Grading colorectal adenocarcinoma: interobserver agreement between American Joint Committee on Cancer and World Health Organization criteria and revisiting diagnostic challenges
Dipti M Karamchandani, Maurice B Loughrey, Reetesh K Pai, Christophe Rosty, Kelsey McHugh, Brian Cox, Rish K PaiAim
The current American Joint Committee on Cancer (AJCC8) manual grades colorectal adenocarcinoma (CRC) based on the percentage of gland formation. The World Health Organization 5th ed (WHO5) and 6th ed (WHO6) grade based on the least differentiated component; WHO6 specifies a minimum 0.196 mm 2 area of the least‐differentiated component away from the invasive front to merit a high‐grade diagnosis. We studied the interobserver agreement and challenges among pathologists when grading CRC using the above grading systems.
Methods and results
Five gastrointestinal pathologists reviewed 100 scanned CRC slides and graded each slide per the provided criteria in (a) WHO5, (b) WHO6, (c) AJCC8 and (d) their clinical practice. A post‐survey questionnaire was also provided. The Fleiss multirater kappa among the five reviewers was similar using WHO6 (0.730, P < 0.001) and two‐tiered (low vs. high) AJCC8 (0.731, P < 0.001) and was higher than WHO5 (0.655, P < 0.001) grading criteria; the agreement was substantial for all 3 grading systems. Significantly more tumours were called high‐grade when using WHO6 (average: 34.4 cases) and WHO5 (average: 33.6 cases) as compared to AJCC8 (average: 11.8 cases) grading criteria ( P < 0.01). All reviewers were unanimous in handling certain scenarios (e.g. discounting invasive front tumour buds with WHO grading), but were discordant on certain scenarios (e.g. assessment of stroma‐rich areas).
Conclusions
While inter‐observer agreement improved with WHO6 versus WHO5, the persistent upgrading of tumours using the WHO scheme may affect treatment algorithms. Synchronization of grading criteria with validation in prospective studies with clinical outcomes is crucial.