Thank you for submitting your work. The authors have systematically reviewed the effects of exercise-based pulmonary rehabilitation on severe to very severe COPD and found that exercise-based pulmonary rehabilitation improved exercise capacity, quality of life, and dyspnea. This review extends the current knowledge of the effects of pulmonary rehabilitation in severe to very severe COPD. However, the following points should be addressed.

・Why were only studies recruited severe to very severe COPD included in this review? I think that many other RCTs recruited for mild to very severe COPD should be included, and the review team is better to contact the authors to request data on severe to very severe COPD. Because I think that including only studies recruited severe to very severe COPD may cause serious selection bias.

・12 months from the last search is a bit long, there might have been valuable studies published since then. Consider updating the literature search.

・The authors limited their search to title or abstract. For instance, “[Title/Abstract]” was used in PubMed. I am not sure that their search was comprehensive. Please explain why you limited the search to title of abstract. Moreover, why were MeSH terms not used in Cochrane Library? Please explain.

・Is a search strategy for Embase correct? It seems not to include any ward for COPD. If it is incorrect, the authors must redo the search.

・Was author contact performed to obtain missing data?

・Meta-analysis on VO2 peak should not be performed. There was only one RCT. Similarly, I think that inconsistency in VO2 peak can not be rated for the same reason.

・The authors rated the risk of bias of 6MWD as not serious. Is it correct? Because I think 2 (Marie-Kathirin Breyer 2010 and Wajdi Mkacher 2015.) out of 5 studies were at higher risk of bias.

・I think that conclusions should be softened because the quality of evidence is low to moderate.