Digital quantification of PRAME for distinguishing melanoma from nevi compared to manual assessment

Abstract

Aims

In this proof-of-concept study, we propose a new method for automated digital quantification of PRAME (PReferentially expressed Antigen of MElanoma) as a diagnostic aid to distinguish between benign and malignant melanocytic lesions. The proposed method utilizes immunohistochemical virtual double nuclear staining for PRAME and SOX10 to precisely identify the melanocytic cells of interest, which is combined with digital image analyse to quantify a PRAME-index.

Methods

Our study included 10 compound nevi, 3 halo nevi, and 10 melanomas. Tissue slides were stained with PRAME, scanned, the cover glass removed, stained with SOX10, scanned again, and finally analysed digitally. The digitally quantified PRAME-index was compared with a manual qualitative assessment by a dermatopathologist using the standard PRAME-scoring system.

Results

The digitally quantified PRAME-index showed a sensitivity of 70 % and a specificity of 100 % for separating melanomas from benign lesions. The manual qualitative PRAME-score showed a sensitivity of 60 % and a specificity of 100 %. Comparing the two methods using ROC-analyses, our digital quantitative method (AUC: 0.931, 95 % CI: 0.834;1.00, SD: 0.050) remains on par with the manual qualitative method (AUC: 0.877, 95 % CI: 0.725;1.00, SD: 0.078).

Conclusion

We found our novel digital quantitative method was at least as precise at classifying lesions as benign or malignant as the current manual qualitative assessment. Our method has the advantages of being operator-independent, objective, and replicable. Furthermore, our method can easily be implemented in an already digitalized pathology department. Given the small cohort size, more studies are to be done to validate our findings.