Timo Soeterik

149 MRI T-stage for LNI Risk Prediction that addition of PI-RADS score improved model discrimination in terms of AUC for both nomograms, increasing from 75% to 86% for Briganti and from 79% to 88% for MSKCC, respectively. 23 Recently, two new nomograms have been introduced, including mpMRI and target biopsy features such as maximum diameter of the index lesion and maximum percentage of tumour involvement in one core. 24,25 Of these, the 2019 Briganti nomogram was recently externally validated showing excellent characteristics, including an AUC of 79% and high agreement between predicted and observed probabilities for risk thresholds below 35%. 26 In their head-to-head comparison, the 2019 Briganti nomogram outperformed the Briganti 2017 and MSKCC 2018 in terms of discrimination, calibration and net benefit. 26 Although these new nomograms potentially enable improved LNI risk prediction due to the addition of mpMRI guided target biopsy, they both include complex features which may not be always available in clinical practice, such as maximum diameter of the index lesion on mpMRI and highest tumour length in millimetres of all biopsy cores taken. 24,25 In addition, more external validation studies are warranted to confirm the accuracy of these new nomograms in external patient populations, as model transportability needs to be adequate to prevent systematic wrong decision-making. Our results do not support the statements in a recent position paper on prostate cancer staging by Paner et al., who suggested that DRE should not be replaced by mpMRI for establishing clinical T-stage. 27 In the present study, mpMRI outperformed DRE in terms of AUC for nomogram-based LNI risk prediction, as the use mpMRI T-stage resulted in higher AUCs for all bootstrap samples. This is most likely the consequence of the main advantage that mpMRI has over DRE for determining local tumour extent, which is the visualisation of the prostate gland as a whole and improved detection of non-organ- confined disease. Our study group has confirmed this in a recent study, as the reported sensitivity for the detection of non-organ-confined disease was significantly lower for DRE compared with mpMRI (12% vs. 51%, p < 0.001). 28 Although our main study results favour the use of mpMRI T-stage for nomogram- based LNI risk prediction, there are arguments against replacing DRE with mpMRI T-stage that should be mentioned. First, disadvantages of MRI include reader interobserver variability and quality differences regarding mpMRI reading. 27 However, a previous study by Angulo et al showed interobserver inconsistency also to be an issue for DRE, resulting in a low ability to reproduce clinical staging by DRE among multiple examiners. 29 Second, use of mpMRI compared with DRE would lead to upstaging of clinical T-stage in one-third of the patients. 28 Although mpMRI can provide valuable prognostic information for specific patients, including those with non-organ-confined disease which was not detected during DRE, the high upstaging rates bear the risk of overstaging and hence overtreatment in patients with genuine low-risk disease. 28 To select patients for ePLND, it remains important to take into account patient’s preferences, age and prognostic tumour parameters other than those included in the nomograms to distinguish the patients who would benefit from additional ePLND, from those in whom this intervention would potentially do more harm than good. 8