In:
Cancer Research, American Association for Cancer Research (AACR), Vol. 71, No. 24_Supplement ( 2011-12-15), p. P3-07-09-P3-07-09
Abstract:
Background The Memorial Sloan Kettering Cancer Center-Breast Cancer Nomogram (MSKCC-BCN) predicts additional nodal metastasis in patients with positive sentinel lymph node (SLN). This statistical tool does not include HER2 status. It has been shown that the interaction covariate between estrogen receptor (ER) and HER2 status was a determinant of SLN positivity. The purpose of our study was to determine if the accuracy of MSKCC BCN could be enhanced with new variables. Patients and methods: Our dataset consisted of 2769 consecutive patients treated for operable breast cancer with SLN biopsies between 2006 and 2009. We selected all the patients (n = 588) with a positive SLN who underwent a completion axillary lymph node dissection (ALND). The MSKCC-BCN was used to calculate the theoretical risk of additional nodal metastasis for all patients. The evaluation of the MSKCC-BCN was performed with calibration test (Cox method) and performance test (Bleeker). Multivariate analysis used a logistic regression model. The input was based on the variables found significant in the univariate analysis. Interaction covariate between ER and HER2 status was included in the model. Our model was then analyzed in terms of discrimination (area under the curve) and of calibration (Hosmer-Lemeshow test). Results: Calibration test showed significant differences between the probability of additional nodal disease predicted by the MSKCC-BCN and the probability observed in our population for the following subgroups of patients: histological grade 3 (p= 0.007), lymphovascular invasion (p=0.03), multifocality (p=0.04), positive ER (p=0.002), micrometastasis in the SLN (p=0.003), isolated tumor cells in the SLN (p=0.02 and positive HER2 (p=0.01). Performance test showed significant differences for the following variables: histological grade (p= 0.02), size of the SLN metastasis (p=0.04) and HER2 status (p=0.01). This shows that the MSKCC-BCN is not well calibrated and cannot be used for our population. A multivariate model to determine the probability of additional nodal metastasis was defined with the following variables: pathologic size of the sentinel node metastasis, interaction covariate between the ER and HER2 status, number of positive SLN and number of SLN removed [Table 1]. This multivariate model resulted in a nomogram tested on the training population. It was discriminating with an area under the curve of 0.76 [0.720−0.808] and well calibrated (Hosmer-Lemeshow test p= 0.51). Conclusions: We showed that HER2 status and pathologic size of the SLN metastasis are determinant to predict additional nodal metastasis after a positive SLN. These two variables were included in a new nomogram that could help in the decision-making concerning further axillary treatment in these patients. Our model has to be validated prospectively in external series to confirm its accuracy. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P3-07-09.
Type of Medium:
Online Resource
ISSN:
0008-5472
,
1538-7445
DOI:
10.1158/0008-5472.SABCS11-P3-07-09
Language:
English
Publisher:
American Association for Cancer Research (AACR)
Publication Date:
2011
detail.hit.zdb_id:
2036785-5
detail.hit.zdb_id:
1432-1
detail.hit.zdb_id:
410466-3
Permalink