K. E. Engelhardt1,2, O. Kutlu3, W. Lancaster1, K. Staveley-O’Carroll4, E. Kimchi4, A. M. Abbott1, E. R. Camp1 1Medical University Of South Carolina,Charleston, Sc, USA 2Northwestern University,Chicago, IL, USA 3University Of Miami,Miami, FL, USA 4University Of Missouri,Columbia, MO, USA
Introduction:
Current recommendations by the National Comprehensive Cancer Network are for consideration of sentinel lymph node biopsy (SLNB) for patients with melanoma <1mm. Clinicians may use presence of ulceration, mitotic rate (MR), and tumor depth to aid in decision making. Low MR has been associated with high false positive rates, however, it is unknown at what value MR becomes a significant predictor of SLN positivity. We hypothesized that higher MR would strongly predict tumor biology in thin melanomas and, therefore, predict SLN involvement.
Methods:
We queried the Surveillance Epidemiology and End Results database for all patients diagnosed with trunk and extremity cutaneous melanoma ≤1mm depth from 2010 to 2013 who underwent SLNB to determine whether MR was an independent predictor of SLN. Patient demographics and tumor characteristics (depth, mitotic rate, ulceration, and tumor location were evaluated). MR was dichotomized at multiple cut-points and estimated multiple stratified logistic regression models were used to identify the ideal cut point for MR as a predictor of SLN status. After determination of the ideal cut-point, we then estimated a hierarchical multivariable logistic regression model to determine the association between high MR and SLN+. We also performed a subset analysis for melanoma at the upper limit of thin categorization (0.75-1mm).
Results:
patient cohort was 51.7% male (n=2,246) with mean patient age of 55.6 years (range 18-85). We identified 4 or more mitosis per high power field as the ideal cut point for dichotomization of MR into high and low risk groups. In our final regression model, MR≥4 (OR 3.67 95%CI 2.66-5.05; p<0.001) and ulceration (OR 2.32 95%CI 1.62-3.31; p<0.001) were significantly associated with SLN+. In a subset analysis of 0.75-1mm melanomas, MR≥4 (OR 4.61 95%CI 2.77-7.66; p<0.001) and ulceration (OR 2.29 95%CI 1.37-3.82 p=0.002) were associated with SLN+. The SLN+ rate of the entire cohort of 0.75-1mm patients was 5.6% (n=122/2184); this number increased to 14.4% (n = 20/139) when the cohort was sub-selected for MR ≥4. Interestingly, depth was not an independent predictor of SLN+ rate in the overall cohort or the 0.75-1mm subset.
Conclusion:
In our analysis, MR≥4 was the strongest independent predictor of SLN+ in thin melanoma. In patients with thin melanoma we found that MR and ulceration, not tumor depth, were independent predictors of SLN involvement. When evaluating patients with thin melanoma for SLNB, MR and ulceration may aid in decision-making analysis more than tumor depth alone.
