Population-based AIF for the quantification of fatty acid amine hydrolase density in PTSD by [11C]CURB PET

Introduction: Positron emission tomography (PET) imaging with [¹¹C]CURB offers in vivo measurements of fatty acid amine hydrolase density (Boileau et al., 2015 – JCBFM). This endocannabinoid catabolizing enzyme has been identified as a therapeutic target of interest for post-traumatic stress disorder (PTSD) due to its role in the response to stress (Mayo et al., 2022 – Biol. Psychiatry; Green et al., 2022 – Biol. Psychiatry; Hill et al., 2023 – Psychol. Med.). Quantification of [¹¹C]CURB by PET requires measuring the arterial input function (AIF); however, arterial sampling can be challenging and alternatives, such as the population-based AIF (PBIF), are needed when sampling issues occur. A PBIF is obtained by calibrating the average AIF from a given population using parameters such as the standardized uptake value (SUV) or manual blood samples (MS) (van der Weijden et al., 2023 – EJNMMI). We hypothesize that a more accurate metabolite-corrected plasma PBIF can be obtained from the MS-scaled blood PBIF by incorporating measurements of parent radiotracer in the plasma-to-blood ratio (PBR). Here we compared and validated three approaches to obtain [¹¹C]CURB PBIFs for a cohort of individuals with PTSD.

Methods: The PET data used in this secondary analysis were acquired from individuals with PTSD (n = 26; mean age 41 ± 12 years, 17/9 females/males) following the injection of 390 ± 30 MBq of [¹¹C]CURB. Arterial blood sampling was performed automatically for approx. 20 minutes and manually throughout the PET acquisition. Metabolite-corrected plasma AIFs were generated using in-house MATLAB scripts. Delay- and dispersion-corrected AIFs were converted to SUV, aligned, and averaged (leave-one-out; Figure 1A-B). Blood and plasma PBIFs for each subject were obtained by scaling the average AIFs by the SUV (method 1) or MS (method 2). For method 3, PBR was fit to a mono-exponential function (f_PBR (t); Figure 1C) and the MS-scaled blood PBIF (C_b (t)) was used to obtain the plasma PBIF ( C_p (t) = f_PBR (t) ∙ C_b (t) ). Leave-one-out cross-validation was performed to evaluate the PBIFs. For this, time-activity curves were extracted from 9 volumes-of-interest (VOIs) and fit (PMOD, v4.203) to extract the composite parameter λk₃ ( λ = K₁/k₂ ) (Rusjan et al., 2013 – JCBFM).

Results: Mean area-under-the-curve (AUC) ratios between blood PBIFs and AIFs were 1.04 ± 0.24 (SUV) and 1.01 ± 0.07 (MS). Respective AUC ratios for the plasma curves were 1.05 ± 0.27 and 0.99 ± 0.29. Average AUC ratio of 1.00 ± 0.11 was obtained for the plasma PBIFs generated by incorporating the PBR fitted curve. Fitting results are summarized in Figure 1D; no significant differences were observed. Significant correlations with respect to AIF results were found across VOIs for the three methods (p < 0.01), with average Pearson correlation coefficient of 0.69 ± 0.04, 0.71 ± 0.03, and 0.95 ± 0.01 for methods 1, 2, and 3, respectively. PBIF λk₃ results are plotted against AIF estimates in Figure 1E-G. PBR PBIF λk₃ estimates presented the best agreement to respective AIF values (y = (1.02 ± 0.03) x + (0.001 ± 0.005); R² = 0.90 ± 0.02; Figure 1G).

Conclusions: A remarkable agreement to AIF was observed when deriving the plasma PBIF using each subject’s PBR measurements. These results suggest method 3 should be preferred over scaling the PBIF using SUV or blood samples for [¹¹C]CURB PET studies involving individuals with PTSD. Future studies are underway to expand this approach to other cohorts and radiotracers, as well as to investigate more accurate methods when PBR is not available.

• Download figure
• Open in new tab
• Download powerpoint