Incorporation of sentinel lymph node metastasis size into a nomogram predicting nonsentinel lymph node involvement in breast cancer patients with a positive sentinel lymph node

Abstract

Background and objective: Sentinel lymph node (SLN) metastasis size is an important predictor of non-SLN involvement. The goal of this study was to construct a nomogram incorporating SLN metastasis size to accurately predict non-SLN involvement in patients with SLN-positive disease.

Methods: We identified 509 patients with invasive breast cancer with a positive SLN who underwent completion axillary lymph node dissection (ALND). Clinicopathologic data including age, tumor size, histology, grade, presence of multifocal disease, estrogen and progesterone receptor status, HER2/neu status, presence of lymphovascular invasion (LVI), number of SLN(s) identified, number of positive SLN(s), maximum SLN metastasis size and the presence of extranodal extension were recorded. Univariate and multivariate logistic regression analyses identified factors predictive of positive non-SLNs. Using these variables, a nomogram was constructed and subsequently validated using an external cohort of 464 patients.

Results: On univariate analysis, the following factors were predictive of positive non-SLNs: number of SLN identified (P < 0.001), number of positive SLN (P < 0.001), SLN metastasis size (P < 0.001), extranodal extension (P < 0.001), tumor size (P = 0.001), LVI (P = 0.019), and histology (P = 0.034). On multivariate analysis, all factors remained significant except LVI. A nomogram was created using these variables (AUC = 0.80; 95% CI, 0.75-0.84). When applied to an external cohort, the nomogram was accurate and discriminating with an AUC = 0.74 (95% CI, 0.68-0.77).

Conclusion: SLN metastasis size is an important predictor for identifying non-SLN disease. In this study, we incorporated SLN metastasis size into a nomogram that accurately predicts the likelihood of having additional axillary metastasis and can assist in personalizing surgical management of breast cancer.

Figure 1. Nomogram to predict likelihood of additional, non-sentinel lymph node metastases in a patient with a positive SLN

The first row (Points) is the point assignment for each variable. Rows 2–8 contain the variables included in the model. For an individual patient, each variable is assigned a point value based on the characteristic. A vertical line is made between the appropriate variable value and the Points line. The assigned points for the 7 variables are summed and the total is found in row 9 (Total Points). Once the total points value is determined, a vertical line is made between row 9 and row 10 to determine the risk of additional positive non-SLNs. SLN maximum metastasis (max met) size is measured in millimeters.

Figure 2. Validation of nomogram performance

The performance of the nomogram was quantified with respect to (A) discrimination and (B) calibration. Discrimination was quantified with the area under the receiver operating characteristic (ROC) curve (AUC). To construct the calibration curve, a histogram of the probabilities calculated using the nomogram was plotted along the horizontal axis. The vertical axis represents the actual incidence on non-SLN positivity.

Figure 2. Validation of nomogram performance

The performance of the nomogram was quantified with respect to (A) discrimination and (B) calibration. Discrimination was quantified with the area under the receiver operating characteristic (ROC) curve (AUC). To construct the calibration curve, a histogram of the probabilities calculated using the nomogram was plotted along the horizontal axis. The vertical axis represents the actual incidence on non-SLN positivity.

Figure 3. External validation

To determine if the nomogram is generalizable to external patient populations we used a cohort of patients from the Mayo Clinic. The AUC of 0.74 confirmed that the nomogram has broad applicability.