Abstract
PURPOSE
Multiple independent laboratories have confirmed the histologic observation that some tumor microcirculation patterns (MCPs) in uveal melanomas are associated strongly with death resulting from metastatic disease. Because these patterns are imageable with confocal indocyanine green angiography (ICG), we designed a prospective study to evaluate whether these angiographically detectable MCPs predict time to tumor growth.
DESIGN
Observational case series, prospective, non-randomized.
PARTICIPANTS
Ninety-eight patients with unilateral, small, choroidal melanocytic tumors.
METHODS
The following information and tumor characteristics were recorded for each patient: demographic parameters, best-corrected visual acuity, intraocular pressure, related visual symptoms, location and dimension of tumor, pigmentation, orange pigment, drusen, tumor-associated hemorrhage, subretinal fluid, and confocal ICG angiographically determined microcirculation patterns-silent (avascularity), normal (preexisting normal choroidal vessels within the tumor), straight vessels, parallel without and with cross-linking, arcs without and with branching, loops, and networks.
MAIN OUTCOME MEASURES
Time to growth of the tumor, with growth defined as an increase in the maximal apical tumor height of 0.5 mm measured by standardized A-scan ultrasonography, photographic documentation of an increase of the largest basal diameter of at least 1.5 mm, advancement of one tumor border of at least 0.75 mm, or a combination thereof.
RESULTS
Twenty-eight of the 98 tumors in this study (29%) met the predetermined criteria for tumor growth. The median time to growth was 127 days (range, 51-625 days). The following tumor characteristics were significantly associated with time to tumor growth: flashes (P = 0.0224), orange pigment (P = 0.012), subretinal fluid (P < 0.001), maximum basal tumor diameter at initial examination (P = 0.015), maximum apical tumor height (P < 0.001), parallel with cross-linking MCP (P < 0.001), arcs with branching MCP (P = 0.006), loops (P < 0.001), and networks (P < 0.001). Of these, the angiographic documentation of any of the complex MCPs (parallel with cross-linking, arcs with branching, loops, networks, or a combination thereof) showed the strongest association with the time to tumor growth in a Cox proportional hazard model.
CONCLUSIONS
The characteristics of our patient cohort are comparable by clinical and echographic parameters with cohorts for predicting tumor growth, described previously in the literature. In addition, we detected a novel clinical predictor of tumor growth: the confocal ICG angiographic detection of complex MCPs.
Collapse