I’m afraid you got taken in on this one. The role of AOBEC in retrovirus mutation has been well known for at least a decade, and the observation of much more frequent hypermutation in viral DNA than in plasma virus RNA is commonplace. For example, we (Mary Kearney and colleagues at the NCI) have seen this effect in almost every patient we have looked at, and we would never have thought it worth remarking on, except in passing, much less a PLoS Biol paper. Indeed, I predicted that defective mutant proviruses should accumulate in cell DNA in my essay in Science 20 years ago. The fact is that it is impossible to calculate mutation rates from this kind of data, except to say that they are greater than 0. Contrary to what the authors of the paper seem to believe, the differences between RNA and DNA rates, and almost certainly between rates in (e.g.) rapid vs slow progressors are entirely due to selection—not on virus replication but cell survival. differences in the underlying rates are probably insignificant by comparison. In vivo, the large majority of cells infected with non-defective provirus produce virus and die within a day or two. By contrast, cells that get highly defective proviruses (especially due to APOBEC3G, which changes just about every Trp codon to stop) can survive for much longer times. The result is that cells with defective proviruses have much longer lifetimes (indeed, that can even become proliferating memory cells) and, over thousands of replication cycles, can become a large fraction of the infected cell population, even if the underlying rate is very small. Given that hypermutations are the result of APOBEC molecules that escape the action of Vif and end up in virions to exert their effect in the next infected cell, they are not distributed randomly, but are clustered on a few proviruses Since intact gag genes are needed to make virions, it is no surprise that the frequency of hypermutation in plasma virus is much less than in proviral DNA, again reflecting selection, and having no bearing whatever on the underlying mutation rate.Similarly, differences between different individuals must likely reflect differences in viral dynamics, not APOBEC mutation.
I hope this is clear. I’ll be happy to discuss further, if you like.