“Sentinel lymph node (SLN) biopsy provides valuable staging information—determining whether cutaneous melanoma has metastasized to regional lymph nodes—that can be used by clinicians to make informed management recommendations, plan appropriate follow-up schedules, and determine clinical trial eligibility,” says Serigne N. Lo, PhD, MSc, AStat. “Although national and international melanoma management guidelines—including those from the NCCN and ASCO/SSO—indicate which patients are eligible for SLN biopsy, the overall positivity rate among those who undergo SLN biopsy is low (16% to 27%), suggesting that SLN biopsy may not be appropriate for every patient with melanoma.” Dr. Lo is an Associate Professor of Biostatistics at Central Clinical School, The University of Sydney; Head, Research and Biostatistics Group, Melanoma Institute Australia.
Updating & Improving an SLN Nomogram
In a study published in the Journal of Clinical Oncology, Dr. Lo and colleagues sought to update and improve a nomogram to predict SLN positivity that was developed at the Memorial Sloan Kettering Cancer Center (MSKCC model) 15 years ago by building a new risk prediction model. “We initially considered all parameters used in the MSKCC model, but in addition, considered new clinicopathologic factors that have emerged in the recent literature as predictors for SLN positivity,” explains Dr. Lo. “We developed the model using information from nearly 3,500 patients who had SLN biopsy performed between January 2003 and December 2014 from our prospectively maintained research database at Melanoma Institute Australia (MIA) in Sydney, Australia.”
With the new nomogram (MIA model), body site and Clark level from the MSKCC model were replaced with mitotic rate, melanoma subtype, and lymphovascular invasion. “We assessed the predictive performance of the nomogram and its generalizability to the wider melanoma population using comprehensive statistical methodology, and validated it by applying it to an independent, similarly large sample of melanoma patients who had undergone SLN biopsy at the MD Anderson Cancer Center,” Dr. Lo adds.
The six input parameters chosen for the MIA model (Figure) were found to have statistically significant differences between patients who were SLN positive and negative in both the MIA and MD Anderson cohorts. “For example, SLN-positive patients had a greater tumor thickness, were younger, were more likely to have ulcerated melanomas, and had a higher mitotic rate than SLN-negative patients,” says Dr. Lo. “Patients with acral or superficial spreading melanoma subtypes were more likely to have metastatic melanoma in their lymph nodes. On the other hand, the pure desmoplastic melanoma subtype (melanoma with ³90% desmoplasia) was associated with a low risk of SLN positivity.”
More Accurate Risk Assessment
“We demonstrated that the combination of six widely available clinicopathologic parameters (age, tumor thickness, ulceration, mitotic rate, melanoma subtype, and lymphovascular invasion) provides a more accurate risk estimate for SLN positivity than the MSKCC nomogram or the NCCN or ASCO/SSO guidelines,” explains Dr. Lo. Indeed, the receiver operating characteristic curve of the MSKCC model had a predictive accuracy of 67.7% (95% CI, 65.3% to 70.0%), compared with 73.9% (95% CI, 71.9% to 75.9%) for the MIA model, an increase in accuracy of 9.2%.
“Importantly, use of the MIA model could spare low-risk patients the inconvenience, cost, and potential risks of SLN biopsy, while ensuring that high-risk patients are still identified,” Dr. Lo notes. “To illustrate this, if the decision to proceed with SLN biopsy had been based on our nomogram, 22.1% of the patients with a negative SLN would not have been biopsied.” That compares with rates of 13.4% with the MSKCC model and 12.4% with the ASCO/SSO criteria.
“We hope that our nomogram will enable clinicians to have better-informed discussions with their patients with melanoma based on each individual’s percentage risk of SLN positivity, rather than relying on previous criteria that provide a much less precise estimate of the risk of disease spread to the regional lymph nodes,” says Dr. Lo. To this end, we have created a freely available online version of the nomogram to facilitate its widespread use in clinical practice. It can be accessed, along with a growing suite of other risk-prediction tools for melanoma, at www.melanomarisk.org.au.”