Track

Clinical Studies

Abstract

We have devised a method of nuclear grading of melanoma, like what was described by McGovern in 1970, that correlates with melanoma progression and other adverse histologic indicators including thickness, mitotic rate, and ulceration. The grade is assigned by subjective assessment of nuclear size, irregularities of nuclear membrane contour, number and size of nucleoli, and chromatid arrangement. Grade 1 nuclei resemble nuclei of nevus cells, Grade 3 nuclei show extremes of nuclear abnormalities and Grade 2 nuclei have abnormalities in between Grade 1 and Grade 3.  We compared interobserver agreement of nuclear grading of melanomas between two dermatopathologists, two dermatology residents, and two medical students. After reviewing training images, each observer reviewed 180, 100x magnification, H&E digital images of melanomas and nevi and assigned nuclear grade to each. The interobserver agreement (Kendall’s W) for overall grade was 0.543 (p<0.001). Among dermatopathologists, coefficient of agreement was 0.715 (p<0.001). Our study describes a novel approach for nuclear grading in melanomas and shows interobserver concordance between dermatopathologists and trainees. Feedback and training tools may increase concordance between observers. Computer assisted nuclear grading and artificial intelligence are also highly likely to support the reproducibility of this histopathologic assessment.  Further study is warranted to evaluate the potential role of nuclear grade in microscopic staging of melanoma.