I dug up the
full text of another paper by the same author, and it appears to have the full data set. 100 mcg/ml was the lowest paeoniflorin concentration that had a statistically-significant effect. (Page 10 of the pdf.) It increased androstenedione, decreased testosterone, and did nothing to estradiol. Just looking at that, you'd think it was just a
17-beta-HSD inhibitor. Glycyrrhetinic acid was the only chemical that increased the estradiol/testosterone ratio, and it looks like that came entirely from a reduction in T. I can't read Japanese and the Google translation is a mess, so I don't know their reasoning behind claiming aromatase was stimulated.
There's also some data on giving the herbs to live rats. Peony is called Shakuyaku here. 90 mg/kg was required to lower T. Estradiol looks like it went up, but the error bars are huge so it wasn't a significant change. When they removed the rats' ovaries, peony didn't work any more, so it apparently doesn't work in the adrenals or fat tissue. Maybe it would work in the testicles?
I also dug up
human pharmacokinetics. 145 mg of paeoniflorin was extracted from 5 g of peony root. 2.9% seems a bit low, so perhaps the extraction process wasn't very thorough. That dose provided a Cmax of 397.5 ng/ml. (But only 103.6 ng/ml if combined with licorice.) So to get 100 mcg/ml you'd need to take 36,478 mg paeoniflorin. That's 521 mg/kg for a 70 kg human, which is sort-of close to the estimate from the mouse study.
So then I did a search on Shakuyaku and found that
2.5-5 g peony + 2.5-5 g licorice does have an effect on T in human women. (I'm assuming the peony and licorice were combined in equal amounts like I've read elsewhere.) It dropped from 50-160 ng/dl to less than 50 ng/dl. That would be 72-145 mg paeoniflorin, which is very reasonable. Maybe it gets super-concentrated in the ovaries so you don't need 100 mcg/ml in the blood?