10 Using the Nodal Ratio of Positive/Excised Nodes Rather Than the Absolute Number of Positive Lymph Nodes Minimizes Differences in Postmastectomy LRR Risk Between the British Columbia (BC) and The University of Texas M. D. Anderson Cancer Center (MDACC) Prospective Data Sets W. A. Woodward 1 , P. T. Truong 2 , H. D. Thames 1 , I. A. Olivotto 2 , J. Ragaz 2 , T. A. Buchholz 1 1 The University of Texas M.D. Anderson Cancer Center, Houston, TX, 2 British Columbia Cancer Agency, University of British Columbia, Victoria, BC, Canada Background: The BC and Danish postmastectomy radiotherapy (PMRT) randomized trials demonstrated improved overall survival with PMRT for patients with one to three positive lymph nodes (+LN). However, data from MDACC and other U.S. cooperative groups have reported lower LRR risk compared to the BC and Danish randomized trials. Differences in the median number of nodes excised may account for the discrepancy in reported LRR between these institutions. Purpose/Objective(s): To examine whether using nodal ratio of positive/excised axillary nodes, which accounts for the differences in number of lymph nodes excised, could be used to identify cohorts of similar risks within these two data sets. Materials/Methods: Data from the no radiation arm of the BC PMRT randomized trial (n=123, all premenopausal) were compared to data from 505 patients aged  50 years treated on sequential prospective chemotherapy trials at MDACC with mastectomy and no radiation. Locoregional recurrences in the MDACC data set were recoded to be consistent with the BC definitions of LRR. LRR was defined as the first site of recurrence involving the chest wall (local) and/or axillary, supra or infraclavicular, and internal mammary nodes (regional), with or without simultaneous distant recurrence. LRR events that occurred 1month after distant recurrence were not recorded. Kaplan-Meier LRR curves were generated and compared according to the absolute number of +LN and according to nodal ratios. Nodal ratio cut points were generated using regression tree analysis in the MDACC data set and tested in the BC data set. Results: The median number of excised lymph nodes was different in the two data sets (11 in BC vs. 17 in MDACC, p0.0001). Distributions of age, T stage, nodal ratio, and ER status were not significantly different between the data sets. Median follow up was 18.8 years in BC vs. 12.2 years at MDACC. There were statistically significant differences in the 15-year risk of LRR between the two data sets for patients with 1-3 +LN (26% BC vs. 11% MDACC, p=0.026) and for patients with 4 to 9 +LN (61% BC vs. 30% MDACC, p=0.003). However, when LRR was assessed using the nodal ratio, there were no statistical differences between the two data sets in the lowest risk cohorts. Specifically, for patients with a nodal ratio of less than 10%, the 15-year LRR rates were 15% BC vs. 9% MDACC (p=0.36) and for those with a nodal ratio of 10%-20% the LRR rates were 25% BC vs. 18% MDACC (p=0.50). On multivariate analysis NR was the strongest predictor for LRR in both data sets. Conclusions: Grouping patients by the nodal ratio rather than by the absolute number of positive lymph nodes partially corrects differences in LRR rates that may exist due to less extensive nodal dissections. The nodal ratio may be a useful method for extrapolating data from randomized trials to clinical practices in which the axillary surgical procedures vary. Author Disclosure: W.A. Woodward, None; P.T. Truong, None; H.D. Thames, None; I.A. Olivotto, None; J. Ragaz, None; T.A. Buchholz, None. 11 Factors Associated With Involvement of Four or More Axillary Nodes for Sentinel Lymph Node Positive Patients A. Katz 1 , B. L. Smith 1 , M. Golshan 2 , A. Niemierko 1 , W. Kobayashi 1 , M. Gadd 1 , M. Specht 1 , L. Rizk 1 , A. Kelada 1 , A. Taghian 1 1 Massachusetts General Hospital, Boston, MA, 2 Brigham and Women’s Hospital, Boston, MA Background: The standard of care for patients with a positive (+) sentinel lymph node (SLN) is axillary dissection, however, for various reasons some SLN + patients do not undergo dissection. Radiation fields for patients with a + SLN who do not undergo axillary dissection are not clearly defined. Treatment of the supraclavicular fossa/ axillary apex is indicated for patients with  4 nodes following axillary dissection. SLN + patients who do not undergo axillary dissection but who are unlikely to have  4 involved nodes might be safely treated with less extensive nodal radiation. The purpose of this study was to define possible predictors of having  4 involved nodes. Materials/Methods: The records of 488 patients with breast cancer and 1-3 involved SLNs, who underwent completion axillary dissection at two academic cancer centers were reviewed. None of these patients received neoadjuvant chemotherapy. Factors associated with having  4 involved axillary nodes (SLNs and nonSLNs) were evaluated by simple and multiple logistic regression analysis. Results: 79 of 488 patients had  4 + nodes. Table 1 lists the univariate logistic regression analysis of factors associated with involvement of  4 axillary nodes for patients with a + SLN. Type of surgery, estrogen or progesterone receptor status were not associated with having  4 nodes. On multivariate analysis, having  4 SLNs was associated with lymphovascular space invasion (LVSI), increasing number of involved SLNs, increasing primary tumor size, invasive lobular histology and the presence of extranodal extension. A nomogram to predict the probability of having  4 nodes based on patients’ pathologic data was developed from this multivariate model. A separate previously published dataset of 246 SLN + patients treated a community hospital in another city was used to validate this model. Conclusions: Patients with a low probability of having  4 nodes can be identified and might be adequately treated with modified high tangential fields rather than comprehensive nodal radiation. Prospective studies to evaluate this approach are warranted. S6 I. J. Radiation Oncology ● Biology ● Physics Volume 66, Number 3, Supplement, 2006