Six hundred seventy-six patients with ductal carcinoma in situ of the breast (DCIS) from 1971 to 1995 were included in the study. Computerized patient files were retrospectively analyzed. Clinical findings were less frequently reported to reveal DCIS after 1989. Positive mammographic findings were obtained in 87% of patients and were mainly represented by microcalcifications (79.4%). Treatment procedures were breast-conserving surgery (BCS) alone (37.5%), BCS followed by radiation (BCSR) (25.5%), or mastectomy (M) (37%). The actuarial local recurrence was 2.6% in the M group (94 months of follow-up), 14.5% in the BCS group (85,7 months of follow-up), and 7.5% in the BCSR group (78.8 months of follow-up). Predictive factors of recurrence in all patients were invaded margin status and age. In the BCS group, grade was also a predictive factor. The analysis per decade shows that the lesions currently diagnosed are less serious than those of the past. All the recurrence in patients with positive margins was in the same quadrant as the original lesion. This further emphasizes the need for clear margins.

#3003 Background: The Scarff Bloom and Richardson (SBR) histologic grade, which assesses tumor differentiation, nuclear pleomorphism and tumor proliferation through a mitotic score, is known to be an independent prognostic factor in breast cancer. Ki-67 is a nuclear protein that is expressed by cells that are engaged in the cell cycle and may be a more accurate proliferation marker than the mitotic score used in SBR grade. The aim of our study was to look at the prognostic value of ki-67 (detected with Mib1 antibody) in breast cancer and if it could improve the prognostic value of SBR grade (SBR G).
 Materials and Methods: We analyzed ki-67 expression by immunohistochemistry using the Mib1 antibody on tissue microarrays comprising four 0.6mm diameter tissue cores of 1033 consecutive invasive ductal carcinomas (290 SBR G1, 470 SBR G2, 273 SBR G3) operated on between 1989 and 1993 (median follow-up: 167 months, 280 metastatic events). Using the median and 75th percentile values of Mib1 expression in the series (10 % and 20%, respectively) as cut offs, we defined a three tiered Mib1 score and used it to replace the mitotic score in the SBR G (Mib1 G). The prognostic values of Mib1 score alone, SBR G, Mib1 G and combined SBR/Mib1 G using the higher of the 2 grades to assign tumors to a risk category were assed by univariate analysis using the Log-rank test with the Kaplan Meier method.
 Results: 53/290 (18%) and 53/470 (11%) of SBR G1 and 2 tumors were reclassified in combined SBR/Mib1 G2 and G3, respectively. In the combined SBR/Mib1 G three groups of tumors could be individualized according to their metastatic risk.
 
 Patients with SBR G1/Mib1 G1 tumors had the lowest metastatic risk. Patients with SBR G1/Mib1 G2 tumors, SBR G2/Mib1 G1 tumors and SBR G2/Mib1 G2 tumors had an intermediate risk. Finally, patients with SBR G2/Mib1 G3 tumors, SBR G3/Mib1 G2 and SBR G3/Mib1 G3 tumors had the highest metastatic risk. Mib1 score, Mib1 G, SBR G, and combined SBR/Mib1 G were all strongly associated with Metastasis-free Interval (p=3.10-10, p=8.10-11, p=2.10-12 and p=4.10-13, respectively).
 Discussion: In the combined grade patient assignment to risk category (low, intermediate or high) is made on the higher of the two grades. Metastatic risk was more accurately estimated when SBR and Mib1 grades were combined compared to SBR G or Mib1 G separately. Mib1 may complement the prognostic information provided by histologic grade. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 3003.