ImportancePatients with melanoma are selected for sentinel lymph node biopsy (SLNB) based on their risk of a positive SLN. To improve selection, the Memorial Sloan Kettering Cancer Center (MSKCC) and Melanoma Institute Australia (MIA) developed predictive models, but the utility of these models remains to be tested.ObjectiveTo determine the clinical utility of the MIA and MSKCC models.Design, Setting, and ParticipantsThis was a population-based comparative effectiveness research study including 10 089 consecutive patients with cutaneous melanoma undergoing SLNB from the Swedish Melanoma Registry from January 2007 to December 2021. Data were analyzed from May to August 2023.Main Outcomes and Measures,The predicted probability of SLN positivity was calculated using the MSKCC model and a limited MIA model (using mitotic rate as absent/present instead of count/mm2 and excluding the optional variable lymphovascular invasion) for each patient. The operating characteristics of the models were assessed and compared. The clinical utility of each model was assessed using decision curve analysis and compared with a strategy of performing SLNB on all patients.ResultsAmong 10 089 included patients, the median (IQR) age was 64.0 (52.0-73.0) years, and 5340 (52.9%) were male. The median Breslow thickness was 1.8 mm, and 1802 patients (17.9%) had a positive SLN. Both models were well calibrated across the full range of predicted probabilities and had similar external area under the receiver operating characteristic curves (AUC; MSKCC: 70.8%; 95% CI, 69.5-72.1 and limited MIA: 69.7%; 95% CI, 68.4-71.1). At a risk threshold of 5%, decision curve analysis indicated no added net benefit for either model compared to performing SLNB for all patients. At risk thresholds of 10% or higher, both models added net benefit compared to SLNB for all patients. The greatest benefit was observed in patients with T2 melanomas using a threshold of 10%; in that setting, the use of the nomograms led to a net reduction of 8 avoidable SLNBs per 100 patients for the MSKCC nomogram and 7 per 100 patients for the limited MIA nomogram compared to a strategy of SLNB for all.Conclusions and RelevanceThis study confirmed the statistical performance of both the MSKCC and limited MIA models in a large, nationally representative data set. However, decision curve analysis demonstrated that using the models only improved selection for SLNB compared to biopsy in all patients when a risk threshold of at least 7% was used, with the greatest benefit seen for T2 melanomas at a threshold of 10%. Care should be taken when using these nomograms to guide selection for SLNB at the lowest thresholds.

中文翻译：

---

用于预测前哨淋巴结状态的 MIA 和 MSKCC 工具的基于群体的验证

重要性根据 SLN 阳性的风险，选择黑色素瘤患者进行前哨淋巴结活检 (SLNB)。为了改善选择，纪念斯隆凯特琳癌症中心 (MSKCC) 和澳大利亚黑色素瘤研究所 (MIA) 开发了预测模型，但这些模型的实用性仍有待测试。 目的确定 MIA 和 MSKCC 模型的临床实用性。 设计、设置和参与者这是一项基于人群的比较有效性研究，包括 2007 年 1 月至 2021 年 12 月瑞典黑色素瘤登记处连续 10 089 名接受 SLNB 的皮肤黑色素瘤患者。数据分析时间为 2023 年 5 月至 8 月。主要结果和措施，预测使用 MSKCC 模型和有限 MIA 模型计算 SLN 阳性概率（使用有丝分裂率作为不存在/存在而不是计数/mm2并排除每位患者可选的可变淋巴管侵犯）。评估和比较了模型的运行特性。使用决策曲线分析评估每种模型的临床效用，并与对所有患者进行前哨淋巴结活检的策略进行比较。结果在 10 089 名患者中，中位年龄 (IQR) 为 64.0 (52.0-73.0) 岁，其中 5340 名患者 (52.9%) ）是男性。Breslow 厚度中位数为 1.8 mm，1802 名患者 (17.9%) SLN 阳性。两种模型在整个预测概率范围内都得到了很好的校准，并且在受试者工作特征曲线下具有相似的外部面积（AUC；MSKCC：70.8%；95% CI，69.5-72.1；有限 MIA：69.7%；95% CI，68.4） -71.1）。在 5% 的风险阈值下，决策曲线分析表明，与对所有患者进行 SLNB 相比，这两种模型都没有增加净收益。在 10% 或更高的风险阈值下，与 SLNB 相比，两种模型都为所有患者增加了净效益。使用 10% 的阈值观察到 T2 黑色素瘤患者获益最大；在这种情况下，与针对所有人的 SLNB 策略相比，列线图的使用导致 MSKCC 列线图每 100 名患者净减少 8 例可避免的 SLNB，有限 MIA 列线图每 100 名患者净减少 7 例。结论和相关性本研究证实了MSKCC 和有限的 MIA 模型在大型、具有全国代表性的数据集中的统计性能。然而，决策曲线分析表明，当使用至少 7% 的风险阈值时，与活检相比，使用这些模型仅改善了所有患者对 SLNB 的选择，在阈值为 10% 时，T2 黑色素瘤的获益最大。使用这些列线图指导选择最低阈值的 SLNB 时应小心。