Tuesday, August 19, 2008

This one gets right up my nose

Or why journalists should read the original research paper and not just quote a scientist looking to hype up their own research. MHC-correlated odour preferences in humans and the use of oral contraceptive was published last week in the journal Proceedings of the Royal Society B. Studies, in animals and in humans, have shown that genes in the major histocompatibility complex (MHC) influence individual odours and that females often prefer odour of MHC-dissimilar males. The study in question here sought to investigate whether oral contraceptive use changes these odour preferences. The study found that women's preferences changed after they went on the pill - such that they went from preferring the odours of men who were MHC dissimlar to those of men who were MHC similar. The researchers also found that single women preferred men who were MHC similar, while women in relationships preferred those who were dissimilar. They speculate as to the reasons for this, suggesting that women in relationships might seek to improve offspring quality by seeking out other pairings. The study doesn't look into this, it's just a suggestion. The study also doesn't account for why single women might prefer MHC similar men, which is fair enough, as it's beyond its remit. Nor does it account for the possibility that the kind of woman who goes on the pill may have some behavioural, genetic, other, or combination of all three characteristics which may account for the change in preference over time; again, fair enough, that's not what it set out to do. The study certainly doesn't, in its own right, look into issues of fertility, relationship breakdown, or the possible consequences of MHC similarity for the health of future offspring; again, all these things are beyond its remit.

However, in the introduction the authors do suggest that
Mate preference for MHC-dissimilar individuals can be adaptive as it would increase offspring MHC heterozygosity, with beneficial influences on offspring viability through increased resistance to infectious disease or avoidance of inbreeding effects

referencing two previous studies, one, Potts & Wakeland, 1993, the other, Milinski, 2006.

The latter study, reviewing the literature on the possibility that choosing an MHC dissimilar mate may be adaptive, reports that findings have been equivocal:
Many of the studies that demonstrated or implied MHC-dependent mate choice found that the choosy sex prefers partners with somewhat dissimilar MHC alleles (e.g., Bonneaud et al. 2006; Egid & Brown 1989; Eklund et al. 1991; Freeman-Gallant et al. 2003; Landry et al. 2001; Ober et al. 1997; Olsson et al. 2003; Potts et al. 1991, 1994; Richardson et al. 2005; Wedekind & Füri 1997; Wedekind et al. 1995; Yamazaki et al. 1976, 1978). MHC disassortative mating may function to increase the resistance of offspring to infectious diseases by increasing their MHC heterozygosity (heterozygote advantage hypothesis; e.g., Apanius et al. 1997, Potts & Wakeland 1990) and/or it may operate to prevent kin-matings (inbreeding avoidance hypothesis; Brown & Eklund 1994, Potts et al. 1994), because this can have fitness benefits (Arkush et al. 2002, Meagher et al. 2000). Some correlative field studies supported the heterozygote advantage hypothesis (e.g., Carrington et al. 1999, Thursz et al. 1997) while others did not (e.g., Hill et al. 1991, Paterson et al. 1998). Also experimental studies beginning with Doherty & Zinkernagel (1975) provided ambiguous results as a recent meta-analysis showed (Penn 2002).

and concludes:
A heterozygote advantage per se has not been found by a number of studies and it is not necessarily expected either. Furthermore, as recent models have shown, a heterozygote advantage on its own fails to explain the high degree of polymorphism of the MHC (De Boer et al. 2004) in contrast to predictions of earlier models. Mate choice just for dissimilar MHC alleles would not necessarily improve the resistance of offspring nor would it help to maintain MHC polymorphism in the population.

There's a reason why journal articles contain references - it's so the reader may, if (s)he so wishes, check out the veracity of statements contained in that article. So, this one line in the introduction of the study under discussion, the only one, in fact, in that study to suggest that pairings of MHC similar mates may result in less than optimal offspring, is not, in fact, supported by the literature.

Which brings me to this:
Going for genetically similar men, detected from body odour, may increase a woman's risk of difficulties trying to conceive, miscarriage and of long intervals between pregnancies.

