Choroidal melanomas have an incidence of 5?7 new cases per one million persons per year (^{Shields et al 1992}; ^{Singh et al 2003}). They occur much more commonly in Caucasians than in African Americans (^{Egan et al 1988}). There are no known risk factors other than a family history or ocular melanocytosis (^{Singh et al 1998}).

In the eye, choroidal melanomas cause visual loss. Typical sites of metastases are the liver, the lung and rarely other organs (^{COMS Report No. 15 2001}).

The risk factors for metastasis have traditionally been related to the size of the tumor and the cell type (^{McLean et al 1983}; ^{COMS Report No. 18 2001}). Another and very important risk factor for metastasis is monosomy 3, although the mechanism of how this is related to increased metastases is still not known (^{Prescher et al 1996}).

In addition, PAS positive loops are an independent risk factor for the development of metastasis (^{Rummelt et al 1995}). These loops have been shown to be vascular loops made of a combination of endothelial cells and tumor cells. An increased number of vascular loops correlate with a worse prognosis.

Aside from the size and location, or cytogenic abnormality (monosomy 3 status) and gene expression profiling information derived from fine need aspiration, all of the other risk factors are usually known only at the time of the enucleation of the eye following examination of the specimen (^{Gunduz et al 1999}; ^{COMS Report No. 18 2001}).

Since vascularity may be an important and independent indicator for prognosis, methods to determine the amount of vascularity short of removal of the eye are needed. ^{Mueller and associates (2002)} have attempted to look at the vascularity by using indocyanine green (ICG). ^{Buerk and colleagues (2004)} have attempted to evaluate intratumoral vascularity with magnetic resonance imaging (MRI) by determining the relative enhancement over time. The problem with this technique is that it is only a relative technique and does not allow absolute quantitation. Additionally, the temporal sampling of the enhancing signal was only approximately 40 seconds using the MRI dynamic enhancement protocol. We elected this approach to match the technique previously used by ^{Buerk and colleagues (2004)} to allow comparison of the computed tomography (CT) perfusion results with previously studied methods. It is important to note, however, MRI perfusion techniques can sample at 1?1.5 images per second, similar to CT.

Once metastasis occurs, the death rate is 80% at one year and 92% at two years (^{Bedikian et al 1995}). Since the uveal melanomas express vascular endothelial growth factor (VEGF), which helps the tumor attract vessels to help nourish the tumor, attempts at the use of VEGF inhibitors are being considered as well (^{Lee et al 2006}).

The use of CT to quantify vascular perfusion within tumors and organs has only recently been developed (^{Dugdale et al 1999}; ^{Seppenwoolde et al 2000}; ^{Fournier et al 2004}; ^{Kiessling et al 2004}; ^{Kan and Kobayashi et al 2005}; ^{Kan and Phongkitkarun et al 2005}; ^{Gandhi et al 2006}; ^{Laghi 2006}). The benefits of CT perfusion include the ability to measure absolute perfusion values and high temporal resolution (measurements can be acquired every 1 sec, or less). Dependent on the acquisition protocol and analysis software various quantitative perfusion indices may be calculated. The goal of this study was to develop an in vivo measure of vascularity for patients presenting with choroidal melanomas, with the ultimate hope being that this will provide improved prognostic information without removal of the affected eye. We report for the first time, in a patient with a large choroidal melanoma, the results of a 64-slice CT perfusion study and compare it to the relative enhancement from MRI and histopathologic findings.

The results of the CT perfusion study were based on tumor voxel time-attenuation curves (TAC) such as the one shown in the lower half of Figure 1. TAC curves are readily measured with a CT scanner and reflect the change in X-ray attenuation (CT number, HU) as the injected iodinated contrast material flows through the tumor. The corresponding Patlak analysis is shown in the upper half of Figure 1. The average perfusion parameters within the choroidal melanoma were determined by region of interest analysis (Figure 2). Blood flow was 118 ml/100 ml/min, blood volume 11.3 ml/100 ml and Ktrans 62 ml/100 ml/min. The resulting permeability surface area product was 48 ml/100 ml/min.

The MRI enhancement showed a maximum enhancement of 111% over baseline levels at 150 sec. The slope was 2.37 MR enhancement units/sec (Figure 3a). There was a slow wash-out of 0.66 units/sec. There appeared to be a correlation between the areas of increased perfusion by CT and increased enhancement by MR (Figures 2, 3b).

Histology showed the presence of a large spindle cell melanoma with no sign of extraocular extension (Figures 4a, 4b). The neoplasm was mitotically active (13 units/40 hpf). Vascular loops and arches were present throughout the tumor. There was no tumor necrosis or tumor infiltration of lymphocytes. The overlying retina was atrophic, and there was a localized exudative detachment.

Present techniques for the diagnosis of choroidal melanoma do not allow determination of vascular flow. Considering that both brachytherapy and possibly anti-angiogenic therapy affect the vascularity of the tumor, it is important to develop methods that can determine the perfusion characteristics of choroidal melanomas.

