Germline mutations in CDKN2A greatly increase risk of developing cutaneous melanoma. We have constructed a risk prediction model, Familial Risk Assessment of Melanoma (FRAMe), for estimating the likelihood of carrying a heritable CDKN2A mutation among Australian families, where the prevalence of these mutations is low. Using logistic regression, we analysed characteristics of 299 Australian families recruited through the Sydney site of GenoMEL (international melanoma genetics consortium) with at least three cases of cutaneous melanoma (in situ and invasive) among first-degree blood relatives, for predictors of the presence of a pathogenic CDKN2A mutation. The final multivariable prediction model was externally validated in an independent cohort of 61 melanoma kindreds recruited through GenoMEL Queensland. Family variables independently associated with the presence of a CDKN2A mutation in a multivariable model were number of individuals diagnosed with melanoma under 40 years of age, number of individuals diagnosed with more than one primary melanoma, and number of individuals blood related to a melanoma case in the first degree diagnosed with any cancer excluding melanoma and non-melanoma skin cancer. The number of individuals diagnosed with pancreatic cancer was not independently associated with mutation status. The risk prediction model had an area under the receiver operating characteristic curve (AUC) of 0.851 (95% CI 0.793, 0.909) in the training dataset, and 0.745 (95%CI 0.612, 0.877) in the validation dataset. This model is the first to be developed and validated using only Australian data, which is important given the higher rate of melanoma in the population. This model will help to effectively identify families suitable for genetic counselling and testing in areas of high ambient ultraviolet radiation. A user-friendly electronic nomogram is available at www.melanomarisk.org.au.

CDKN2A 中的种系突变大大增加了患皮肤黑色素瘤的风险。我们构建了一个风险预测模型，黑色素瘤家族风险评估 (FRAMe)，用于估计在这些突变的流行率较低的澳大利亚家庭中携带可遗传的CDKN2A突变的可能性。使用逻辑回归，我们分析了通过 GenoMEL（国际黑色素瘤遗传学联盟）悉尼站点招募的 299 个澳大利亚家庭的特征，在一级血缘亲属中至少有 3 例皮肤黑色素瘤（原位和侵袭性），作为存在的预测因素致病性CDKN2A突变。最终的多变量预测模型在通过 GenoMEL Queensland 招募的 61 名黑色素瘤亲属的独立队列中进行了外部验证。与CDKN2A的存在独立相关的家族变量多变量模型中的突变是 40 岁以下被诊断患有黑色素瘤的个体数量，被诊断患有一个以上原发性黑色素瘤的个体数量，以及与被诊断为除黑色素瘤以外的任何癌症的一级黑色素瘤病例相关的血液个体数量非黑色素瘤皮肤癌。诊断出患有胰腺癌的个体数量与突变状态没有独立关联。风险预测模型在训练数据集中的接受者操作特征曲线 (AUC) 下面积为 0.851 (95% CI 0.793, 0.909)，在验证数据集中为 0.745 (95% CI 0.612, 0.877)。该模型是第一个仅使用澳大利亚数据开发和验证的模型，鉴于人群中黑色素瘤的发病率较高，这一点很重要。该模型将有助于有效识别适合在高环境紫外线辐射地区进行遗传咨询和检测的家庭。用户友好的电子列线图可在 www.melanomarisk.org.au 上获得。