Population-Based Input Function Modeling for [18F]FMPEP-d2, an Inverse Agonist Radioligand for Cannabinoid CB1 Receptors: Validation in Clinical Studies

BackgroundPopulation-based input function (PBIF) may be a valid alternative to full blood sampling for quantitative PET imaging. PBIF is typically validated by comparing its quantification results with those obtained via arterial sampling. However, for PBIF to be employed in actual clinical research studies, its ability to faithfully capture the whole spectrum of results must be assessed. The present study validated a PBIF for [18F]FMPEP-d2, a cannabinoid CB1 receptor radioligand, in healthy volunteers, and also attempted to utilize PBIF to replicate three previously published clinical studies in which the input function was acquired with arterial sampling. MethodsThe PBIF was first created and validated with data from 42 healthy volunteers. This PBIF was used to assess the retest variability of [18F]FMPEP-d2, and then to quantify CB1 receptors in alcoholic patients (n = 18) and chronic daily cannabis smokers (n = 29). Both groups were scanned at baseline and after 2–4 weeks of monitored drug abstinence. ResultsPBIF yielded accurate results in the 42 healthy subjects (average Logan-distribution volume (VT) was 13.3±3.8 mL/cm3 for full sampling and 13.2±3.8 mL/cm3 for PBIF; R2 = 0.8765, p1 receptors in alcoholic subjects, without significant change after abstinence. However, a small PBIF-VT bias of −9% was unexpectedly observed in cannabis smokers. This bias led to substantial errors, including a VT decrease in regions that had shown no downregulation in the full input function. Simulated data showed that the original findings could only have been replicated with a PBIF bias between −6% and +4%. ConclusionsDespite being initially well validated in healthy subjects, PBIF may misrepresent clinical protocol results and be a source of variability between different studies and institutions.背景 基于人群的输入函数（Population-based input function, PBIF）或许可作为定量正电子发射断层扫描（Positron Emission Tomography, PET）成像中全血采样的有效替代方案。PBIF的有效性通常通过将其定量结果与动脉采样所得结果进行对比来验证。然而，若要将PBIF应用于实际临床研究，必须评估其能否如实覆盖所有结果区间。本研究针对大麻素CB1受体放射性配体[18F]FMPEP-d2，在健康志愿者中验证了一款PBIF，并尝试利用该PBIF复现三项此前已发表的临床研究——这些研究均采用动脉采样获取输入函数。 方法 本研究首先基于42名健康志愿者的数据构建并验证了该PBIF。使用该PBIF评估[18F]FMPEP-d2的重测变异性，随后对酒精成瘾患者（n=18）及长期每日大麻吸食者（n=29）的CB1受体进行定量分析。两组受试者均在基线期及接受2~4周监督下药物戒断后接受扫描。 结果 在42名健康受试者中，PBIF的定量结果准确可靠：全采样组的平均Logan分布容积（VT）为13.3±3.8 mL/cm³，PBIF组为13.2±3.8 mL/cm³；决定系数R²=0.8765，p<0.001。酒精成瘾患者的CB1受体定量结果在戒断后无显著变化。然而，在长期每日大麻吸食者中意外观察到PBIF-VT存在-9%的小幅偏差。该偏差引发了显著误差，包括在全输入函数分析中未出现下调的脑区中出现VT值被错误降低的情况。模拟数据显示，仅当PBIF偏差处于-6%至+4%区间时，才可复现原始研究结果。 结论 尽管PBIF最初在健康受试者中得到了良好验证，但其仍可能无法准确反映临床研究方案的结果，并成为不同研究与机构间结果存在差异的潜在来源。