Abstract 165P
Background
ILC represents about 10% of all breast cancers. Both Ki67 cut-off to discriminate prognostic (px) groups and genomic signatures (GS) used in early disease may capture the biology of the most frequent invasive carcinoma of non-special type. Different performance of EPclin, OncotypeDx (ODx) and Prosigna in ILC have been seen. Our aim was to define the association of Ki67 with the risk categories defined by these GS.
Methods
Retrospective study including patients (pts) with pT1-2N0-1 Luminal HER2- ILC from 2 sites diagnosed between (b/w) 2017 and 2021 and tested by EPclin/EndoPredict, ODx or Prosigna. EndoPredict, EPclin (EndoPredict + pT + pN) and ODx pts: low vs high risk, being the cut-off value 3.3 with EPClin, 5 with EndoPredic, and 26 with ODx if >50y and 21 if ≤50y. EProsigna: low-risk (ROR 0-40 if pN0, 0-15 if pN1), intermediate-risk (ROR 41-60 if pN0, 16-40 if pN1) or high-risk (ROR >60 if pN0 and >40 if pN1).
Results
70 pts (EPClin 25, ODx 21, Prosigna 24) included, median (M) age of 61.5y, no differences in main pathological features b/w cohorts. EPclin and ODx high-risk: 48% and 9.5%, respectively. Prosigna: 33.3% low, 16.7% intermediate and 50% high risk. ODx pts were less likely to be classified as high-risk (OR 0.11 vs both EPClin/ODx). EndoPredict high-risk and low-risk groups had different M Ki67 values (11% vs 3%, p=0.05), but neither ODx (13.5% vs 8%, p=0.53) nor EPClin (10%vs 10%). M Ki67 levels in Prosigna low, intermediate and high-risk groups were 7.5%, 6.5% and 23.5%, respectively. There was a difference regarding M Ki67 b/w Prosigna low/intermediate vs high groups (p<0.01), but not b/w low vs intermediate groups. Using EndoPredict, the AUC for Ki67 was 0.79 with an optimal cut-off value of 4% (sensitivity 90%, specificity 80%). With Prosigna the AUC for Ki67 was 0.84 with an optimal cut-off value of 10.5% (sensitivity of 83%, specificity of 75%).
Conclusions
In our study, Ki67 levels were related to EndoPredict (EPClin molecular part) groups and Prosigna modified categories (low/intermediate vs high). The best cut-offs to differentiate risk groups were 4% and 10%, respectively, both lower than the generally accepted (14% or 20%). Our findings support prior data suggesting that lower Ki67 threshold should be established to discriminate prognostic groups in ILC.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.