Page 216 - Quantitative Imaging of Small Tumours with Positron Emission Tomography

with high-volume mCPRC who underwent both [18F]DCFPyl and [18F]FDHT PET before and during systemic treatment. For both tracers, we demonstrated that there was a large tracer sink effect over time, due to substantial increases in tumour volumes. Moreover, changes in TBR evidently correlated with the clinical progression, whereas changes in SUV erroneously indicated stable disease. We also evaluated whether application of PVC in compliance with EARL2 affected these conclusions, but found no substantial change. As we previously discussed, use of TBR for response assessment is often unwanted due to the added noise of blood activity measurements, especially when PVC is applied. However, despite its worse precision, TBR did accurate reflect the true progressive disease during treatment, in contrast with SUV (Figure 9.2 and 9.4). Hence, this clinical example illustrates that, in some cases, high accuracy (use of TBR) is more important than high precision (use of SUV), even in response assessment. Finally, we must acknowledge that the progression quantified on both [18F] DCFPyL and [18F]FDHT PET was confirmed by basic prostate specific antigen (PSA) measurements. Initial studies on response assessment with PSMA PET have shown a similar correlation between PSA response and PET response, warranting further determination of the true added value of PET-based response assessment in metastatic prostate cancer (45,46). Here, the advantage PET-based assessments will be capturing interlesional heterogeneity of response. PSMA-expression quantified on PET as novel biomarker? In Chapter 6 we validated and optimized the use of PET radiomics from [18F] DCFPyL for risk stratification in primary prostate cancer. In this analysis, we found that these radiomics features were highly predictive when optimized according to tumour delineation and PVC, and that these features may outperform basic PET parameters. In Chapter 10 these results were briefly discussed with respect to clinical relevance for the urologist community. We summarized results from our and two other studies on PSMA-expression quantified on PET for prediction of lymph node involvement (LNI), high Gleason score, and/or extracapsular tumour extension (47,48). Interestingly, all studies observed high AUCs for prediction of LNI, high Gleason score, and (in our study) extracapsular tumour extension. We also benchmarked the radiomics AUCs to clinical prediction models, and found that the [18F]PSMA radiomics outperformed these models. All taken together, PSMA-expression quantified on PET is a very promising new player in the group of biomarkers in primary prostate cancer (e.g. PSA, Gleason score). However, 11 Summarizing discussion 215