Improving NPI for breast cancer prognosis by including PR and HER-2 expression: own data and external validation set.

Abstract

Abstract Abstract #1080 The Nottingham Prognostic Index (NPI) classifies primary operable breast cancer patients into good (NPI ≤ 3.4) , intermediate (NPI 3.4-5.4) and poor (NPI ≥ 5.4) prognostic risk groups [Br J Cancer 1987; 56: 489–92]. Studies indicate a significant short term prognostic role for estrogen (ER), progesterone (PR) and HER-2 receptor on disease free survival (DFS). We proposed at SABCS 2007 to add PR and HER-2 expression to the NPI called improved NPI (iNPI) defined as (NPI) + (HER-2) – (PR), with HER-2 and PR '1' in case of amplified HER-2 and any PR-expression and '0' otherwise.
 Internal validation of the iNPI was performed on 1928 breast cancer patients, primary operated between Jan 2000 and June 2005 at Univ Hospital Leuven. All patients were classified into any of the three NPI or iNPI risk group. Mean follow-up was 3.5 yrs and DFS was defined as previously stated [Breast Cancer Research and Treatment 2007; 106: S247-S247]. Out of the 601 patients classified as NPI low risk, 59 shifted towards the intermediate iNPI group. Comparison of DFS for the 59 shifted with the non-shifted patients from the low NPI class did not reveal a significant difference (p=0.129, log-rank test). 57 patients out of 935 and 136 out of 392 respectively classified as intermediate and high NPI group became high and intermediate risk according to the iNPI. In both cases, DFS of the shifted patients differed significantly from DFS of the non-shifted patients (p=<.001 en p=0.029). DFS of the shifted patients with the iNPI group they where shifted to was not/marginally significant (p=0.856, p=0.462 and p=0.047 for low, intermediate and high NPI respectively).
 For external validation a dataset from the Breast Cancer Micrometastasis group in Oslo with 677 patients and known NPI, PR and HER-2 was used. Figure 1 confirms a significant difference between DFS of the shifted and the non-shifted patients within the highest NPI risk group (p=0.021). The difference in DFS between shifted and non-shifted patients was marginally non-significant (p=0.068) in the intermediate and non-significant in the low NPI group (p=0.116). The non-significant difference between DFS of the shifted groups with the DFS of the group they were shifted to was confirmed, with exception of the low NPI group (difference in systemic treatment) (p=0.040, p=0.621 and p=0.076 for low, intermediate and high NPI respectively).
 We confirm the results from an internal study at the Univ Hospital in Leuven on an external dataset. Inclusion of the PR and HER-2 status into the NPI results in more homogenate risk groups concerning DFS.
 
 Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 1080.