Abstract
262
Objectives: In order to standardize interpretation of PSMA PET, a molecular imaging PSMA scoring system (miPSMA) was recently introduced, considering degree of radiotracer uptake, in relation to reference structures [1]. To allow accurate quantification of radiotracer uptake on PET, several corrections are needed including partial volume effect (PVE) [2]. PVE originates from the finite spatial resolution of the PET and causes underestimation of the measured activity concentration especially for lesions at or below the resolution of the scanner [3]. Clinically, this may become relevant when assessing radiotracer uptake in small lesions. In the setting of staging/restaging prostate cancer (PCa), the identification of all disease sites may have significant implications to tailoring the most appropriate therapeutic approach. The aim of the current study was to investigate the impact of PVE on the miPSMA score in PCa metastases detected in patients with biochemical recurrence with negative or equivocal CT and bone scintigraphy. This was performed as part of an ongoing PSMA PET registry.
Methods: A total of 27 patients who completed 18F-DCFPyL (PSMA) PET/CT scans for assessment of recurrent PCa (mCT40, Siemens Healthcare). Of these patients, 7 where excluded (no metastatic disease), and 1 who has lesions which could not be accurately contoured on the unenhanced CT, resulting in a final cohort of 20 patients with 71 suspected metastases. Of the 71 lesions included, the majority were small lymph nodes (n=68), and the remaining lesions were bone metastases (n=2) and a lung deposit (n=1). All lesions were manually contoured on CT and PET separately, using Mirada Vision workstation (Mirada Medical) and calculated volume (cm3), SUV max and SUV mean recorded. For the purpose of reference, the SUV max/mean of parotid gland, liver and blood pool were also obtained. PET images were corrected by segmented anatomical CT, smoothed by the simulated PET scanner point-spread function. Dividing the smoothed volume by the original mask, while considering the appropriate lesion contrast, provides a correction map, which was transformed into the PET image voxel space and applied to the PET images. The result was a new PET image with the same total activity as the original, but with local, deconvolution-like PVE correction [Fig.1]. The measured vs corrected SUVmax and SUVmean were compared. Furthermore, the prospectively assigned miPSMA score was compared to the score obtained from the corrected dataset. All quantitative and qualitative assessment was correlated for lesion volume.
Results: Of the 71 available lesions, 4 lymph nodes were excluded from final analysis due to PVE correction errors (usually due to proximity to ureter). There were 67 corrected lesions with a measured contoured volume of 0.31 cm3 (median; range: 0.06-10.35 cm3). Before and after PVE correction, SUVmax was 13.0 (median; range 1.8-89.5) vs 29.3 (median; range 6-182.2), respectively; SUVmean was 6.5 (median; range 1.4-31.3) vs 11.7 (median; range 2.9-55.3), respectively; average miPSMA score was 1.98 and 2.53, respectively. The estimated SUVmax correction factor for this cohort was 2.52 [Fig.2]. After PVE correction, the corrected miPSMA score were upgraded in 33 lesions (50%) from score 1 to 2 in 10 lesions (30%); from score 1 to 3 in 4 lesions (12%) and from score 2 to 3 in 19 lesions (58%). When assessing lesions with a volume < 0.3 cm3 (volume of a sphere with a diameter of 8mm), 2/3 of lesions (21/32; 66%) had a higher miPSMA score after correction, compared to 1/3 (12/34; 35%) of lesions with volume ≥ 0.3 cm3. [Fig.3-4] Conclusions: Partial volume effects have an impact on quantitative assessment of metastatic lesions on PSMA PET. For small lesions, and especially those with a volume < 0.3 cm3, this also impacts qualitative assessment, resulting in a higher miPSMA score in nearly 2/3 of lesions. This should be considered in future revised versions of the miPSMA score.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.