Abstract

Oral melanoma is a common canine cancer with a historically poor prognosis. Recent evidence suggests that a subset of cases may have a more favorable outcome, defined as long-term survival in the absence of intervention other than initial surgery. Traditional histological parameters have had prognostic significance in some studies but not in others, potentially due to inter-observer variation. We evaluated the prognostic utility of Ki67 immunohistochemistry in a group of 79 canine oral melanomas using a technique easily applied in a veterinary diagnostic laboratory. A threshold Ki67 value of>19.5 had a sensitivity and spe-cificity of 87.1 % and 85.4%, respectively, at predicting death or euthanasia due to melanoma by 1 year postdiagnosis. Threshold values for classical histological parameters were also identified for most cases and were>4 (>30%; sensitivity 83.9%, speci-ficity 86.0%) for nuclear atypia and>4/10 hpfs (sensitivity 90.3%, specificity 84.4%) for mitotic index. In this study, the percentages correctly classified with respect to death by 1 year postdiagnosis were comparable for Ki67 (86.1%, 68/79), nuclear atypia (86.3%, 63/73), and mitotic index (86.8%, 66/76). High pigment (>50%) had a high negative predictive value of 90.9 % (18/20), but overall, only 61.0 % (47/77) of cases could be correctly classified by this parameter. Based on these results, we recommend a panel of prognostic parameters, including nuclear atypia, mitotic index, Ki67, and pigment to more accurately predict the likely outcome of canine oral melanomas. Keywords