The Management of Lesions to Exclude Melanoma schema may be a clinically useful and accurate classification system for melanocytic lesions

by Jake Engel, Michael Pratte

Originally published by 2 Minute Medicine® (view original article). Reused on AccessMedicine with permission.

1. Pathologists demonstrated high accuracy in diagnosing melanocytic lesions using the Management of Lesions to Exclude Melanoma (MOLEM) classification system on high-risk patients

2. Furthermore, the numerical classification of the MOLEM system may be more easily integrated into artificial intelligence algorithms compared to other qualitative and language-based schemas.

Evidence Rating Level: 2 (Good)

Study Rundown:

Melanoma is diagnosed with a morphological assessment of a biopsied lesion. There appears to be a discordance between pathologists in diagnosing melanoma from other lesions called “borderline lesions.” The Management of Lesions to Exclude Melanoma (MOLEM) schema is separated into five classes (I-V) and used to differentiate melanocytic from nonmelanocytic lesions. This classification system may be utilized to improve diagnoses and potentially used in the future as algorithms to train artificial intelligence. This cohort study investigated the accuracy and confidence of pathologists in diagnosing melanoma in high-risk patients using the MOLEM classification system. Study patients were from a primary care skin clinic in New South Wales, Australia, from April 2019 until December 2019. Lesions were reviewed by a minimum of five and up to nine independent pathologists; they reported their MOLEM class and confidence rating (1 [most] – 5 [least]) in their diagnoses. The reference standard was defined as the most frequent MOLEM class for each case. The outcomes were interrater agreement and diagnostic confidence of excised lesions by the pathologists. One hundred ninety-seven high-risk patients for suspected melanoma were included in the study, with a total of 1516 histological diagnoses from 217 lesions. The overall quadratic weighted interrater agreement for MOLEM class ratings was 91.3% (Gwet AC1 coefficient: 0.76 [95% CI: 0.72-0.81]) with the lowest agreement in class II of 82.1% (Gwet AC1 coefficient: 0.55 [95% CI: 0.44-0.66]) and the highest agreement in class V of 96.6% (Gwet AC1 coefficient: 0.94 [95%CI: 0.88-1.00]). The overall quadratic weighted interrater agreement for confidence ratings was 95.2% (Gwet AC1 coefficient: 0.92 [95% CI: 0.90-0.94]), with the lowest agreement in class II of 93.5% (Gwet AC1 coefficient: 0.85 [95% CI: 0.79-0.91]) and the highest agreement in class V of 97.5% (Gwet AC1 coefficient: 0.97 [95% CI: 0.93-1.00]). This study demonstrated that the MOLEM classification is clinically useful with a high level of agreement between pathologists; in cases that were more in doubt, pathologists were aware of the difficulty and admitted to diagnosing with less confidence. Notably, it is important to consider that the methodology was limited by selection bias based on the included cohort and by observation bias as the pathologists’ subjective criteria and cognitive biases may have played a role in their diagnoses.

©2021 2 Minute Medicine, Inc. All rights reserved. No works may be reproduced without expressed written consent from 2 Minute Medicine, Inc. Inquire about licensing here. No article should be construed as medical advice and is not intended as such by the authors or by 2 Minute Medicine, Inc.