Pattern of invasion and lymphovascular invasion in squamous cell carcinoma of the floor of the mouth: an interobserver variability study

• Published in: Histopathology 2016-12, Vol.69 (6), p.914-920
• Main Authors: Beggan, Caitlin, Fives, Cassie, O'Leary, Gerard, Sheahan, Patrick, Heffron, Cynthia C, Feeley, Linda
• Format: Article
• Language: English
• Subjects: Aged; Carcinoma, Squamous Cell - mortality; Carcinoma, Squamous Cell - pathology; Disease-Free Survival; Female; floor of the mouth; Humans; interobserver variability; Kaplan-Meier Estimate; Lymphatic Metastasis - pathology; lymphovascular invasion; Male; Middle Aged; Mouth Neoplasms - mortality; Mouth Neoplasms - pathology; Observer Variation; pattern of invasion; Retrospective Studies; squamous cell carcinoma
• ISSN: 0309-0167; 1365-2559
• DOI: 10.1111/his.13014

Aims Lymphovascular invasion (LVI) and the histological pattern of invasion (POI) at the invasive tumour front have been reported as adverse prognosticators in oral squamous cell carcinoma (SCC). However, assessment of these parameters is hampered by variation in the criteria used for their evaluation. Our objective was to evaluate interobserver variability in the assessment of the POI and LVI in SCC of the floor of the mouth (FOM), and to study the impact of the POI on clinical outcomes by using varying quantitative cut‐offs. Methods and results Fifty‐eight cases of FOM SCC were independently evaluated for the POI and LVI by three pathologists. Interobserver variability was analysed by the use of Fleiss kappa statistics. Interobserver agreement was substantial for the assessment of LVI [κ = 0.64, 95% confidence interval (CI) 0.60–0.68]. Interobserver agreement was moderate for evaluation of the POI with a 50% cut‐off (κ = 0.58, 95% CI 0.54–0.62), a 20% cut‐off (κ = 0.58, 95% CI 0.54–0.62) cut‐off, and worst POI (κ =0 .43, 95% CI 0.39–0.46). A consensus diagnosis of the POI was a significant predictor of locoregional recurrence (LRR), disease‐specific survival (DSS) and overall survival (OS) on univariate analysis when a 50% cut‐off was used (LRR, P = 0.01; DSS, P = 0.01; OS, P = 0.01) and when a 20% cut‐off was used (LRR, P = 0.02; DSS, P = 0.02; OS, P = 0.03), but was not significant when worst POI was used (LRR, P = 0.18; DSS, P = 0.16; OS, P = 0.17). Conclusions Interobserver agreement in the diagnosis of LVI was substantial. The POI at the 50% and 20% cut‐offs is moderately reproducible, and has prognostic value in FOM SCC. Further studies are necessary to establish the optimum quantitative cut‐off for the POI.