<scp>PRAME</scp> immunohistochemistry for melanoma diagnosis: A <scp>STARD</scp>‐compliant diagnostic accuracy study
Michaela K. O’Connor, Hongyan Dai, Garth Fraga
Abstract
BACKGROUND: Preferentially expressed antigen in melanoma (PRAME) is a tumor-associated antigen that is frequently expressed in cutaneous melanoma and can be evaluated by immunohistochemistry. Earlier studies on PRAME utilized case-control study designs that may misestimate diagnostic accuracy and lack generalizability. METHODS: Using retrospective cohort selection, a cross-sectional study of diagnostic accuracy of PRAME was conducted according to standards for reporting diagnostic accuracy studies requirements. RESULTS: Mean PRAME positive fraction was higher in 42 malignant melanocytic lesions than 101 benign melanocytic lesions (0.71 ± 0.30 vs. 0.13 ± 0.20, p < 0.01). Receiver operating characteristic curve showed the test was effective (area under the curve = 0.90). Global PRAME 4+ scores (>75%) were associated with sensitivity of 0.63, specificity of 0.97, accuracy of 0.87, and excellent interrater concordance (Kappa = 0.83). Lower cutoffs for PRAME of 2+ (>25%) and 3+ (>50%) produced higher joint sensitivity/specificity (Youden index) than PRAME 4+, but lower accuracy. CONCLUSION: PRAME as it is used in clinical practice is an effective test for melanoma. PRAME is best used as an ordinal variable to calculate the posttest probability of melanoma. PRAME ≤25% (0/1+) favors nevus, PRAME 26%-75% (2/3+) is noncontributory, and PRAME >75% (4+) favors melanoma.