It is difficult to compare the diagnostic accuracy of [68 Ga]Ga-DOTA-TOC PET/CT and PET/MRI due to the high sensitivity and specificity of the radiopharmaceutical by itself [13].
[68 Ga]Ga-DOTA-TOC PET/MRI, has certain advantages over conventional PET/CT. High-energy positron emitters such as 68 Ga are especially suitable for PET/MRI. This isotope has a shorter range before its annihilation due to the influence of the magnetic field of MRI, improving the spatial resolution compared to PET/CT [14]. In contrast to PET/CT, in PET/MRI the acquisition of both images is simultaneous. This fact allows for an increase in the acquisition time of PET in PET/MRI, reduces noise, improves the detection sensitivity in small lesions, and avoids artifacts in co-registration caused by respiratory movements, particularly in the upper abdominal organs.
In addition to the superior soft-tissue contrast due to MRI, [68 Ga]Ga-DOTA-TOC PET/MRI offers double functional information, SSR expression by PET acquisition, and water diffusion if the DWI is acquired [15].
The incorporation to the new PET/MRI equipment of the combination of TOF, which reduces image noise and improves contrast, and PSF, which locates the annihilation points along the line-of-response at the true geometric locations, improves the detection of small lesions considerably [16, 17].
Thanks to the information provided by PET, the improved TOF and the PFS, fewer MRI sequences are necessary in PET/MRI, therefore shortening the exploration time. Nevertheless, the examinations are still significantly longer and more uncomfortable than those of PET/CT [18]. This fact, added to the few PET/MRI pieces of equipment available, and their high cost, are its main drawbacks.
The mean activity of [68 Ga]Ga-DOTA-TOC corrected by radioactive decay at the start of PET/MRI acquisition was 36.37% lower than the radioactivity at the start of PET/CT. Despite the lower proportional activity of [68 Ga]Ga-DOTA-TOC in PET/MRI, it detected 14.6% more lesions than PET/CT, meaning that the radiation to be administered in PET/MRI could be adjusted downwards in the future, without affecting its diagnostic performance. Also, the absence of ionizing radiation from CT makes PET/MRI especially feasible in young patients who will require lifelong follow-up, due to active disease or because they are carriers of a genetic syndrome, although the reduction of accumulated radiation is a benefit whose real future impact is still unknown.
The most discrepant locations between PET/CT and PET/MRI were liver and bone. Both in the liver and in the bone, PET/MRI detected 5 more lesions than PET/CT, followed by abdominopelvic lesions (3 lesions), with a millimetric size (mean size 3.73 mm) and always in patients with known metastatic disease. In addition to the technical improvements of PET, these differences are probably due to the higher capacity of MRI in the detection of hepatic liver and bone marrow small lesions [19]. Apart from location, size was a key factor in the detection of new lesions on PET/MRI. With a size ranging between uptakes impossible to characterize morphologically to lesions of maximum 12 mm, the new lesions were below the spatial resolution of PET/CT.
Both SUVmax and TLR were higher on PET/MRI than on PET/CT (38.27 vs 19.45 and 4.12 vs 2.44 respectively). Coura-Filho et al., evaluated the [68 Ga]Ga-DOTA-TATE temporal variation of SUVmax in normal tissues and NETs. They acquired studies before 15 min, between 45–90 min and after 90 min post-injection, concluding that there was stability in SUVmax values over time both in tumors and normal tissues [20]. This study, in addition to the fact that our TLR values also increased proportionally to the SUV, minimizes the possibility that the higher uptake in PET/MRI is due to physiological changes in the biodistribution of the radiopharmaceutical, since the PET/MRI is a later study.
The most pronounced increases in uptake intensity were observed in larger lesions (size 11-20 mm and > 20 mm). However, in these sizes, this variation had no clinical impact, as both PET/CT and PET/MRI showed high uptake intensity (TLR in size 11-20 mm changed from 4.67 to 7.04 from PET/CT to PET/MRI and in size > 20 mm from 2.04 to 4.39). It is in the smallest lesions where, this difference had more diagnostic impact, despite a smaller variation in the uptake intensity between both scanners. For < 5 mm lesions, the mean TLR increased from 0.6 on PET/CT to 1.85 on PET/MRI. This means that the lesions changed from having notably lower uptake than baseline hepatic uptake (on PET/CT), to having intense uptake equal or greater than hepatic uptake on PET/MRI, leading to a change in KS and getting the lesions over the line for suitability of PRRT.
