 89% of patients had a single organ as the site of first metastasis. Although liver was the most common site, 39.5% of patients had nonliver sites, most commonly lung, as the first site of metastasis.
CONCLUSION: A substantial subset of patients with metastatic uveal melanoma survive more than 4 years with metastatic disease. Data on variates of survival and site of first metastasis may guide strategies for screening patients, although our data failed to show a survival advantage in discovering asymptomatic metastatic disease.
INTRODUCTION
Uveal melanoma arises from melanocytes located in the choroid, the layer between the sclera and the retina. Although it is the most common primary malignant primary tumor of the eye, there are only about 1,500 diagnoses per year in the United States.1 Despite optimal treatment (surgery or radiation), metastases often develop, although the true incidence of metastases from uveal melanomas has been difficult to determine. Differences in study design, diagnostic procedures, data collection, inaccuracies in biopsies of metastases, and even incorrectly diagnosed ocular tumors have led to disagreement in the literature. All studies agree that long-term prognosis is poor. Two retrospective analyses reported median survival from the time metastases was documented to be 6 months or less.2,3 Aside from surgery,4 treatment has not been shown to improve survival, and it is unclear whether early detection leads to improved survival.
Few studies have looked specifically at patients with metastatic uveal melanoma and studied the characteristics of these patients by analyzing the sites of metastasis, method of discovery of metastatic disease, treatment received by patients with metastatic disease, overall survival of patients presenting with metastatic disease, and correlates of survival. We have reviewed the Memorial Sloan-Kettering (New York, NY) experience with patients who had metastatic uveal melanoma seen at our institution over a 10-year period. We found that our patients with metastatic uveal melanoma survived substantially longer than the patients in these prior studies. We describe the natural history of uveal melanoma among our large patient cohort, quantitate survival from time of metastatic disease, describe factors that correlate with long-term survival in patients with metastatic uveal melanoma, and evaluate the value of screening blood tests, x-ray tests, and routine physical examinations on time of detection and survival.
PATIENTS AND METHODS
This is a retrospective analysis of stage IV uveal melanoma patients seen at Memorial Sloan-Kettering Cancer Center. All patients with stage IV uveal melanoma were included and a total of 119 patients (67 women and 52 men) who developed stage IV melanoma between 1994 and 2004 were identified from the melanoma database. The patients initially had been diagnosed with uveal melanoma between 1972 and 2003. Date of initial diagnosis, date of stage IV diagnosis, and date of last follow-up or death were recorded. We also recorded the site of the first metastasis, how the first metastasis was discovered, treatment of stage IV disease, and outcome of therapy. This retrospective analysis was performed with Institutional Review Board approval, which determined that this was exempt research under 45 Code of Federal Regulations 46.101.b.
Statistical Analysis
Time to stage IV was defined as the time from the date of initial diagnosis to the date of stage IV diagnosis. Survival from stage IV diagnosis was defined as the time from date of stage IV diagnosis to the date of death or last follow-up. Overall survival distributions from stage IV diagnosis were estimated using Kaplan-Meier methodology and comparisons between categoric variables were assessed using the log-rank test. A Cox proportional hazards model was used for multivariate analysis.
To determine whether tumor response was associated with improved survival, a landmark analysis was performed in which survival time was defined as the time from 2 months after treatment start (approximately the time of first response assessment) to the date of death or last follow-up. This was done to correct for patients who died from melanoma before they had an opportunity to be assessed for tumor response.5 This eliminated nine patients from the analysis. Without this correction, the comparison of survival between responders and nonresponders could be biased toward showing a benefit in the responder group.
RESULTS
The median age at diagnosis of metastatic melanoma for this cohort of 119 patients was 63.1 years (range, 25.7 to 86.4 years). As of last follow-up, 31 of 119 patients were still alive; the median follow-up of survivors was 17 months (range, 0 to 96 months). Estimated median time from initial diagnosis of uveal melanoma until detection of metastatic disease (time to stage IV) for all patients was 53 months (range, 0 to 359 months). Because our information was taken mainly from chart notes written at the time of diagnosis of stage IV, we did not have detailed information about size or location of the primary tumor.
