Dynamic ⁶⁸Ga-DOTA PET/CT and Compartmental Modelling to Non-invasively Estimate the Glomerular Filtration Rate

Introduction: The determination of the glomerular filtration rate (GFR) is decisive for a variety of clinical issues such as ensuring an adequate renal function prior to radioligand therapy. Renal scintigraphy using radiotracers that are almost exclusively extracted by glomerular filtration instead of active tubular secretion allow the assessment of urine efflux as well as the determination of the GFR. However, repeated blood sampling over a period of several hours is required for an accurate GFR estimation. In contrast, dynamic renal PET imaging using glomerular filtrated PET tracers like ⁶⁸Ga-labelled 1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetraacetic acid (⁶⁸GA-DOTA) bears the potential to estimate the GFR without venous blood sampling by compartmental kinetic modelling. We here present the to our knowledge first investigation of the feasibility of a GFR estimation by single-compartmental-modelling using dynamic ⁶⁸Ga-DOTA PET/CT data. The aim of the study was to correlate the PET-derived GFR_PET with the clinically established serum creatinine-derived GFR_crea and to compare the visual assessment of urinary obstruction with renal scintigraphy results.

Methods: Data sets of 10 patients who underwent dynamic ⁶⁸Ga-DOTA PET imaging prior to ¹⁷⁷Lu-DOTATOC or ¹⁷⁷Lu-PSMA therapy were included. The mean applied activity was 112.2 MBq of ⁶⁸Ga-DOTA. Dynamic list mode PET data were acquired for 30 min starting at the time point of injection using a Siemens Biograph Vision PET/CT system; the data were resampled and reconstructed in 12 frames of 30s and 18 frames of 90s. A single compartmental-model (Fig. 1) was used to estimate the influx rate constant K1 and the tissue blood volume fraction v_B from the tracer kinetic in the functional renal cortical volume V_RC. V_RC was manually segmented in the dynamic PET images. GFR_PET was calculated as the product of K1, V_RC, and the vascular volume fraction (1-v_B). The arterial input function was determined from the PET signal in the abdominal aorta. GFR_crea was calculated using the established CKD-EPI-formula. The Pearson correlation coefficient was determined to describe the association between GFR_PET and GFR_crea in a linear regression model; the confidence interval (CI) was estimated by bootstrapping with 1000 replicates. Time-activity-curves were visually assessed for urinary obstruction and compared with previously determined renal scintigraphy results.

Results: In 8/10 patients with undisturbed urinary efflux in the visual analysis of the time-activity-curves (Fig. 3), the reproducibility analysis revealed a good agreement and a linear correlation between GFR_PET and GFR_crea with a Pearson coefficient of R² = 0.97 (95% CI: 0.7685 - 0.9987, Fig. 4). In 2/10 patients with urinary obstruction, the determination of GFR_PET led to a significant underestimation indicated by a GFR_PET/GFR_crea fraction of 0.32 and 0.29, respectively. These were the same 2/10 patients who were diagnosed with urinary obstruction in previous renal scintigraphy examinations.

Conclusions: The preliminary data suggest that a non-invasive accurate GFR estimation by single-compartmental-modelling of dynamic ⁶⁸Ga-DOTA PET data is feasible in patients without urinary obstruction. Further investigation, such as a validation against a clinical gold standard, is warranted. The visual assessment of urine outflow revealed comparable results with routine renal scintigraphy.