Inter-rater concordance of basal cell carcinoma subtypes: influences on reporting format and opportunities for further classification modifications

Abstract

Diagnosis of basal cell carcinoma (BCC) higher risk subtypes influences management strategies because of their propensity to recur locally. Subtyping is prone to inter-observer variability, and subtyping definitions are inconsistently applied. This study sought to compare the interobserver reproducibility of individual BCC subtypes using the 4th edition World Health Organization (WHO) Classification of Skin Tumours (CoST) definitions, with classification into lower and higher risk histological subtype groups. Ninety-one BCC cases were rated by seven pathologists, noting the presence of BCC subtype(s), and providing a higher or lower risk subtype grouping per case. Raters were provided with definitions as per the 4th edition WHO CoST for 10 listed BCC subtypes. Surgical specimen type was noted. Subgroup analysis was performed to exclude cases when the tumour deep front was not well visualised, or there was tangential sectioning (n= 6). Light's kappa was used to assess inter-rater reliability. From the total group (n= 91), five BCC subtypes showed a sufficient number of ratings for computing a κ statistic. From these five subtypes, superficial subtype showed substantial inter-rater agreement (κ= 0.64), and the other four subtypes showed moderate inter-rater agreement [nodular (κ= 0.45), sclerosing/morphoeic (κ= 0.45), infiltrating (κ= 0.49) and micronodular (κ= 0.57)]. Two-tiered rating into either higher or lower risk subtype showed substantial inter-rater agreement (κ= 0.72). Our results suggest a need to more precisely define BCC subtypes. We suggest reporting BCC subtype using a two-tiered risk grouping, followed by specific subtypes present. Further studies examining the inter-rater reliability of less common BCC subtypes are required.