Open Access

Additional Information: The original online version of this article was revised: Following the publication of this article, the authors noted that the information on lymph-node metastases in the ARAMIS trial was unfortunately not correct in the original article by Fizazi et al (N Engl J Med 2019;380:1235-1246; DOI: 10.1056/NEJMoa1815671), but has since been corrected (N Engl J Med 2022;387:860, DOI 10.1056/NEJMx220007). As this information is cited in Table 1 and the text of the article by Wenzel et al, we have made the following changes in order to be consistent with the corrected source. Original: Table 1: “The proportion of N1 (control vs. treatment)… 29% vs. 17%”.

Text pages 140–141: “Patients with regional lymph node metastases were only allowed in the SPARTAN and ARAMIS trials and their proportions differed (Table 1). In the SPARTAN trial 16 and 16% harbored regional lymph node metastases, in respective control and treatment arms vs. 29 and 17% respectively in the ARAMIS trial”. Corrected: Table 1: “The proportion of N1 (control vs. treatment) … 12% vs. 10%”.

Text pages 140–141: “Patients with regional lymph node metastases were only allowed in the SPARTAN and ARAMIS trials and their proportions differed (Table 1). In the SPARTAN trial 16 and 16% harbored regional lymph node metastases, in respective control and treatment arms vs. 12 and 10% respectively in the ARAMIS trial”.

Original: Text pages 145–146: “However, more pronounced differences were recorded in rates of regional lymph node metastases and the study findings need to be interpreted in their light. However, it should also be noted that the distribution of lymph node metastases was virtually the same in the treatment and control arms in the SPARTAN and PROSPER trials. Although the distribution of lymph node metastases was unequally divided in the ARAMIS trial (control: 29 vs. treatment: 17%), a higher prevalence of lymph node metastases in the control arm certainly did not prevent darolutamide from ranking first in overall efficacy analyses. In consequence, it is not clearly evident that lymph node metastases rate differences resulted in important heterogeneity that confounded the study findings”.

Corrected: “Differences in patient characteristics that exist between the three-phase III RCTs are also important to consider in the interpretation of the current, as well as all previous NMAs. Very similar distribution of PSA, PSA-DT, baseline Gleason scores and regional lymph node metastases most likely had marginal if any contribution to population heterogeneity, within the three- phase III RCTs. However, study designs differed with respect to PSA- DT definitions. Specifically, the PROSPER trial relied on PSA-DT of less than 6 months. Conversely, the SPARTAN and ARAMIS trials included patients with PSA-DT for up to six months. Such difference may be marginal. However, it requires mention. In addition, all studies relied on conventional imaging. Although the use of conventional imaging did not differ between studies, it is of importance to emphasize that patient inclusion in the category of high-risk nmCRPC was much higher than if PSMA PET/CT was systematically obtained. Moreover, the timing of AE capture and their definitions may have also demonstrated small, albeit potentially important differences that influenced AE rates of the three ARATs. However, it is unlikely that study design differences have induced important confounding variables that prevent valid direct or indirect comparisons between the three RCTs since the endpoint of interest corresponds to OS. In all three RCTs, the assessment of this endpoint is the same. In addition, differences with respect to patterns of PSA-progression-free survival and metastatic progression-free survival (Supplemental Fig. 2) exist between the three RCTs. All of the above potential differences, regardless of their marginal or more important magnitude, were not and could not be formally addressed or adjusted for within the NMA methodology”.

The original article has been corrected.