个性化参考区间（prRI）已被提议作为一种诊断工具，用于评估具有高度个性化的被测量。在这里，我们使用癌胚抗原（CEA）评估用于癌症检测的 prRI 的临床表现，并将其与参考变化值（RCV）和临床指南推荐的其他标准进行比较（例如连续 CEA 结果（RV25）和临界点为 5μg/L (CP5))。对 19 年来收集的 2,638 名患者的临床和分析数据进行了回顾性评估。总共研究了 15,485 个 CEA 结果。对于每位患者，我们使用基于不同策略组合的计算机算法计算 prRI 和 RCV，以评估所需 CEA 结果的数量，考虑参考区间的一两个限制以及所使用的个体内生物变异估计 (CVI)：（a ) 公开可用 (CVI-EU)，(b) 使用间接方法计算的 CVI (CVI-NOO) 和 (c) 内部 BV (CVP)。对于每个确定超出 prRI 范围、超过 RCV 间隔、RV25 或 CP5 的新结果，我们搜索了在测试后 3 个月和 12 个月时确定肿瘤存在的记录。计算了每种策略的敏感性、特异性和预测能力。使用 CVI-EU 得出的 PrRI 方法，参考区间的两个限值在所有评估标准的 3 个月和 12 个月时均达到最佳敏感性 (87.5%) 和 NPV (99.3%) 。每个患者只需 3 个结果就足以计算达到此诊断性能的 prRI。PrRI 方法可能是在患者主动监测期间排除新肿瘤学发现的有效工具。© 2024 Walter de Gruyter GmbH，柏林/波士顿。

Personalized reference intervals (prRI) have been proposed as a diagnostic tool for assessing measurands with high individuality. Here, we evaluate clinical performance of prRI using carcinoembryonic antigen (CEA) for cancer detection and compare it with that of reference change values (RCV) and other criteria recommended by clinical guidelines (e.g. 25 % of change between consecutive CEA results (RV25) and the cut-off point of 5 μg/L (CP5)).Clinical and analytical data from 2,638 patients collected over 19 years were retrospectively evaluated. A total 15,485 CEA results were studied. For each patient, we calculated prRI and RCV using computer algorithms based on the combination of different strategies to assess the number of CEA results needed, consideration of one or two limits of reference interval and the intraindividual biological variation estimate (CVI) used: (a) publicly available (CVI-EU), (b) CVI calculated using an indirect method (CVI-NOO) and (c) within-person BV (CVP). For each new result identified falling outside the prRI, exceeding the RCV interval, RV25 or CP5, we searched for records identifying the presence of tumour at 3 and 12 months after the test. The sensitivity, specificity and predictive power of each strategy were calculated.PrRI approaches derived using CVI-EU, and both limits of reference interval achieve the best sensitivity (87.5 %) and NPV (99.3 %) at 3 and 12 months of all evaluated criteria. Only 3 results per patients are enough to calculate prRIs that reach this diagnostic performance.PrRI approaches could be an effective tool to rule out new oncological findings during the active surveillance of patients.© 2024 Walter de Gruyter GmbH, Berlin/Boston.