Sites of Metastases
Given that information on the site of first metastasis is critical for determining a rational screening procedure and we could find no previous studies reporting the location of the first metastasis, we were interested in identifying the site of first metastasis in our patient population (Table 1). We found that 89% of patients had a single organ involved as the first metastatic site and that in 68% of these patients (60.5% of all patients), the liver was the first metastatic site. Surprisingly, however, almost 40% of all patients presented with nonliver sites as first metastasis; the most common site was the lung. This suggests that screening procedures that examine only the liver would be expected to miss a substantial proportion of patients.
Over the course of their metastatic disease, a total of 77.3% of patients eventually developed liver involvement and 42.9% never had disease beyond the liver. Consistent with autopsy studies,2 we found that a small percentage of patients developed metastases to bone (17%) and brain (10%). Although our data confirm that liver is the most common site of metastasis in uveal melanoma, a significant proportion of patients had nonliver organs as the site of first metastasis.
Methods of Discovery of Metastatic Disease
In our attempt to evaluate the value of screening procedures, we were interested in discovering how the first metastases were discovered in these patients. This information was available in 113 patients (Table 2). More than half of the patients were found to have metastatic disease based on symptom presentation. These patients presented with metastatic disease a median of 79 months from the time of initial diagnosis of primary uveal melanoma. A substantial proportion (37.2%) of patients were discovered to have metastatic disease based on radiographic imaging while they were asymptomatic. This cohort of patients was found to have metastatic disease much earlier (median, 30 months from the time of initial diagnosis) than patients who presented with symptoms (P = .0016). Blood tests and routine physical examinations rarely picked up the first metastasis in asymptomatic patients. These findings suggest that routine radiographic screening led to early discovery of metastatic uveal melanoma, but that physical examinations and blood tests in asymptomatic patients were of little value.
Treatment Received for Metastatic Uveal Melanoma
Data on treatment were available for 93 patients; outcome data were available for 88 of 119 patients. Of 19 patients (20%) who underwent surgical resection, 10 patients had complete surgical resections. Of 18 patients (19%) who received intrahepatic chemotherapy, one patient had a complete response to the therapy. Systemic therapy was given to 41% of patients (chemotherapy, immunotherapy, or other investigational treatments) and 19% received no treatment for their metastatic disease. Few of the patients were treated within the context of a clinical trial, and although partial responses and stable disease were reported by some of the treating physicians, we could not formally assess these responses. Complete responses, although rare, were only achieved with local therapy, such as surgery or intrahepatic chemotherapy.
Overall Survival
The survival curve showed an estimated median overall survival of 12.5 months (95% CI, 9.5 to 15.7 months; Fig 1). A significant minority of patients appears to have had a more indolent course, indicated by the observation that after 18 months, the slope of the survival curve flattens out. As a result, an estimated 22% of patients were alive at 4 years. This suggests that metastatic uveal melanoma consists of at least two distinct subpopulations with different tumor biology.
Univariate Analysis of Survival
To identify characteristics that correlated with survival, we tested a set of variates in a univariate analysis (Table 3). We found that age younger than 60 years, female sex, first metastases involving lung or soft tissue, a longer time interval from initial diagnosis to development of metastatic disease, and a history of having undergone surgery or intrahepatic chemotherapy all correlated with prolonged survival. Patients who received surgery or intrahepatic chemotherapy (local treatment) had a particularly improved survival (median, 32.4 months), compared with patients who received systemic chemotherapy (median survival, 9.5 months). This is presumably related to patient selection; patients who were candidates for surgery or intrahepatic chemotherapy would have had more localized metastatic disease.
We also noted that patients who had had a complete or partial response to chemotherapy or surgery had a much longer estimated median survival (49.1 months) compared with treated patients who did not have a response (9.5 months). Although this difference was highly statistically significant, we did not include tumor response in the multivariate analysis because tumor responses were not assessed formally in this patient set.
We were surprised to note that neither the method of discovery of metastatic uveal melanoma (symptoms v routine screening) nor the person who discovered the metastasis (patient v physician) affected survival significantly.
