Study selection

References from the PubMed and EMBASE search strategies were combined and duplicates were manually removed from EndNote, with the preference of PubMed over EMBASE. All screening phases were performed by two independent reviewers (MP and VG) and reported according to the “Preferred Reporting Items for Systematic Reviews and Meta-Analysis” (PRISMA) guidelines (Higgins & Green, 2011). References were first screened based on title and were excluded based on the following criteria: (1) titles showed no relevance to regeneration of articular (hyaline) cartilage, (2) it was specifically stated in the title that the conducted experiment was an in vitro study only, (3) osteoarthritis animal models were used, (4) only ex vivo studies were performed, and (5) deceased animals were used. In case of doubt or disagreement, references were included for further screening. The second screening phase consisted of a title/abstract screening in Early Review Organizing Software (EROS, Institute of Clinical Effectiveness and Health Policy, Buenos Aires, Argentina; www.eros-systematic-review.org). References were included based on the following inclusion criteria: (1) primary study, (2) animal model, (3) bone marrow stimulation by microfracturing or creation of an osteochondral defect, and (4) biomaterial implantation. Articles were only excluded when it was specifically stated in the abstract that the study was performed without healthy animals or acellular biomaterials, or if biomaterials were not implanted in the knee or ankle joint. Articles were not excluded in case important information in the abstract was missing. These articles were assessed in the full-text screening phase. For the full-text screening, articles were included if they met all of the following inclusion criteria: (1) primary study, (2) animal model, (3) healthy animals, (4) articular cartilage regeneration, (5) knee or ankle joint, (6) bone marrow stimulation by microfracturing or creation of an osteochondral defect, and (7) implantation of an acellular biomaterial. In general, if results of the two reviewers were different, articles were discussed until consensus was reached. In case of double publication, one of the studies was removed. During the screening phase, no selection was made based on publication language. The risk of bias assessment and meta-analysis was applied to studies with a comparison between a non-treated empty defect control and biomaterial implantation, and with semi-quantitative histological scoring system results as outcome data.