Digital vs. analog PET/CT: intra-subject comparison of the SUVmax in target lesions and reference regions

The use of digital PET/CT is expected to increase in the following years due to improved image quality, volumetric resolution and sensitivity gains, as well as faster timing resolution compared to analog PET/CT. Such improvements are based on the development of digital photon counting technology, with a 1:1 coupling between scintillation crystal and detector, and a faster time of flight technology [1‐7].

Progressive introduction of digital PET/CT will raise the question of whether follow up PET/CT studies can be interchangeable between digital and analog systems, especially in oncological patients in whom SUVmax values are often used to evaluate treatment response.

The aim of our study was to assess whether digital photon counting technology in digital PET/CT influences the quantification of SUVmax in target lesions and regions of reference in comparison to analog PET/CT in the same patient. Such comparison is necessary before an interchangeable use of either system in follow-up studies.

Table 1 shows mean SUVmax values of the total sample, regardless of the sequence order of PET/CT acquisition. Mean SUVmax of the target lesions and mediastinal blood pool obtained with the digital system were significantly higher than with the analog system, whereas mean SUVmax of the liver did not significantly differ between both systems.

Table 2 shows mean SUVmax values of group 1 (and degree of difference). Mean SUVmax of the target lesions and mediastinal blood pool were significantly higher in the digital system than in the analog one, whereas mean SUVmax of the liver did not significantly differ between both systems (Fig. 1).

Table 3 shows mean SUVmax values of group 2 (and degree of difference). Mean SUVmax of the target lesions and mediastinal blood pool were significantly higher in the digital system than in the analog one, whereas mean SUVmax of the liver did not significantly differ between both systems (Fig. 2).

Table 4 shows the degree of significance of the analog-to-analog and digital-to-digital SUVmax comparison between groups 1 and 2. No significant differences were found between groups except for mean SUVmax in target lesions with the analog PET/CT (group 1 [6.64 ± 4.71] vs group 2 [10.16 ± 7.76]; p = 0.024).

Time delay between both studies in group 1 was 50.38 ± 16.31 min, and in group 2 it was 43.71 ± 24.37 min (P = 0.184). Simple linear regression analysis (R²) between acquisition time delay and SUVmax difference in target lesions was 0 in group 1 and 0.29 in group 2.

Our results consistently show a higher mean SUVmax value with the digital PET/CT as compared to the analog PET/CT. When the analog PET/CT was done first, significant differences were found in the mean SUVmax of target lesions and mediastinal blood pool between both systems, whereas no significant differences were found in the liver. When the digital PET/CT was done first, we also found significant differences in mean SUVmax of target lesions and mediastinal blood pool, but no significant differences in the liver.

We believe that the higher SUVmax of target lesions in the digital scanner, when the analog study had been done before, was due to the higher sensitivity of the digital system plus the effect of a delayed increased uptake and retention (mean percentage difference −50.93). On the other hand, when the digital PET/CT was done first, the mean SUVmax difference and percentage difference were of lower magnitude, probably because in such situations the effect of a delayed uptake played in favour of the analog system. Nevertheless, a delayed uptake effect was not sufficient to produce higher SUVmax values than the digital PET/CT. Also, the effect of the delayed uptake in the analog system is probably the reason for the significant difference in target lesions of group 1 vs 2 when performing the analog-to-analog comparison.

The reasons why the digital PET/CT has a higher maximum count rate, and thus sensitivity, can be explained by the incorporation of a digital photon count rate, a 1:1 coupling between detectors and scintillation crystals and a faster TOF technology [2‐5]. The digital photon counter detector in the digital PET/CT consists of 3200 single photon avalanche diodes (cells). These cells are coupled to a digital photon counter, which produces a binary count from the individual breakdowns of each single photon avalanche diode. This technology eliminates the need for the amplification of the readout to produce a summed analog signal and therefore the need of an analog-to-digital processing (required in traditional “analog” PET/CT systems) [3, 5‐7]. These changes have achieved a very low dead time, a higher count rate time and thus, a better timing resolution with a higher sensitivity [4, 5].

It is well known that the effect of a delayed increased FDG uptake occurs in tissues with high glycolysis such as tumours that have an increased expression of glucose transporters and hexokinase activity and therefore, an increased production of FDG-6-phosphate which is trapped inside the cells [8, 9]. As more ¹⁸F-FDG is captured by the tumour from the interstitial and intravascular activity, more FDG-6-phosphate is synthesized and trapped, thus increasing the SUVmax [8, 9].

