Most spam is predicated on attacking those too technologically literate to vet the attack. A smart scammer could create an email that would go under the radar of more modern client users, who are likely to report the message as spam and reduce the scammer's reach. If the message only works in older clients used by softer targets, then their chances of success are increased.

If this is the intended method of the attack, it's quite clever. Imagine if you could still blanket spam every email you come across, but only target users with old, outdated clients, who are likely older, less technologically savvy, etc... It would be well worth the R&D time.

Such reasoning follows the theory that spam messages include many typos in order to weed out 'smart' users, who are not desirable targets anyway.