联合快速免疫分析法快速诊断肝素诱导的血小板减少症
本研究评估快速免疫分析法(侧流免疫分析法LFIA与乳胶免疫分析法LIA)检测抗PF4/肝素抗体的诊断性能,并在267例推导队列中比较两种基于临床概率联合LIA与化学发光免疫分析法(CLIA)的新算法与已发表的Lausanne算法及Hamilton算法。结果显示LFIA在冻存血浆中漏诊6/30例HIT,LIA灵敏度90.9%、特异度93.5%;采用LIA与CLIA序贯的贝叶斯方法诊断准确性最高,基于回顾性数据LIA/CLIA联合方案可作为前瞻性验证的候选。
要点
- LFIA在冻存血浆中漏诊6/30例HIT,不能用于排除或预测HIT。
- LIA诊断HIT的灵敏度为90.9%,特异度为93.5%。
- 序贯联合LIA与CLIA的贝叶斯方法诊断准确性最高。
核心结论
- LFIA在冻存血浆中不能用于排除或预测HIT。
- 采用两种快速免疫分析法序贯检测抗PF4/肝素抗体的贝叶斯方法对HIT的准确诊断最为有效。
- 基于回顾性数据,LIA/CLIA联合方案是前瞻性验证的候选。
研究设计与样本
- 诊断性能研究,在预试验队列(n=74)和推导队列(n=267)中分析LFIA(STic Expert HIT)与LIA(HemosIL HIT-Ab)的诊断性能。
- 比较两种基于HIT临床概率联合序贯LIA与CLIA(HemosIL AcuStar-HIT-IgG)的新算法与已发表的Lausanne算法(序贯CLIA与PaGIA)及Hamilton算法(同时进行LIA与CLIA)。
主要发现
- LFIA漏诊6/30例HIT。
- LIA的灵敏度为90.9%,特异度为93.5%。
- Lausanne算法正确预测HIT 19/267(7.1%),排除HIT 240/267(89.9%),未解决8/267(3%)。
- 序贯联合CLIA与LIA的算法预测HIT 19/267(7.1%),其中1/19预测错误,排除HIT 236/267(88.4%),未解决11/267(4.1%)。
- 以LIA作为首选检测的算法预测HIT 22/267(8.2%),排除HIT 235/267(88%),未解决9/267(3.4%)。
关键数值与推荐
- LFIA漏诊6/30例HIT。
- LIA灵敏度90.9%,特异度93.5%。
- Lausanne算法:正确预测19/267(7.1%),排除240/267(89.9%),未解决8/267(3%)。
- 序贯CLIA与LIA算法:预测19/267(7.1%),1/19预测错误,排除236/267(88.4%),未解决11/267(4.1%)。
- 以LIA为首的算法:预测22/267(8.2%),排除235/267(88%),未解决9/267(3.4%)。
临床怎么用
- LFIA在冻存血浆中不能用于排除或预测HIT。
- 作者认为,采用两种快速免疫分析法序贯检测抗PF4/肝素抗体的贝叶斯方法对HIT的准确诊断最为有效。
- 作者认为,基于回顾性数据,LIA/CLIA联合方案是前瞻性验证的候选。
局限性
- 基于回顾性数据。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: Early recognition and treatment of heparin-induced thrombocytopenia (HIT) are key to prevent severe complications. OBJECTIVE: To assess the diagnostic performance of rapid immunoassays (IA) in detecting anti-PF4/heparin-antibodies. METHODS: Diagnostic performances of lateral-flow IA (LFIA; STic Expert HIT) and latex IA (LIA; HemosIL HIT-Ab) were analyzed in pilot (n = 74) and derivation cohorts (n = 267). Two novel algorithms based on the combination of HIT clinical probability with sequentially performed LIA and chemiluminescent IA (CLIA; HemosIL AcuStar-HIT-IgG) were compared with published rapid diagnostic algorithms: the "Lausanne algorithm" sequentially combining CLIA and particle-gel IA (PaGIA) and the "Hamilton algorithm" based on simultaneously performed LIA and CLIA. RESULTS: LFIA missed 6/30 HIT. The sensitivity and specificity of LIA were 90.9% and 93.5%. The Lausanne algorithm correctly predicted HIT in 19/267 (7.1%), excluded it in 240/267 (89.9%), leaving 8/267 (3%) cases unsolved. The algorithm sequentially combining CLIA and LIA predicted HIT in 19/267 (7.1%) with 1/19 wrong prediction, excluded it in 236/267 (88.4%), leaving 11/267 (4.1%) cases unsolved. The algorithm employing LIA as a first assay predicted HIT in 22/267 (8.2%), excluded it in 235/267 (88%), leaving 9/267 (3.4%) cases unsolved. Finally, the Hamilton algorithm correctly predicted HIT in 10/267 (3.7%), excluded it in 229/267 (85.7%), leaving 28/267 (10.5%) cases unsolved. CONCLUSION: LFIA cannot be used to exclude or predict HIT when using frozen plasma. A Bayesian approach sequentially employing two rapid immunoassays for anti-PF4/heparin antibodies is most effective for the accurate diagnosis of HIT. Based on retrospective data, the combination LIA/CLIA is a candidate for a prospective validation.
引用
- Rittener-Ruff L, Marchetti M, Matthey-Guirao E, et al. Combinations of rapid immunoassays for a speedy diagnosis of heparin-induced thrombocytopenia. J Thromb Haemost. 2022. PMID: 35808841. doi:10.1111/jth.15811
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。