Multivariate Analysis of Survival
We tested these variates in a multivariate analysis and found five variates that correlated independently with prolonged survival (Table 4). Lung or soft tissue as only site of first metastasis correlated with an improved survival, as did treatment with surgery or intrahepatic therapy, female sex, age younger than 60, and a longer time interval from diagnosis to metastatic uveal melanoma. Whether metastases were first discovered by screening tests (computed tomography scans, blood tests, or routine physical examination) or as a result of symptoms did not correlate with the likelihood of having surgery or with overall survival.
DISCUSSION
Metastatic uveal melanoma is a substantial problem, but relatively little has been published characterizing the prognosis and variates that correlate with survival. Except for a small single-institution study,6 no one has looked carefully at correlates of survival in metastatic uveal melanoma. We have described our single-institution experience in 119 patients with metastatic uveal melanoma. We observed that the estimated median overall survival was 12 months, but that a substantial proportion of patients demonstrated a more indolent course, resulting in a 4-year survival of 22% from the time metastatic disease is diagnosed. Multivariate analysis showed that survival correlated with female sex, age younger than 60 years, metastases to lung or soft tissue sites only, prolonged interval between initial diagnosis and subsequent metastatic disease, and surgical or intrahepatic treatment of metastatic disease. Two previous studies have reported survival data for metastatic uveal melanoma and both have reported far poorer survival than we observed. Gragoudas et al3 reported 145 patients from the Massachusetts Eye and Ear Infirmary up through 1988. They found a median survival of only 3.7 months with a 1-year survival of only 13%. The poor prognosis of these patients was underscored by the fact that 8% of the patients died within 2 weeks of diagnosis of metastatic uveal melanoma. They too found that survival correlated with a younger age. However, they did not see a correlation between survival and the other variates we found to be significant.
The other report describing the prognosis of metastatic uveal melanoma was from the Collaborative Ocular Melanoma Study (COMS) group, in which 714 patients with metastatic uveal melanoma were described.7 Here again, a poor prognosis was observed, with a 1-year survival of 19% and a two year survival of only 8%. They did not study correlates of survival. Neither of these two prior reports included a Kaplan-Meier analysis of overall survival in the entire cohort, so it is impossible to know if a biphasic survival curve was observed as we report here.
It is unclear why our patients with metastatic uveal melanoma experienced a longer survival than the patients in the previous reports. One potential source of bias in our report is referral bias. Our patients may have been subtly selected simply by the fact that they sought treatment at a tertiary care cancer center. For the two prior reports, many of the patients were treated by community physicians. Another possibility is that in our patient population, the median time from primary uveal melanoma to metastasis was 53 months compared with 29 months in the data set of Gragoudas et al3 and approximately 32 to 42 months in the Collaborative Ocular Melanoma Study data set.7 The relatively short interval between diagnosis and metastatic disease in the data set of Gragoudas et al may explain why they did not find that the interval to metastatic disease was a significant prognostic variable, as we did.
A unique aspect of our report is the description of the site of first metastases. This is important in considering a rational screening strategy. In our cohort of patients, almost 40% presented with a first metastasis outside of the liver, meaning that screening strategies that focus only on the liver are likely to fail. Perhaps this contributed to our inability to detect any survival benefit associated with detection of asymptomatic metastases. Patients who presented with symptoms had similar survival as patients whose metastases were detected before they were symptomatic. We also noted that physical examinations and blood tests were extremely inefficient in detecting asymptomatic metastases. One of the reasons metastatic disease should be detected early would be to increase the chance of detecting surgically resectable disease. Indeed, treatment with surgery did correlate with improved survival in our data set. However, metastatic disease detected before it became symptomatic was not any more likely to be treated with surgery or intrahepatic chemotherapy than metastatic disease discovered because of symptoms.
The survival benefit associated with local treatment of liver disease in the metastatic setting as well as the survival benefit gained by the absence of metastasis to the liver might argue for the use of adjuvant hepatic treatment. Intrahepatic melphalan in the metastatic setting has shown a response rate of 62%,8,9 and could be considered as an adjuvant treatment strategy in patients with sufficiently high risks of developing metastatic disease.
