口腔白斑（OLK）是一种常见的潜在恶性疾病。早期识别恶性肿瘤的可能性可以更好地治疗 OLK 并预测口腔鳞状细胞癌 (OSCC) 的发展。然而，目前还没有有效的生物标志物来评估 OLK 的恶性肿瘤风险。基因组拷贝数改变（CNA）是基因组中复杂的染色体结构变异，已被确定为多种癌症的潜在生物标志物。本研究旨在通过拷贝数分析建立 OLK 恶变风险的预测模型。对来自多个学术中心的总共 431 个长期随访（中位随访 67 个月）的 OLK 样本进行了 CNA 分析。 CNA事件随着增生、轻度不典型增生、中度不典型增生和重度不典型增生的严重程度而增加。后来发展为 OSCC 的 OLK 患者比未发展为 OSCC 的 OLK 患者存在更多的 CNA 事件。多因素Cox回归分析显示，CNA高分组OLK恶变风险高于CNA低分组（P < 0.001）。开发了 CNA 评分模型来准确预测预后（受试者工作特征曲线下面积 [AUC] = 0.879；95% 置信区间 [CI]，0.799-0.959），并使用来自 2 个外部中心的数据进行验证（AUC = 0.836） , 95% CI, 0.683-0.989; AUC = 0.876, 95% CI, 0.682-1.000)，并且在评估 OLK 转化风险时，所有这些都表现出比组织病理学分级更好的预测性能。此外，我们对增生、轻度不典型增生、中度不典型增生和重度不典型增生的 4 个 OLK 亚组进行了 CNA 模型，发现 CNA 评分可以准确预测不同亚组的恶变。 CNA评分可能是预测OLK恶变的有用生物标志物。根据 CNA 评分对 OLK 进行亚型分类有助于更好地管理 OLK 并预测 OSCC 的发展。

Oral leukoplakia (OLK) is a common type of potentially malignant disorder. Early identification of the malignancy potential leads to a better management of OLK and prediction of development of oral squamous cell carcinoma (OSCC). However, there has been no effective biomarker to assess the risk of malignancy in OLK. Genomic copy number alteration (CNA) is a complex chromosomal structural variation in the genome and has been identified as a potential biomarker in multiple cancers. This study aimed to develop a predictive model for the malignant transformation risk of OLK by copy number analysis. A total of 431 OLK samples with long-term follow-up (median follow-up of 67 mo) from multiple academic centers were analyzed for CNAs. CNA events increased with the severity of hyperplasia, mild dysplasia, moderate dysplasia, and severe dysplasia. More CNA events were present in patients with OLK who later developed OSCC than in those with OLK who did not. By multivariate Cox regression analysis, the OLK of the CNA scorehigh group showed an increased risk of malignant transformation than the CNA scorelow group (P < 0.001). A CNA score model was developed to accurately predict the prognosis (area under the receiver operating characteristic curve [AUC] = 0.879; 95% confidence interval [CI], 0.799-0.959) and was validated using data from 2 external centers (AUC = 0.836, 95% CI, 0.683-0.989; AUC = 0.876, 95% CI, 0.682-1.000), and all of them showed better prediction performances than histopathological grade in assessing the transformation risk of OLK. Furthermore, we performed CNA models among 4 subgroups of OLK with hyperplasia, mild dysplasia, moderate dysplasia, and severe dysplasia and found that CNA score can accurately predict malignant transformation of different subgroups. CNA score may be a useful biomarker to predict malignant transformation of OLK. Subtyping of OLK by the CNA score could contribute to better management of OLK and predicting development of OSCC.