Apart from which mechanisms regulate VEGF-A in UM, several studies have observed VEGF expression to correlate with development of experimental metastasis.
30–32 On the contrary, Sheidow et al.
33 found no correlation between VEGF immunoreactivity in UM samples of enucleated eyes and the occurrence of metastatic disease. We compared VEGF-A mRNA expression in 27 UM samples with clinical and prognostic data for metastatic disease and found no relation (
Table 3). Additional
t-test and survival analysis of the data demonstrated no association between VEGF-A expression and formation of metastasis. Because of the small number of samples used and, in some cases, the short follow-up, these findings may not be conclusive. That the primary tumor samples showed a great variation in VEGF-A mRNA expression, may be due to sampling bias, as some tumor samples may have originated from ischemic regions and some of the samples from fully oxygenated areas. It has been demonstrated that analysis of UM paraffin-embedded sections showed genetic heterogeneity.
34 Determination of VEGF-A mRNA expression in a sample obtained from primary UM may therefore not be representative of the whole tumor, due to this heterogeneous distribution. To avoid such a sampling bias, examination of the amount of VEGF-A protein in sera of UM patients is an alternative. In several tumors—colon carcinoma, soft tissue sarcomas, and gastric cancer—serum VEGF-A levels have been found to be a marker of disease stage and an indicator of metastasis.
35–37 Until now, lactate dehydrogenase (LDH) and alkaline phosphatase (AP) have been the most indicative serum markers for metastatic disease in UM, in combination with liver ultrasonography.
38,39 Elevated serum osteopontin, melanoma-inhibitory activity (MIA), and S-100β levels showed a correlation with metastatic UM to the liver in some studies.
40,41 However, serum markers that indicate micrometastases at an early stage would be clinically preferable. When we indeed determined the amount of VEGF-A in the sera of UM patients, we observed that VEGF-A levels were not increased at the time of enucleation and that only the patients with manifest metastases had high VEGF-A serum levels (
Fig. 7). A similar study was recently presented with the same results by Barak et al.
42 Thus, although we found no correlation between VEGF-A expression by the primary tumor and histologic parameters, VEGF-A levels were significantly higher in patients with metastatic disease. It may be that the size of the hypoxic areas (the small eye versus the large liver) influences the VEGF-A levels in the blood.