Overall, from our data, there was no survival advantage associated with screening for metastatic disease. This observation is in contrast to the only other prior report examining the association of screening for metastatic disease with survival. Gragoudas et al3 described 145 patients who developed metastatic disease among 1,116 patients treated for primary uveal melanoma who had been screened annually with chest x-ray and liver function tests. Despite screening, a majority of patients (64%) were diagnosed with metastatic disease on the bases of symptoms, a proportion similar to our observations (59%). In contrast to our findings, Gragoudas et al noted that asymptomatic patients detected by routine screening had a significantly longer survival than patients presenting with symptoms, although metastases detected by screening were not discovered significantly sooner than metastases detected due to symptoms. This suggests that screening did not detect metastases earlier but may have simply identified a cohort with biologically less aggressive behavior. This is in contrast to our data in which patients who presented with symptoms as their first sign of metastatic disease had a much longer disease-free interval from the time of initial diagnosis compared with asymptomatic patients discovered to have metastatic disease on routine radiographic screening tests. This suggests that screening did pick up metastatic disease earlier. However, once metastatic disease was noted, survival was not different in these two groups in our data set. This may be because overall survival from metastatic disease was more prolonged in our patients than in the patients described by Gragoudas et al.
Identification of asymptomatic patients with metastatic disease might be expected to result in apparent improvement in survival due to lead time bias. In fact, our univariate analysis showed a survival benefit for asymptomatic patients of 4.7 months, although it was not statistically significant. This suggests that in most patients, metastatic uveal melanoma progresses sufficiently quickly that lead time bias has a small effect and could not be detected with statistical significance in 119 patients.
Our data also indicate that, although liver is the most common site of first metastasis, organs other than the liver are often the site of first metastasis. This suggests that screening tests focusing only on the liver are likely to miss a substantial proportion of patients developing metastatic disease.
Our observations provide evidence that there is a subset of patients with metastatic uveal melanoma who can expect prolonged survival. With this in mind, it is of interest that Onken et al10 recently showed that primary uveal melanomas could be clustered into two distinct molecular classes based on a three-gene expression profile. This molecular classification strongly predicted metastatic death and suggested that there is a subtype of uveal melanoma that is more indolent. In the future, it would be of interest to examine gene expression profiles of metastatic uveal melanomas to see if the subset of long-term survivors can be identified by differential gene expression.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Presented in part at the 41st Annual Meeting of the American Society of Clinical Oncology, May 13-17, 2005, Orlando, FL.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Abramson D, Dunkel I, McCormick B: Neoplasms of the eye, in Bast R Jr, Kufe D, Pollock R, et al (eds): Cancer Medicine (ed 5). Hamilton, Ontario, Canada, BC Decker Inc, 2000
Assessment of metastatic disease status at death in 435 patients with large choroidal melanoma in the Collaborative Ocular Melanoma Study (COMS): COMS report 15. Arch Ophthalmol 119:670-676, 2001
Gragoudas ES, Egan KM, Seddon JM, et al: Survival of patients with metastases from uveal melanoma. Ophthalmology 98:383-390, 1991
Salmon RJ, Levy C, Plancher C, et al: Treatment of liver metastases from uveal melanoma by combined surgery-chemotherapy. Eur J Surg Oncol 24:127-130, 1998
Anderson JR, Cain KC, Gelber RD: Analysis of survival by tumor response. J Clin Oncol 1:710-719, 1983
Kath R, Hayungs J, Bornfeld N, et al: Prognosis and treatment of disseminated uveal melanoma. Cancer 72:2219-2223, 1993
Diener-West M, Reynolds SM, Agugliaro DJ, et al: Screening for metastasis from choroidal melanoma: The Collaborative Ocular Melanoma Study Group Report 23. J Clin Oncol 22:2438-2444, 2004
Alexander HR, Libutti SK, Bartlett DL, et al: A phase I-II study of isolated hepatic perfusion using melphalan with or without tumor necrosis factor for patients with ocular melanoma metastatic to liver. Clin Cancer Res 6:3062-3070, 2000
Alexander HR Jr, Libutti SK, Pingpank JF, et al: Hyperthermic isolated hepatic perfusion using melphalan for patients with ocular melanoma metastatic to liver. Clin Cancer Res 9:6343-6349, 2003
Onken MD, Worley LA, Ehlers JP, et al: Gene expression profiling in uveal melanoma reveals two molecular classes and predicts metastatic death. Cancer Res 64:7205-7209, 2004(Petra Rietschel, Katherin)