Did you mean
this blog post by Coyne, explaining how the sheer number of people involved in PACE mean that the reviewers will almost certainly have had social or institutional connections to members of the PACE team (and therefore not be truly independent)? Or
this one where he looks at Peter White's declared (and undeclared) conflicts of interest?
I don't think there was a second review - I have no recollection of ever reading about a review having taken place - although if you have evidence that there was one then I'm obviously mistaken.
The biggest complaint I've read about the peer review process is the way that the Lancet has tried to pass it off as being way more robust than it actually is. Richard Horton claimed the PACE paper underwent 'endless' rounds of peer review - yet the paper was actually fast-tracked, meaning that it was published 28 days (or less) after it was received.
This Lancet article from 1998 explains how, right from the introduction of the fast-track process, peer-reviewers were expected to respond within 48 hours - which doesn't exactly encourage detailed forensic analysis, and perhaps goes some way to explaining how the reviewers might have missed some of the issues with PACE (although not the most glaring ones).
By early 2015
the Lancet was proudly proclaiming that it had got the fast-track process for RCTs down to 20 days - just 10 days to conduct the peer-review and make a decision (!) and then another 10 days from acceptance to publication. It definitely appears that speed is being prioritised over thoroughness. How long did they spend on the peer-review process for PACE? A leisurely 28 days or an almost indecently hasty 20? And how many rounds of robust peer review can you realistically cram into that window?