TY - JOUR
T1 - Use Pre- and Intra-Operative Data To Predict Probability of Positive Non-Sentinel Lymph Nodes.
JF - Cancer Research
JO - Cancer Res
SP - 302
LP - 302
M3 - 10.1158/0008-5472.SABCS-09-302
VL - 69
IS - 24 Supplement
AU - Chagpar, A.
AU - Blumencranz, P.
AU - Whitworth, P.
AU - Deck, K.
AU - Rosenberg, A.
AU - Simmons, R.
AU - Reintgen, D.
AU - Beitsch, P.
AU - Julian, T.
AU - Saha, S.
AU - Mamounas, E.
AU - Giuliano, A.
AU - Cook, E.
AU - Wang, S.
Y1 - 2009/12/15
UR - http://cancerres.aacrjournals.org/content/69/24_Supplement/302.abstract
N2 - Abstracts: Thirty-Second Annual CTRC‐AACR San Antonio Breast Cancer Symposium‐‐ Dec 10‐13, 2009; San Antonio, TXIntroduction: Many sentinel lymph node (SLN) positive patients have no further positive nodes, and benefit little from completion axillary node dissection (ALND). Prediction of non-SLN metastases (mets) based on pre- and intra-operative data remains a challenge. We sought to determine if this could be improved using a novel intraoperative RT-PCR based assay.Methods: 728 patients were enrolled in 2 prospective studies of the GeneSearch™ BLN Assay (Veridex LLC) for SLN mets. Of these, 205 had ≥1 positive SLN. Of these, 116 underwent ALND and had complete data for tumor size, BLN assay and frozen section (FS)/touch imprint cytology (TIC). These patients formed the cohort of interest for this study. A BLN Score was calculated using the cycle time (CT) values for mammaglobin (MG) and CK19 based on the BLN assay. These CT values were divided by that of the internal control (IC), and the least of the two was used. The BLN Score=10.2586-2.8053 x min(MG_CT/IC_CT, CK19_CT/IC_CT)+1.7164 x BLN positive rate in SLNs. Stepwise logistic regression models were evaluated for their ability to predict non-SLN status. A final score was created to determine the probability of non-SLN mets based on pre- and intra-operative data alone.Results: The median tumor size in this cohort was 2.20 cm (range: 0.19-22.00). The median number of SLNs removed was 2; the median number of positive SLNs was 1. The median number of non-SLNs removed was 12 (range; 1-38).As tumor size can be estimated from preoperative imaging and/or measured intraoperatively, this formed our initial model. Tumor size was a statistically significant predictor of non-SLN mets (AUC=0.631, p=0.027). In a model with the proportion of positive SLNs by intraoperative analysis (FS/TIC positive rate), both tumor size and FS/TIC positive rate were significant (p=0.022 and 0.006). In the final model that included tumor size, FS/TIC positive rate, and BLN score, BLN score was statistically significant (p=0.012) while tumor size and FS/TIC positive rate became borderline significant (p= 0.071 and 0.084, respectively). The final model brought AUC to 0.750 (p=0.048 compared to the original model). This was somewhat better than the model with BLN score alone (AUC=0.72) and that with BLN score and tumor size (AUC=0.742), although not significantly so.A logistic regression model with dichotomized tumor size data was then fitted. When tumor size was >2 cm, FS/TIC positive rate was ≥50%, and BLN score was >10, the probability of non-SLN mets is >62.5%. When tumor size was ≤2cm, FS/TIC positive rate was <50%, and the BLN score was <7, the probability of non-SLN mets is <5%. Even when tumor size >2 cm or FS/TIC positive rate was ≥50%, the probability of non-SLN mets is < 8% if BLN score<7 (see figure).Conclusions: BLN score is a significant predictor of non-SLN metastases. When combined with tumor size and proportion of SLNs positive by FS/TIC, BLN score may provide accurate estimate of probability of non-SLN mets.Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 302.
ER -