DISCUSSION
In clinical localised prostate cancer, 67% to 87% of the tumours are detected multifocally at the time of diagnosis. The index tumour, which is the lesion with the highest Gleason score or the highest volume, is the determinant for the prognosis [9,10]. Therefore, it is crucial to detect the index tumour even in multifocal disease.
Besides the local staging of prostate cancer, the first steps towards identifying index tumours have also started with the increasing use of mpMRI. In a meta-analysis including 9796 patients, pooled data for extraprostatic extension, seminal vesicle invasion and overall stage T3 detection showed sensitivity and specificity of 57% and 91%, 58% and 96%, and 61% and 81% for mpMRI, respectively [11] Contrary to local staging data, in a Cochrane meta-analysis which compared mpMRI to template biopsies in biopsy-naive and repeat-biopsy settings, mpMRI had a pooled sensitivity of 91% and a pooled specificity of 37% for ISUP grade > 2 cancers, and had a pooled sensitivity of 95% and a pooled specificity of 35% for ISUP grade > 3 cancers, respectively [12]. Jesse et al. examined the RP specimens of 122 patients and found the lesion-based sensitivity of mpMRI to be 47%. It was demonstrated that the sensitivity of mpMRI was 72% for lesions of >1 cm and 72% for tumours with an ISUP grade group of ≥ 2 in subgroup analysis. The sensitivity of mpMRI was also found to be related to tumour size, Gleason grade, and prostate volume [13]. In our study, high volume and PI-RADS score of the lesion, high PSA, and high PSAD were found to increase the sensitivity of mpMRI, although this was not statistically significant.
Hartenbach et al. [14] designed a prospective, lesion-based study in which 128 lesions were assessed side by side after matching the angulation of the histological step sections and the MRI slices using PET with 18F -ethyl choline combined with endorectal MRI. Assessments were correlated with the histologically dot-marked tumour lesions. Sensitivities and specificities, respectively, were found to be 48% and 40% for endorectal MRI, 71% and 42% for18F-ethyl choline PET, and 66% and 82% for combined18F-ethyl choline PET and endorectal MRI. Combined18F-ethyl choline PET and endorectal MRI had significantly higher diagnostic accuracy compared with the individual modalities.
In studies investigating the diagnostic accuracy of PET/MRI in the literature, we see that there are many differences in terms of the methodology and radiotracer used. Albayati et al. used a lesion-based approach in a study including 22 patients with 41 lesions. They created two models of dichotomisation for the calculation of lesion-based sensitivities and specificities using transrectal biopsy and RP specimens. The results were dichotomised by defining as malignant either a PI-RADS score of 4–5 and a PET score of 4–5 or a PI-RADS score of 3–5 and a PET score of 3–5. Sensitivities and specificities, respectively, of mpMRI and 68Ga-PSMA PET/MRI were 59% vs 88% and 67% vs 100% for dichotomisation scores 1–3 vs 4–5 and 94% vs 94% and 23% vs 56% for dichotomisation scores 1–2 vs 3–5 [15].
Table 5 summarises the studies in which region-based evaluations were applied in terms of evaluated regions per patient, PET/MRI tracers in terms of sensitivities and specificities of mpMRI and PET/MRI [5,16-20]. Radical prostatectomy specimens were used for histopathological evaluation in all these studies. The evaluation was performed by simply dividing the prostate into regions, except for in the study designed by Hicks et al. who used two different 30-region-based approaches. In addition to the simple region-based approach known as ‘raw stringent’, Hicks et al. used radiological–pathological correlation, which was classified as tumour being present in the same region or any of the immediately adjacent regions, known as the ‘neighbouring approach’. Hicks et al. found that68Ga-PSMA PET/MRI was significantly superior to the mpMRI in terms of sensitivity in both raw stringent and neighbouring approaches (67% vs 42% and 74% vs 50%, respectively). The mpMRI was significantly superior to the 68Ga-PSMA PET/MRI in terms of specificity using the raw stringent approach (79% vs 71%, respectively). However, there was no significant difference in terms of specificity using the neighbouring approach (90% vs 88%, respectively) [18].
In our study, which was a simple four-region-based study,68Ga-PSMA PET/MRI had higher sensitivity and specificity compared with mpMRI, although this was not statistically significant (60.8% vs 55.7% and 94.3% vs 91.8%, respectively). However, combined imaging had significantly higher diagnostic accuracy compared with 68Ga-PSMA PET/MRI and mpMRI (change in AUC: 0.084 and 0.046, p < 0.001 and p = 0.028, respectively). In the study by Hicks et al., the tumours, which were detected with an SUVmax of ≥ 6.9 on68Ga-PSMA PET/MRI, had a Gleason score of ≥7 [18]. In our study, we determined the cut-off value for SUVmaxto be 7.1 in predicting a Gleason score of ≥ 7. However, there were two patients, who had a Gleason score of 3+3, with an SUVmaxof 9.5 and 12.
In a study conducted by Lee et al., PET/MRI was performed using both18F- FDG and 18F-choline as radiotracers. Using a six-region-based approach, it was shown that18F-choline PET/MRI had higher sensitivity and specificity compared with 18F-FDG PET/MRI [17]. Eiber et al., who designed a six-region-based evaluation study, showed that both mpMRI and 68Ga-PSMA PET/MRI had significantly better diagnostic performance in large tumours (>5 mm) compared with small tumours (≤5 mm) [5]. In our study, we observed that the sensitivity increased using both methods with a lesion size of >15 mm and lesion volume of >1.5 ml, and this increase was more prominent in68Ga-PSMA PET/MRI. However, we did not find a statistically significant difference between the lesion size and lesion volume groups in terms of diagnostic accuracy for both imaging modalities. Eiber et al. also showed that mpMRI had higher sensitivity in PI-RADS 3 lesions compared with 68Ga-PSMA PET/MRI [5]. Contrary to this finding, in our study,68Ga-PSMA PET/MRI had higher sensitivity in PI-RADS 3 lesions compared with mpMRI (34.5% vs 55.2%, respectively). In combined usage, we showed that sensitivities increased significantly in terms of all parameters.
In a meta-analysis by Li et al., which included nine studies and 353 patients, PET/MRI had significantly higher diagnostic accuracy compared with mpMRI. Sensitivities were found to be 60% vs 78% and specificities to be 90% vs 88% for mpMRI and PET/MRI, respectively. However, when considering the results of this meta-analysis, it should be noted that lesion-based and region-based studies using different radiotracers were evaluated together [21].