Relative MRI enhancement and absolute CT perfusion imaging studies may allow the noninvasive evaluation of the vascularity of tumors. These techniques have been used to look at tumors in other organs (^{Seppenwoolde et al 2000}; ^{Fournier et al 2004}; ^{Laghi 2006}). This is the first study of CT perfusion studies in ocular tumors. One advantage of using CT compared to MRI is better spatial resolution (0.29 ? 0.29 ? 2.4 mm³ voxel size for CT and 0.63 ? 0.83 ? 3.0 mm³ voxel size for MR). It would be very challenging from a technical standpoint to perform MRI enhancement studies of the orbit with such high spatial resolution. Furthermore, the temporal resolution of CT (we used 1 second intervals between measurements although this could be as short as 0.33 seconds) is superior to that of MRI perfusion techniques which has best temporal resolution of approximately 1 second. The potential drawback of using CT is that the patient is exposed to ionizing radiation, which has been shown to induce lens opacities at doses above 500 mGy and cataracts at doses above 5000 mGy (^{International Commission on Radiological Protection 1991}). For this study, we sought to minimize dose to the unaffected eye by tilting the head and using a superficial absorber. With this technique, the dose to the surface of the unaffected eye was 120 mGy, which is well below the thresholds for induction of lens opacities or cataracts. Thus, CT offers advantages over MRI in spatial and temporal resolution with negligible risk of radiation injury to the unaffected eye. Furthermore, the CT technique can provide absolute quantitative measurements of perfusion, whereas the MRI technique provides relative values that are not able to be compared between patients. This lack of an absolute measure limits the correlation of histology findings to tumor perfusion for prognostic purposes.

It is important to note, however, that the dynamic MR enhancement technique used in this study gives different and complementary information about the tumor compared with the CT perfusion technique. The CT perfusion technique used in this study relies on intravascular indicator dilution analysis and two-compartment modeling, which may be used to characterize blood flow, blood volume and vascular permeability of the tumor. The MRI dynamic enhancement technique provides an estimate of the degree of endovascular permeability of the intrinsic tumor vascularity (^{Mueller et al 2002}). The clinical significance of these measures of tumor vascularity remains to be further elucidated though interestingly, in this patient, there was fast development of metastatic disease. One hypothesis would be that more rapidly growing tumors have higher vascular permeability than that of slower growing. Interestingly, in this case, there was close correlation in the location and degree of increased perfusion in the tumor with the area of increased enhancement (permeability) (Figures 2, 3b).

This particular tumor had a marked number of small complex vessels forming loops and arcs by histological evaluation. CT perfusion studies have already been used as a predictive determinant of whether a lesion in the lung or liver is cancerous (^{Seppenwoolde et al 2000}; ^{Fournier et al 2004}; ^{Laghi 2006}). It also has been used in patients before and after treatment to determine if there was a response to therapy. It also has been used to determine the effect of antiangiogenic agents on tumor blood flow parameters. Considering that VEGF receptors have been noted on iris and choroidal melanomas, future studies to determine if chemotherapeutic agents including antiangiogenic therapies may have an effect on the tumors will require the use of diagnostic methods that help to determine the blood flow parameters to the lesion (^{Lee et al 2006}). In addition, it may be an earlier determinant of metastases since CT perfusion has been reported to have a high sensitivity in predicting malignancy of solitary pulmonary nodules and liver lesions (^{Kiessling et al 2004}; ^{Kan et al 2005}). Furthermore, this technique has been shown to have a potential in assessing the grade, activity and treatment response of lymphoma (^{Dugdale et al 1999}). Perfusion values of liver metastases as measured with CT perfusion have been described to correlate with patient survival (^{Kan et al 2005}), whereas in head-and-neck cancer, CT perfusion appeared to be an independent predictor of local outcome after irradiation (^{Gandhi et al 2006}). Functional CT perfusion techniques have previously been successful in demonstrating decreasing effects of different angiogenesis inhibitors on tumor blood flow parameters (^{Kan and Kobayashi et al 2005}; ^{Kan and Phongkitkarun et al 2005}; ^{Wong et al 2006}). The use of this technique before and after successful radiotherapy may add further data as to the validity of the thought that brachytherapy affects the tumor by ablating its vascularity (^{Kaiserman et al 2004}). Future studies should be performed to compare perfusion with other available techniques, including monosomy 3 testing and ICG studies, as well as to determine the clinical value and cost-effectiveness of each technique.

In summary, we present the first attempt at determining the perfusion characteristics of a large choroidal melanoma using state of the art 64-slice CT perfusion. We estimated tumor permeability using dynamic enhancement MRI. In addition, we show the relationship to histology of the tumor.

The authors would like to thank the following colleagues for their assistance and support: Edward P Lindell, MD, Robert J Witte, MD, John I Lane, MD, J Douglas Cameron, MD, John Woog, MD, Glen Sturchio, MS, Chris Tipka, RT, Theresa A Nielsen, and Kristina M Nunez.