However, it is also known that delayed increased FDG uptake does not occur in all tumours. Tumour heterogeneity such as proliferation rate, hypoxia and blood flow play a part in the effect of a delayed uptake. Tumours with an initial low FDG uptake show less prominent delayed increased uptake. This has been proven for breast, small lung nodules and some pancreatic cancers [8]. Thus, some of our findings may not have been influenced by imaging time and can only be explained by differences in technology detection.

Delayed increased uptake of 18F-Choline has also been reported in malignant lesions of prostate cancer [10‐14] as seen with the ten prostate cancer patients included in our analysis. Oprea-Lager et al. studied lymph node uptake in 25 prostate cancer patients by early and delayed imaging and later assessed the malignancy by histopathology and/or follow-up studies. All of the benign nodes showed a decreased uptake, whereas 95% of the pelvic lymph nodes assessed as malignant showed a stable or increased uptake. They suggest that this pattern can be used as an indicator of malignancy with a positive predictive value of 97% [10]. Similar findings were reported in a study of Beheshti et al., where false positive lymph nodes showed a rapid washout on the dynamic images and a decreased SUVmax in delayed images, whereas malignant lymph nodes showed a constant or increased uptake pattern [11]. Finally, Kwee et al. found that dominant malignant regions in the prostate (histologically proven after prostatectomy) showed a significant increase of SUVmax between initial and delayed scans [12].

Because the digital PET/CT is a recent technology, there are few studies comparing it with the traditional analog PET/CT systems. Nguyen et al. studied 21 oncological patients, with a total of 52 representative lesions. All of them underwent the analog PET/CT first and the digital afterwards. They found that the SUVmax difference in the representative lesions was 36% higher for the digital system [4]. This is similar to our findings, where the SUVmax difference values were also higher with the digital system. The aforementioned study did not find a statistical association between the SUVmax difference (comparing GeminiTF and Vereos systems) and acquisition time delay [4]. Likewise we did not find any statistical association with the acquisition time delay when the analog PET/CT was performed first (R² = 0). When the digital PET/CT was performed first, we found only a very small association (R² = 0.29) between acquisition time delay and the SUVmax difference in target lesions. To our knowledge, there are no studies where the digital PET/CT is performed first.

Chin et al. studied the SUV changes in normal tissue imaging at 1 and 3 h post FDG injection in 99 patients [9]. They found no significant activity differences in liver and lung over time, hypothesizing that the absence of significant differences could be due to an integrated similar compartmental influx-efflux [9]. It could also be that the high levels of glucose-6-phosphatase in the liver should result in a reduction in the levels of FDG-6-phosphate that it contains. This could also explain our findings in the liver, where no significant differences were found between systems or acquisition times.

Even though the difference of the mean SUVmax in mediastinal blood pool was not very high, we did find significant differences between both systems. Despite vascular activity being expected to decrease in time due to the continuous clearance of FDG by the kidneys [8, 9, 15], the SUVmax was always higher with the digital system (regardless of the sequence order of PET/CT acquisition). This finding might be in part explained by the better resolution and count rate of the vessel wall by the digital PET.

The main limitation of our study is the impossibility of doing both studies at the same time, so the influence of delayed increased uptake cannot be ruled out. Another limitation is the use of a different number of iterations during the image reconstruction. With higher iterations, noise increases and therefore the SUVmax also increases [16‐18]. Nevertheless, the analog PET/CT was the one which used more iterations and that did not result in a higher SUVmax. In addition, point spread function modelling was used in the reconstruction algorithm with the digital PET/CT. This reconstruction algorithm is known to improve signal-to-noise ratio and lesion detectability, but it also can increase the SUVmax, especially in small lesions [18, 19]. Finally, this is a heterogeneous series of patients with various primary tumours, disease stage and treatment conditions. Furthermore, ten studies were done with Choline, which increases the heterogeneity of the sample. Although this small group of patients studied with Choline does not allow a proper separate statistical analysis, a similar trend to FDG was observed, which indicates that the increased SUVmax obtained in the digital system is not limited to FDG. Future comparisons in more selected groups of patients may bring additional information on the relative advantages and performance of the digital vs analog PET/CT.

Improved photon counting technology in the digital PET/CT, and the effect of delayed increased uptake and retention significantly increases SUVmax values. This has to be taken into account before interchangeable use of either system in follow up studies.