There is a school of thought, as championed by Simon Baron-Cohen, that thinks of autistic disorders as ‘an exaggerated version of maleness’. The theory goes that autists are systematic, unable to articulate feelings, not great at empathising. They tend to gravitate towards logical subjects like maths and science. They often don’t have close friends. The same things are broadly more true of men than of women. Female diagnoses of high-functioning autism are much rarer (4:1 for profound autism, maybe even 9:1 for Aspergers). Ergo, maybe autism = extreme maleness. Girls are less likely to be diagnosed because a particularly “male” girl is… well, a tomboy. Whereas an extra-“male” boy is more of an anomaly.
It’s a nice theory, but has not been conclusively shown to have a physiological or endocrinological basis. The empathising / systemising nature of humans in general has been shown to correlate well with testosterone levels in foetuses. Testosterone has been shown in vitro (i.e. in a petri dish) to have an inhibitory effect upon a transcription factor called RORA, and biopsies show that autists have lower levels of RORA in their brains than the general population. There could very well be a link, but it’s not conclusive. It also doesn’t tell us much about why there is a link.