Timo Soeterik

40 CHAPTER 2 patients undergoing radical treatment following AS discontinuation were comparable between the PRIAS-eligible and PRIAS-ineligible subgroups (73% vs 73%). Some 11% of PRIAS-eligible patients who discontinued AS switched to watchful waiting, while this percentage was 14% in the PRIAS-ineligible group. Given the comparable percentages in both groups, we concluded that the presence of bias that would affect our study results was very unlikely. In addition, to guarantee data quality, all data inserted by abstractors were checked by the principal investigator. In case of uncertainty, the treating urologist involved in the clinical process was consulted. Furthermore, the number of missing values was within an acceptable range, since 8% of patients had one or more missing baseline values that were inserted using multiple imputation. Sensitivity analyses of the original dataset and imputed datasets showed no significant differences regarding outcomes and, therefore, did not alter our final conclusions . Second, the study has a risk of measurement bias, since not all patients underwent the same diagnostic testing during follow-up. Thus, metastasis could only be confirmed in patients undergoing, for example, pelvic lymph node dissection or imaging; therefore, the number of patients with metastasis may in fact be underestimated. Third, other factors that may influence AS outcomes, such as pursuing AS despite significant tumour progression and follow-up intensity, were beyond the scope of this study. CONCLUSIONS Weighing the advantages of potentially avoiding radical treatment against the risk of the tumour progressing to a stage with worse prognosis remains the most important dilemma when selecting (higher-risk) patients with PCa for AS. In the present cohort, half of all patients on AS did not meet the PRIAS criteria. These selected “off protocol” patients have earlier disease progression and a higher risk of unfavourable outcomes following deferred surgery. PSAD ≥0.2 ng/ml/ml was an important individual predictor of tumour progression and was associated with worse disease prognosis.