The finding of these changes in the smaller lesions had an important impact on the modifications of the KS. KS was introduced into the functional SSR imaging of NETs, with the aim of semi-quantitatively grading the pathological uptake of lesions. A KS with a value of 0 means no uptake; KS1 very low uptake; KS2 uptake less than or similar to the liver; KS3 uptake greater than liver; KS4 uptake greater than the spleen [21]. This score was initially introduced to assess planar octreotide scintigraphy and was later adapted for PET with [68 Ga] Ga-DOTA-related radiopharmaceuticals. A KS ≥ 3 has important clinical implications. The high expression of SSRs in the functional image means that there is a therapeutic target for “cold” somatostatin analogs or PRRT. In PRRT, the reports recommend a KS ≥ 3 for effective therapy [22]. In our series, the PET/MRI implied an upward elevation of the KS. In up to 60% of the patients, the KS changed in some of the lesions. 30.4% of the evaluated lesions experienced a change in the KS, 18.4% of the lesions increased their KS from 2 on PET/CT to a KS ≥ 3 on PET/MRI and most importantly, a 24.96% of the lesions per patient with multifocal disease displayed a KS ≥ 3 on PET/MRI, that were not detected or showed lower KS on PET/CT. From a theragnostic point of view, these lesions would change from not candidates for SSR-targeted therapies to being suitable for these treatments.
The mean size of the lesion’s changing KS was 11.89 mm. This result implies a reduction in the size of the lesions that can be characterized by [68 Ga]Ga-DOTA-TOC PET/MRI compared to previous studies with PET/CT. Hope et al. described that the change from using [111In]In-octreotide scintigraphy to [68 Ga]Ga-DOTA-TATE PET/CT led to an increase in KS, particularly in lesions smaller than 2 cm. In our study, the change from conventional [68 Ga]Ga-DOTA-TOC PET/CT to PET/MRI technology supposes an increase in the KS of lesions of approximately 1 cm [23].
In 3 of the 25 patients, we consider that PET/MRI modified the clinical management (12% of patients). This result is highly influenced by the reason for patient referral to the scan. 22 of the 25 patients were referred for staging or follow-up. In the context of follow-up, the appearance of a millimetric lesion, or the increase in uptake intensity, was considered as a consequence of a better resolution of the PET/MRI and not as a real progression of the disease. In the context of staging, all the patients in whom there was a mismatch were already metastatic patients, so they already were referred to systemic treatment if required.
The major changes occurred in patients referred for therapeutic decision (changes applied to 2 out of 3 patients), but this was the smallest group of patients. Considering the changes in the KS in our overall series, if this group had been more numerous, we believe PET/MRI would have probably modified the strategy, or at least reinforced the therapeutic decision in more cases.
The third case in which PET/MRI had a clinical impact was patient 24. This patient with a metastatic PHEO, previously received two [131I]MIBG treatments, with persistence of sclerotic bone lesions of a few millimeters, below spatial resolution capacity of PET/CT. In this case, PET/MRI, which in our series has shown great accuracy in quantifying small lesions, confirmed a very low uptake and therefore active disease, so a follow-up was decided.
To our knowledge, no similar series have been published on PGGLs. There are few series comparing in a same-day protocol PET/CT and PET/MRI with [68 Ga]Ga-DOTA related radiopharmaceuticals in NETs, but in most of them, the intensity of uptake is not assessed. Berzaczy et al. published a series comparing [68 Ga]Ga-DOTA-NOC PET/CT and subsequent PET/MRI on the same day in 28 patients, but the study included patients with NETs (with different degrees of differentiation) of small bowel, lung, pancreas, colon, unknown primary location and two PGLs. Berzaczy et al. observed comparable results between PET/CT and PET/MRI with a minor advantage for PET/MRI in terms of overall accuracy and sensitivity. However, in their study Berzaczy et al., observed differences in the SUV between both techniques, but with higher values measured on PET/CT than on PET/MRI, unlike in our series. These results do not fit the evolution of uptake in normal and tumor tissues that has been described previously, and it is probably due to the characteristics of the scanners used. The PET/CT equipment performed in this study incorporated TOF while the PET/MRI did not. Differences between both systems result in overall better image quality with higher background activity and small uptake volumes in the equipment using TOF [24].
