Revisiting nuclear grade as a high-risk marker for predicting upgrade in ductal carcinoma in situ

Ductal carcinoma in situ (DCIS) diagnosed by needle biopsy harbors invasive carcinoma in about one-third of cases. Nuclear grade (NG) has long been used to stratify risk, with NG3 regarded as "high risk," particularly for upgrade and local recurrence. However, the extent to which NG contributes to predicting upgrade remains uncertain. We retrospectively analyzed 1015 patients diagnosed as DCIS by needle biopsy at two Japanese institutions (2013-2024) and subsequently treated surgically. Biopsy specimens were evaluated for NG and other pathological factors including Ki67 labeling index. Logistic regression was performed to identify predictors of upgrade. Upgrade to invasive carcinoma was observed in 347 of 1015 surgical specimens (34%). Upgrade rates increased only modestly with NG (NG1, 32.8%; NG2, 34.3%; NG3, 38.8%) and were not significantly associated with upgrade. When NG3 was used as a binary predictor, it showed high specificity (88.9%) but very low sensitivity (13.5%), indicating limited discriminatory ability. Both univariable and multivariable analyses revealed that older age (P = 0.002) and a higher Ki67 labeling index (P < 0.001) were independently associated with upgrade. Ki67 showed the highest area under the curve (0.593; cutoff 12%), although its discriminatory ability remained limited when used alone. Despite its long-standing role as a high-risk marker, NG showed limited value in predicting upgrade from DCIS to invasive carcinoma. These findings suggest that risk stratification relying on NG warrants reconsideration. Ki67 may be a more informative pathological factor than NG.