A. Kelsall1, T. Novice1, B. Chang1, K. Ogu1, J. Noda1, R. Hoogmoed1, J. Loh1, H. Cheriyan1, C. Ky1, M. Martin1, B. Sunkara1, M. S. Cohen1 1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA
Introduction: Secondary lymphedema is a significant complication after axillary (ALND) or inguinal lymphadenectomy (ILND) in melanoma patients. While prior studies have noted pre- and postoperative risk factors, no clinically useful statistical tool exists for predicting the likelihood of a patient developing lymphedema following ALND or ILND. In this context, we used data from the largest single institution cohort of melanoma patients who underwent ALND or ILND to develop a lymphedema risk assessment tool and tested its efficacy in predicting the development of lymphedema in another cohort of patients undergoing similar procedures
Methods: We retrospectively reviewed our prospective database of melanoma patients undergoing either ALND or ILND. The model cohort (N=524;302 ALND,222 ILND) contained patients undergoing procedures between June 2005 and June 2015. The test cohort (N=98;53 ALND,45 ILND) contained patients between November 2015 and June 2016. Patients having bilateral lymphadenectomy, iliac dissections, or preoperative chemotherapy were excluded. Demographic and clinical data were collected from the electronic medical record (EMR). We used stepwise logistic regression to model the impact of various preoperative and postoperative risk factors on the likelihood of developing lymphedema in the model cohort. These models were then used to calculate risk scores for patients in the test cohort, and procedure-specific tercile thresholds were used to assign patients to “low”,“moderate”, and “high” risk categories and clinical outcome incidence was used to evaluate the accuracy of the risk prediction tool.
Results:Key preoperative factors (ever smoking, stage, and peripheral vascular disease) were included in the preoperative risk estimation model. Three additional postoperative factors (# nodes dissected, adjuvant therapy, and 30-day non-lymphedema complications) were added when estimating risk postoperatively. In the test cohort, “low”, “moderate” and “high” risk patient groups experienced significantly different (each p<0.01) incidences of lymphedema when estimating risk both preoperatively and postoperatively (see table 1). Both the postoperative and preoperative models equally were able to predict and stratify the incidence of lymphedema.
Conclusion:Our findings demonstrate that this novel risk assessment tool can accurately use either pre- or postoperative factors to reasonably predict the risk of developing secondary lymphedema in melanoma patients undergoing either an ILND or ALND. Using such a risk-stratification tool may provide the patient and surgical team important information for surgical decision-making and discussion.