Savicki et al. compared PET/CT and PET/MRI with [68 Ga]Ga-DOTA-TOC in 30 patients with proven NETs, demonstrating that out of 70 liver lesions, 10 were not detected by PET/CT, and that all of them measured < 1 cm. These results are similar to those of our protocol [25].
Seith et al. compared retrospectively in 29 patients, fast non‑enhanced abdominal examination protocols in PET/MRI to PET/CT in NETs. They concluded that non-enhanced contrast PET/MRI is an alternative for follow-up examinations in patients with unresectable NET and kidney failure [26].
Beiderwellen et al. studied the usefulness of PET/MRI with [68 Ga]Ga-DOTA-TOC in gastroenteropancreatic NETs in 8 patients, with similar results between both techniques, observing high soft tissue contrast in MRI and special utility in patients with chronic renal failure with whom iodinated contrast could not be used [27]. Hope et al. reported a simultaneous study with [68 Ga]Ga-DOTA-TOC in 10 patients with known liver metastases, with better liver disease detection in PET/MRI, just as in our study, but with improved detection of nodal involvement in PET/CT [28].
In economic terms, Mayerhoefer et al. performed a cost comparison between PET/CT and PET/MRI, in a mixed oncologic patient population, using two measures of effectiveness: the percentage of diagnostic accuracy and changes in clinical management. They observed that PET/MRI led to changes in clinical management or therapy in 8% of patients (+ 13.4% accuracy compared with PET/CT). These results included NETs due to the superior detection of liver metastases and are of a similar range as our results (12% of changes in clinical management). The cost per PET/MRI examination was nearly 50% higher compared with that of PET/CT, but it must be contrasted with the higher diagnostic accuracy of PET/MRI and the ability to affect management and therapy decisions in a significant fraction of patients. The incremental cost-effectiveness ratios of PET/MRI were 14.26 EUR per percent of diagnostic accuracy, and it was 23.88 EUR per percent of correctly managed patient. They concluded that a histology-based triage of patients to either PET/MRI or PET/CT could be meaningful, being one of the reasons that promoted our study [29].
But in addition to the cost, accessibility to PET/MRI should be assessed, considering that it is not available in all medical centers and that it is still a longer and more uncomfortable examination. Therefore, it is necessary to determine which patients will benefit from PET/MRI versus PET/CT, especially when only a 13.4% difference in accuracy has been described between both techniques.
After evaluating our results, based on the lower proportional activity of the radiopharmaceutical used in PET-MRI, the greater discrepancy in certain locations between both scans, the size of the lesions that showed change in quantification and the reason for referral of the patient, the clinical scenarios that could increase the benefit of PET/MRI over PET/CT would be in our opinion (Fig. 7):
-
Young patients, with active disease or carriers of a genetic syndrome, with good tolerance to the MRI examination, which require lifelong surveillance, in whom the ionizing radiation of the CT would be avoided, and the activity of the administered radiopharmaceutical could probably be reduced.
-
Patients in follow-up with liver and bone predominance of small-sized metastatic disease (a few millimetres).
-
Patients undergoing evaluation prior to SSRs-targeted treatment (somatostatin analogs or PRRT), with the presence of small-sized metastatic disease (around 1 cm).
-
Patients with millimetric metastatic disease who have received systemic therapy, and in whom PET/CT cannot determine the presence of active disease.
Within the limitations of the study, the reference standard was based mostly on clinical follow-up and imaging follow-up, and to a lesser extent on the histopathology of surgical specimens and biopsies. Intravenous contrast agents were not used in either PET/CT or PET/MRI. This was solved by using [68 Ga]Ga-DOTA-TOC, which by itself has high accuracy, and can provide information as if it were a contrast agent, in addition to the DWI sequence as described above. All patients underwent PET/CT first and PET/MRI next, which could constitute a bias in uptake quantification. Technological improvements of PET/MRI over PET/CT may have increased the differences in the quantification of lesions. The detection of small lung nodules has traditionally been one of the weaknesses of MRI. However, lung involvement is rare in PGGLs (in our series, 2 out of 25 patients), our patients underwent respiratory monitoring to motion correction, and PET uptake of the lesions facilitated their location, so we did not observe differences between the two tests in this location [30]. Finally, the patients were referred for examination for varied reasons, so their change in clinical management was more difficult to assess. However, given the homogeneity of our series, we believe that the results can be extrapolated to the separate groups, as described in Table 1, according to the reason for referral.