University of Twente Student Theses


Validation of a nomogram to predict axillary lymph node status in Dutch early stage breast cancer patients

Aarnink, Merel (2016) Validation of a nomogram to predict axillary lymph node status in Dutch early stage breast cancer patients.

[img] PDF
Abstract:Introduction: Axillary lymph node (ALN) status is a an important prognostic factor in patients with primary breast cancer. Sentinel lymph node biopsy (SLNB) is standard procedure for staging the axillary lymph nodes. However, the procedure is related to some risks on morbidity. To avoid axillary surgery in patients with low risk on metastases, a nomogram to predict ALN status based on a Chinese population is developed in 2016. The model showed good performance in an internal validation population. The aim of this study was to externally validate this model in a Dutch population. Methods: Early stage breast cancer patients from six Dutch hospitals, diagnosed between January 2011 and December 2015, and with positive ultrasound findings were included. Patients who received primary systemic therapy, patients with bilateral breast cancer or patients with incomplete data on the variables in the nomogram were excluded. The validation population was compared to the development population. The area under the receiver operating characteristic (ROC) curve (AUC), false negative rates (FNR) and false omission rates (FOR) were calculated to determine the predictive accuracy. The Hosmer-Lemeshow goodness of fit (HL) test and a calibration plot were used to assess its goodness of fit. The model was updated using logistic regression. Results: 1,416 patients were included in this validation study. Large differences in tumour- and ALN characteristics were found in the development- and validation population. Only transverse diameter showed no significant difference. 24.93% of all patients in the validation population had ALN metastases, compared to 50.62% of the patients in the development population. The AUC was 0.77. The HL- test showed a significant difference between the predicted probability and the observed event rate. However, the calibration plot showed a good fit. In the updated model, the FNR was 5.67% at a cut-off point of 7%. Discussion: Based on predictive accuracy and calibration, the model seems to perform good in the Dutch patient population. The updated model has better predictive accuracy for selecting low-risk patients when compared to the original model. Although the model cannot serve as preoperative information tool in the Netherlands, the model can be of possible value in other Western European countries. For the Dutch situation, it would be worthwhile to adapt the model or to develop a new model with preoperative variables only. Keywords: breast cancer; axillary lymph node status; ultrasound; predictive model; validation
Item Type:Essay (Master)
Faculty:TNW: Science and Technology
Subject:44 medicine
Programme:Health Sciences MSc (66851)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page