Original ArticlesNew Criteria for Histologic Grading of Colorectal CancerUeno, Hideki MD, PhD*; Kajiwara, Yoshiki MD, PhD*; Shimazaki, Hideyuki MD, PhD†; Shinto, Eiji MD, PhD*; Hashiguchi, Yojiro MD, PhD*; Nakanishi, Kuniaki MD, PhD†; Maekawa, Kazunari MD†; Katsurada, Yuka MD†; Nakamura, Takahiro PhD‡; Mochizuki, Hidetaka MD, PhD*; Yamamoto, Junji MD, PhD*; Hase, Kazuo MD, PhD*Author Information *Department of Surgery ‡Laboratory for Mathematics, National Defense Medical College †Department of Laboratory Medicine, National Defense Medical College Hospital, Japan Conflicts of Interest and Source of Funding: The authors have disclosed that they have no significant relationships with, or financial interest in, any commercial companies pertaining to this article. Correspondence: Hideki Ueno, MD, PhD, Department of Surgery, National Defense Medical College, 3-2, Namiki, Tokorozawa, Saitama 359-8513, Japan (e-mail: email@example.com). The American Journal of Surgical Pathology: February 2012 - Volume 36 - Issue 2 - p 193-201 doi: 10.1097/PAS.0b013e318235edee Buy SDC Metrics Abstract Conventional tumor grading systems based on the degree of tumor differentiation may not always be optimal because of difficulty in objective assessment and insufficient prognostic value for decision making in colorectal cancer (CRC) treatment. This study aimed to determine the importance of assessing the number of poorly differentiated clusters as the primary criterion for histologic grading of CRC. Five hundred consecutive patients with curatively resected stage II and III CRCs (2000 to 2005) were pathologically reviewed. Cancer clusters of ≥5 cancer cells and lacking a gland-like structure were counted under a ×20 objective lens in a field containing the highest number of clusters. Tumors with <5, 5 to 9, and ≥10 clusters were classified as grade (G)1, G2, and G3, respectively (n=156, 198, and 146 tumors, respectively). Five-year disease-free survival rates were 96%, 85%, and 59% for G1, G2, and G3, respectively (P<0.0001). Poorly differentiated clusters affected survival outcome independent of T and N stages and could help in more effective stratification of patients by survival outcome compared with tumor staging (Akaike information criterion, 1086.7 vs. 1117.0; Harrell concordance index, 0.73 vs. 0.67). The poorly differentiated cluster-based grading system showed a higher weighted κ coefficient for interobserver variability (5 observers) compared with conventional grading systems (mean, 0.66 vs. 0.52; range, 0.55 to 0.73 vs. 0.39 to 0.68). Our novel histologic grading system is expected to be less subjective and more informative for prognostic prediction compared with conventional tumor grading systems and TNM staging. It could be valuable in determining individualized postoperative CRC treatment. © 2012 Lippincott Williams & Wilkins, Inc.