Interobserver variation in the histopathologic reporting of key prognostic parameters, particularly clark level, affects pathologic staging of primary cutaneous melanoma.

BACKGROUND: Accurate reporting of prognostically important histopathologic parameters is critical for optimal management of melanoma patients. Previous studies have shown that assessment of Breslow thickness and ulceration is more reproducible between pathologists than estimation of Clark level. However, detailed quantification of the magnitude of variation between observers and its effect on melanoma staging has not been reported.
METHODS: Nine hundred twelve melanoma patients referred to the Sydney Melanoma Unit (SMU) with pathology reports from referring pathology services and reports following histopathologic review by SMU pathologists were studied. Breslow thickness, Clark level, ulceration, and pathologic staging (American Joint Committee on Cancer, 2002) data were extracted from each matched pair of SMU and non-SMU reports. The degree of concordance for these parameters was assessed.
RESULTS: There was excellent agreement between SMU and non-SMU pathologists in the assessment of Breslow thickness (kappa = 0.883; intraclass correlation coefficient = 0.968) and ulceration (kappa = 0.832), while that for Clark level was only moderate (kappa = 0.627). The degree of concordance in the assessment of overall pathologic stage was excellent (96.2%), but there was poor agreement for the evaluation of substages IB (77.3%) and IIB (72.4%), due largely to poor Clark level concordance.
CONCLUSIONS: Although there was a high level of agreement between pathologists in the assessment of overall pathologic stage, the significant level of discordance in pathologic substaging suggests that accurate and reliable substage-based prognostic prediction may not be possible in all cases. It will be important to consider these issues in future revisions of the American Joint Committee on Cancer melanoma staging system.