149PD - Prognostic impact of proliferation for resected early stage breast cancer according to histology: Cut-off analysis of Ki67 in 859 patients with pur...

Date 08 October 2016
Event ESMO 2016 Congress
Session Breast cancer, early
Topics Breast Cancer
Presenter Luisa Carbognin
Citation Annals of Oncology (2016) 27 (6): 43-67. 10.1093/annonc/mdw364
Authors L. Carbognin1, I. Sperduti2, M.V. Dieci3, I. Zampiva1, G. Griguolo3, S. Pilotto1, V. Guarneri3, M. Brunelli4, R. Nortilli1, E. Manfrin4, E. Fiorio1, V. Parolin1, E. Filippi1, F. Pellini5, F. Bonetti4, G.P. Pollini5, P.F. Conte3, G. Tortora4, E. Bria1
  • 1Medical Oncology, AOU Integrata Verona "Borgo Roma", 37134 - Verona/IT
  • 2Biostatistics Unit, Istituto Regina Elena, 00144 - Roma/IT
  • 3Department Of Surgery, Oncology And Gastroenterology, University of Padova, Istituto Oncologico Veneto IRCCS, 35128 - Padova/IT
  • 4Pathology And Diagnostics, Azienda Ospedaliera Universitaria Integrata Verona-"Borgo Roma", 37134 - Verona/IT
  • 5Breast Unit, AOU Integrata Verona "Borgo Roma", 37134 - Verona/IT



While the prognostic impact of Ki67 has been extensively investigated in IDC, despite the absence of standardized cut-off, its role in ILC has not been fully validated. Thus, the aim of this analysis was to investigate the prognostic potential of Ki67 in a multi-center series of patients (pts) affected by early stage pure ILC comparing with IDC.


Clinicalpathological data of consecutive pts affected by pure ILC and IDC, referring to 2 institutions, were correlated with overall survival and disease-free survival (OS/DFS) using a Cox model. The maximally selected Log-Rank statistics (MSLRS) analysis was applied to the Ki67 continuous variable to estimate the appropriate cut-off according to histology.


Data from 457 ILC and 402 IDC pts were gathered (median age 61/59 years [yrs]). At a median follow-up of 75 months, 10-yrs OS and DFS for ILC and IDC were 81.7%/83.4%, and 71.4%/76.2%, respectively. The MSLRS analysis identified 4% and 18% as optimal Ki67 cut-offs for OS for ILC and IDC, respectively. At the multivariate analysis Ki67, Performance Status (PS), nodal status (N), and TNM-tumor-size (T-size) were independent predictors for OS in ILC pts. Ki67 highly replicated at the internal cross-validation analysis. For IDC pts, PS, age, estrogen receptor expression and T-size were independent predictors for OS. With regard to DFS, the MSLRS analysis identified 4% and 14% as optimal KI67 cut-offs for ILC and IDC, respectively. PS and N were independent predictors for ILC, while PS, age, grading and T-size were predictors for IDC. Log-rank analysis is shown in the table:

Histotype Ki67 5-yrs OS (%) 10-yrs OS (%) Log-Rank
ILC  ≤ 4% 96.9 89.9 p = 0.008
 > 4% 90.1 77.2
IDC  ≤ 18% 97.4 95.8 p = 0.002
 > 18% 93.6 62.6
5-yrs DFS (%) 10-yrs DFS (%)
ILC  ≤ 4% 88.2 79.4 p = 0.03
 > 4% 81.1 69.2
IDC  ≤ 14% 96.0 87.0 p = 0.002
 > 14% 89.2 61.8


Despite the retrospective and exploratory nature of the study, the prognostic relevance of Ki67 (as well as its optimal cut-off) seems to significantly differ according to histology. In particular, a very low cut-off of Ki67 (4%) may significantly discriminate the prognosis of pts with ILC.

Clinical trial identification

Legal entity responsible for the study

University of Verona, Verona


University of Verona, Verona


All authors have declared no conflicts of interest.