(from The Independent, UK - Why women can't sniff out Mr. Right when they're on the pill)
It's that 'detected from body odour' that's the outrage. It suggests that this study found this to be so. It didn't. While the science may support the idea that genetic similarity in partners is not good for their offspring, the science behind MHC similarity/dissimilarity does not show this to be so; as demonstrated above, the literature is equivocal, and certainly does not run to finding an association between MHC similarity and such dramatic outcomes as fertility problems and miscarriage. This is outright nonsense. Dangerous nonsense.

Which brings me to the reason I'm so very angry about this article. Of course, had the journalist done some research he would have turned up what I'd turned up. Had he read just the original research paper he would have realised that extrapolating wildly was inappropriate; had he read just the last three lines of the paper (again! journalists! read the conclusions! they're so very short, and so terribly important) he would have found his entire article written for him, And in this case, the conclusion even helpfully exaggerates the findings somewhat, so he wouldn't even have had to hype it himself:
We do not know whether the change in preferences related to pill use is sufficiently strong to influence partner choice, but it could do so if odour plays a significant role in actual human mate choice. Some studies have suggested that women consider the olfactory domain to be an important factor in their assessment of potential partners (e.g. Havlicek et al. 2008). Although we were unable to replicate the effect, Wedekind et al.'s (1995) demonstration of an association between MHC dissimilarity and the reminiscence of current or previous partners suggests that the influence of MHC-odour cues may extend beyond the laboratory. If this is the case, our results indicate that use of the contraceptive pill could lead to choice of an otherwise less preferred partner.

But look what we have here, a quote from Dr. Craig Roberts, one of the researchers on the study:
The preferences of women who began using the contraceptive Pill shifted towards men with genetically similar odours. Not only could MHC similarity in couples lead to infertility problems but it could ultimately lead to the breakdown of relationships when women stop using the Pill, as odour perception plays a significant role in maintaining attraction

Had he included this bit of speculation in the research paper submitted to Proceedings of the Royal Society B he would have found it rapidly run through with the editor's pen. His findings do not support this conclusion; the scientific literature does not support this conclusion; there is nothing to support this conclusion but his desire to hype his own research.
There are hundreds of reasons why a woman might choose a mate; there are equally hundreds of reasons why this mate may or may not choose to be chosen; these choices carry with them hundreds of implications, and we're a long way from unpicking the resultant 100sx100sx100s (that's lots and lots then) of possibilities that go into making babies and what those babies are like. It's unlikely many women will go off the pill as a result of this study, but if even one does, and finds herself with an unwanted pregnancy because the condom broke, then Jeremy Laurence and Dr. Craig Roberts should both be very very ashamed.

The Times does a reasonable job of reporting on this story, by the way.


Jon said...

More worryingly, ERV reports that the research area and significance of these results is somewhat circumscribed by the lack of an essential organ and mechanism.

"So just like there are specialized cells in your eyes and nose using specialized GPCRs to detect light or smells, mice (and snakes, and some other animals) have a structure in their noses, the vomeronasal organ (VNO), to detect a different kind of 'smell'. But the VNO is not just an extension of the nose (think of it as if you had a second pair of eyes that only saw in the UV spectrum). The GPCRs in the VNO do not recognize the same chemicals/molecules as the main olfactory system, and its not connected to your brain the same way as your nose. 'Normal' smells go to the olfactory lobe to be sorted into 'WHOO! Thats popcorn!' or 'BLECH! Thats a pig feed-lot!'. Vomeronasal smells end up at the amygdala and hypothalamus.

So its not too surprising that mice and snakes use vomeronasal smells for mating and aggression and stuff the amygdala and hypothalamus control. So why cant humans use our 'second smell' for mating too?

Well, thats where we hit that bump in the road.

Humans dont have a vomeronasal organ."

Anonymous said...

Interesting stuff. Re: "why journalists should read the original research paper" - I think they like to conduct journalism-by-press-release. It's a lot easier than reading the actual paper and they probably know that not many readers will actually pick them up on it.

Enjoyed the ERV take on this too. The lack of a vomeronasal organ would seem to be a bit of a problem. Shame the authors of the MHC study didn't address it.