I'm thinking the appearance of a polar group where once there was an ether, reduces lipid solubility, and thus brain penetration. Sort of the reverse of morphine to heroin, but I'm sure there are examples where that seems to help.
*cough* morphine being O-desmethyl-codeine *cough*
Sure demethylation reduces BBB penetration and probably even more so oral BA, but increased receptor affinity/efficacy more than makes up for that sometimes.
Oxymorphone has 2 hydroxyls AND a carbonyl, but while oral BA is atrocious, it is still lipophilic enough to penetrate your BBB and fuck those mu receptors even harder than hydromorphone due to its added 14-hydroxy clitoral stimulator 8(
I bring up the development as a drug, because if there wasn't some catch with nortramadol, why not market that one? It's got a vaguely pethidine shape to it--probably came out of massive screens through a very detailed pharmacophore, and wasn't accidental. There's a chance that the developers were angling for lower dependence risk, but the same phenyl methoxy remains in venlafaxine.
The "catch" was, presumably, simply the significantly shorter half-life for nor-tramadol. Synthing stuff is cheap. Getting stuff approved by the FDA and doing enough marketing to actually get doctors to prescribe it is expensive as hell. To use another RC example, Etizolam is a perfectly fine benzo (or technically "thieno") that is enjoying widespread popularity in Asia, similar to what alprazolam is in the West. So why not market it over here? Well, alprazolam was already firmly entrenched in the market, so there was little reason to do it.
And I can't imagine it's ever cost-effective to build up from scratch what's got to be easier to just divert. The ultra-high potency opioids might make sense, but even then you'll notice typically include more, or bigger, substituents, never a slight shift in positioning of some functional group. There's like 100,000 documented tropane analogs, the same for every known brain receptor, and yet RC vendors always seem to go with ones brought to legitimate market and technically available.
Diverting a controlled substance isn't necessarily "easier", though. Diverting tends to be pretty illegal, and while you can usually smooth things over with a bribe or two in China, you don't want to be one of the poor saps they make an example of when they're trying to demonstrate their anti-corruption stance.
Synthing a non-controlled substance in your medium-scale lab, though? Nobody in China cares. And when that non-controlled O-DSMT is also significantly more potent than the controlled version, that's basically a win-win... just not enough to be anywhere as attractive as U-47700 or the fentanyl analogues.
Also, which "tropane analogs" are you talking about? You mean dichloropane and troparil? The only (recreational) tropane analogue that was actually "brought to legitimate market" is cocaine, unless you are talking about small amounts of investigational tropane-based SNDRI's being tested on monkeys... but I doubt anyone could afford to use these recreationally if people had to divert them from Sigma-Aldrich or Caymanchem. Not sure tropanes are a good example though, since even the ones that made it to the RC market involved extremely hard-to-source precursors, and were thus way too expensive to be a major success.
And sure, they're usually going to go with compounds that have already been described in the literature to some extent, but that doesn't mean they necessarily have to be "diverted" from somewhere. Where would the Chinese have gotten all that mephedrone from? I'm not aware of this stuff actually being used clinically, but somehow people managed to flood Europe's drug markets with huge amounts of it at dirt-cheap prices.
Give the Chinese some credit... if they can pull off the synth for acetildenafil (the unregulated sildenafil analogue they put into all those "herbal"
male enhancement supplements), they can sure as hell do O-DSMT.