Grading of Head-Neck Squamous Dysplasia – A Multi-Insitutional Study
H Mani, M Richardson, R Seethala, J Lewis, J Hunt, E Schaefer, H Crist. PSMSHMC, Hershey; MUSC, Charleston; UPMC, Pittsburgh; WUSTL, St Louis; UAMS, Little Rock
Background: Grading of squamous dysplasia on head-neck mucosal biopsies plays a central role in management, but is subjective. None of the limited studies on observer variation in grading has included community pathologists. This study compares observer variation in grading between subspecialized head-neck pathologists (HNP) and general surgical pathologists (GSP). This is the first study employing digital imaging as a media for such evaluation at this site.
Design: Aperio scanned images of a random selection of 50 head-neck mucosal biopsies were circulated to 7 HNP (4 surgical and 3 oral) and 7 GSP (3 academic and 4 community). Participants were asked to assign 3-tier (negative, mild, moderate, severe) and 2-tier (negative, low, high) grades on each case. Results were analyzed for observer agreement using Krippendorf alpha (KA), a measure more appropriate than kappa statistics for ordinal data. 5 cases rated suboptimal were excluded.
Results: Percent cases graded as no/mild/moderate/severe dysplasia were 18/18/19/45 by HNP and 20/26/18/36 by GSP. HNP graded 18/27/55 cases as no/low/high grade dysplasia compared to 21/33/46 by GSP. Severe/high grades were assigned more often by oral HNP (54/61%) than by surgical HNP (38/51%), academic GSP (46/54%) and community GSP (28/40%). HNP made a diagnosis of moderate dysplasia in 61 instances, of which 29 (48%) were classified as low and 32 (52%) as high grade. GSP made a diagnosis of moderate dysplasia in 56 instances, of which 23 (41%) were classified as low and 33 (59%) as high grade. On 3-tier grading, agreement was only moderate among all pathologists (KA 0.56), and was better among GSP (0.62) than HNP (0.56). 2-tier grading resulted in less agreement with KA values being 0.48 (all pathologists), 0.47 (HNP) and 0.54 (GSP). Agreement did not improve even when restricting analysis to a single site (true cord, 23 cases), with KAs for 3-tier/2-tier being 0.54/0.48 (all pathologists), 0.52/0.47 (HNP) and 0.58/0.54 (GSP). All 7 HNP agreed on a grade in only 9 cases, while 6/7 HNP agreed on a grade in 15 cases. 6/7 GSP gave the same grade as HNP in only 11 of these 15 (73%) cases.
Conclusions: Overall agreement in grading dysplasia is only moderate at best. Oral pathologists tend to assign higher dysplasia grades, while community pathologists tend to assign lower grades. HNP divide moderate dysplasias equally between high grade and low grade. A better defined and more objective classification system is likely required to improve reproducibility among all pathologists.
Category: Head & Neck
Monday, March 19, 2012 2:30 PM
Platform Session: Section F, Monday Afternoon