Abstract
97
Objectives: The ability to quantify β-cell mass (BCM) in vivo in type 1 diabetes mellitus (T1DM) patients would provide more accurate evaluation of disease progression and efficacy of therapies, such as islet transplantation or pharmacologic intervention. We previously published a study demonstrating the potential utility of positron emission tomography (PET) imaging with 3,4,4a,5,6,10b-hexahydro-2H-naphtho[1,2-b][1,4]oxazin-9-ol (11C-(+)-PHNO), a dopamine D2/D3-receptor agonist with 25-50-fold higher affinity for D3 receptors, to differentiate between healthy control (HC) and T1DM subjects (Bini, et al, JNM, in press). In the current study, we examined a larger cohort of HC and T1DM subjects and compared kinetic analysis methods to assess possible quantitative measures to differentiate between HC and T1DM subjects. METHODS. Ten subjects (5 HC and 5 T1DM) underwent PET/CT imaging after injection of 11C-(+)-PHNO. Mean injected activity was 310±54 MBq and mean injected mass was 0.02±0.01 µg/kg. Subjects were scanned for 2 h with 1-bed position centered over the pancreas. Regions-of-interest were drawn manually for the pancreas and spleen on a summed 0-10 min PET image and applied to the dynamic PET images to generate time-activity-curves (TACs). SUV and SUVR (using spleen as reference) were calculated at 20-30 min and 60-70 min for all scans. Binding potential (BPND) was calculated using SRTM (spleen as reference) with 30, 70 or 120 min of the TACs and the Logan reference model (t[asterisk]=0, 10 or 20 min, and tmax=30, 70 or 120 min).
Results: Differences between HC and T1DM were detectible with virtually all pancreatic outcome measures. Mean pancreas SUV at 20-30 min was lower in T1DM subjects versus HCs (HC: 14.7±2.4; T1DM: 9.9±3.8, 33% difference; p=0.05). T1DM subjects demonstrated regional variations of reduced SUV in pancreas head (34%; p=0.07), pancreas body (49%; p=0.01), and pancreas tail (33%; p=0.06). No significant difference was seen in mean spleen SUV at 20-30 min between groups (HC: 2.1±0.5; T1DM: 2.0±0.4; p=0.67). Mean pancreas SUVR-1 at 20-30 min showed a similar reduction between groups (HC: 6.2±1.0; T1DM: 3.9±1.3; 37% difference; p=0.02) versus HCs. Mean pancreas SUVR-1 at 60-70 min showed a similar reduction in T1DM versus HCs (HC: 3.4±0.9; T1DM: 2.1±0.5; 38%; p=0.04), as seen at shorter durations. BPND calculated using SRTM from 0-70 min (HC: 3.2±0.5; T1DM: 2.2±0.7; p=0.04) demonstrated a 31% reduction in T1DM. BPND calculated using the Logan reference model (t[asterisk]=0 min, tmax=70 min) demonstrated a 36% reduction in T1DM (HC: 3.1±0.5; T1DM: 2.0±0.6; p=<0.01). SUVR-1 from 60-70 min was more highly correlated with SRTM BPND from 0-70 min (y=1.0904x-0.1736, R2=0.79, p<0.01) and BPND from the Logan reference model (t[asterisk]=0 min, tmax=70 min; y=1.0757x+0.0383, R2=0.75, p=<0.01) than SUVR-1 at shorter durations (tmax=30 min). In general, SRTM and Logan reference methods provided very similar BPND values. BPND calculated using SRTM with tmax=30 min and tmax=70 min were highly correlated (y=0.99x+0.0227: R2=0.99, p<0.01). BPND calculated using SRTM 0-30, 0-70 and 120 min provided similar values; however, TAC fits were progressively worse at longer durations. Logan plots with t[asterisk]=20 min and tmax=30 min did not have enough data points to fit the model. Changing of t[asterisk] between 0, 10 and 20 min had minor impact on BPND when Logan reference plots used tmax= 70 or 120 min.
Conclusions: These studies suggest that multiple modeling techniques and SUVR can be useful to provide quantitative outcome measures for 11C‑(+)-PHNO, a D3-receptor-preferring agonist PET radioligand, to differentiate BCM in T1DM and HCs. Longer scans did not provide a quantitative advantage, i.e., scan times of 30-70 min are sufficient. The good agreement between SUVR-1 and BPND suggests that this simplified measure may also be useful.