The authors regret that the wrong structure of abiraterone D4A metabolite was used for molecular docking. The docked molecule had double bond between 5 and 6 carbon atoms of steroid fragment instead of 4 and 5 carbon atoms. That led to incorrect docking poses in Fig. 1, Fig. 2 and Fig. 3. The corrected versions of the figures are shown below. The obtained binding parameters of CYP21A2 with D4A were also corrected: the distance between the nitrogen and iron ion in the obtained complex was found to be approximately 2.4 Å (instead of 2.7 Å) and the distance between the C3 keto-group of the ligand and the Arg234 residue was found to be 2.54 Å (instead of 2.52 Å). We have also found that amino acid residues forming hydrophobic interactions with D4A are Val101, Val198, Trp202, Val287, Thr296, Val360, Leu364 (instead of Val101, Val198, Trp202, Val287, Val360, Leu364). The corrected data are presented in Table 1. The corrections do not alter the main conclusions of our paper. The authors would like to apologize for any inconvenience caused.