Dr Sadiq Muhammad
- A Genetic Basis for Homosexuality?
- Neurological and Genetic Mechanisms of Homosexuality Examined
- Evidence from DNA
- Can Epigenetics Explain Homosexuality?
- Twin studies, which give us the best indicator of whether homosexuality is genetically determined or not, show that if your identical twin is homosexual, you have an 80% likelihood of being heterosexual. This demonstrates that homosexuality is not genetically determined.
- Rates of homosexuality between identical and non-identical twins show little difference (5-7%), which is in keeping with other biological traits that have little to no genetic component, such as Parkinson’s Disease.
- Attempts to demonstrate that homosexual men and women have more “feminine” and “masculine” brains, respectively, as a result of hormonal differences in-utero, have failed, amid contradictory results. This was due to the fact that the method used was confounded by ethnicity of participants.
- All attempts to demonstrate a neurological correlate of homosexuality have similarly failed, with significant differences in small sample sizes, evaporating when larger sample sizes are utilised.
- Regions of DNA, such as Hamer’s Xq28 region, which seem to be more prevalent among homosexual individuals have similarly been shown to be artefacts as a result of small sample sizes. The methods utilised for seeking a “gay gene” have been flawed insofar as the method used – genome wide association – is unsuited to the study of homosexuality.
- Epigenetic differences cannot play any significant role in determining homosexuality, as identical twins, who show large differences in rates of homosexuality, are born with indistinguishable epigenomes.
A Genetic Basis for Homosexuality?
Identical twins (known as “monozygotic”), unsurprisingly, have identical DNA. Not only that, but they also share similar uterine environments. As such, they provide a powerful method of judging the relative contribution of genetics for biological traits. Through the diligence of various countries, large databases of identical, adult twins now exist, who often happily participate in genetic studies.
Analysis of 4,901 identical twins from the Australian birth registry1 found that the calculated likelihood of experiencing homosexual attraction if your identical twin does, is 20% for men, rising to 24% for women. In other words, if your identical twin experiences feelings of homosexual attraction, there is around an 80% probability that you will not. Thus, as regards homosexual attraction, the evidence shows that if an individual has homosexual feelings, the likelihood is that his or her identical twin will not have such similar feelings.
While the above study looked at homosexual feelings, another large twin study using population data from 3,826 monozygotic or identical (2320) and dizygotic or non-identical (1506) pairs of adult twins, aged 20-47 years old in Sweden, between 2005-2006, looked at homosexual partners. The results were similarly startling2. This study demonstrated that the probability of having had same sex relationships if one’s identical twin has are 18% for men and 22% for women. Again we see around an 80% chance of having only had heterosexual partners if one’s identical twin has had homosexual partners.
It should be also noted that this study was useful insofar as it provided data for us to compare the rates of homosexuality between non-identical and identical twins. This is important, since identical twins have identical DNA and non-identical twins have different but similar DNA, as siblings do. Thus, by comparing the rates of homosexuality between the two types of twins, we can see to what extent genes contribute to this trait. The results are clear: non-identical and identical twins have virtually the same rates of homosexuality: if a man’s identical twin has had homosexual partners, the likelihood of him having a lifetime homosexual partner is 18%. This drops to merely 11% in non-identical twins. For women the difference is smaller: 22% between identical twins and 17% in non-identical twins. Thus, the difference is between 5-7% in men and women together.
To give some perspective, in conditions that have strong genetic components, we see a much larger difference between identical and non-identical twin concordance rates: in Type 1 (insulin-dependent) diabetes, for example, non-identical twins have 57% lower concordance than identical twins3. Non-identical twins have a 47% lower concordance for autism spectrum disorders than identical twins4 and a 58% lower rate for Crohn’s disease5. These are examples of traits that do have a strong genetic component to them. As compared to such figures, the 5-7% concordance differences for homosexuality are negligible. Indeed, such a figure fits better with traits such as Parkinson’s disease, which, according to the largest population based heritability study shows a 7.9% concordance difference between non-identical and identical twins6, from which the researchers concluded that genetic factors do not play a major role in causing typical PD (Parkinson’s disease). If that is the case for Parkinson’s disease, why not homosexuality, when they show precisely the same difference in concordance rates?
Comparison of concordance rates or genome wide association studies (the pitfalls of which in respect to homosexuality are discussed shortly) are the two main methods by which genetic contribution to homosexuality are generally determined. Despite this, the Swedish study neglected such calculations and did not present these results up front. Instead, through “model-fitting” estimates, they obtained inflated figures, claiming that male and female homosexuality had 39% and 36% genetic components, instead of 5-7%. Media outlets subsequently reported these inflated figures, thus conveying the message to the public that homosexuality has a genetic component approaching 50%. However, what were not advertised as much were the confidence intervals (CI) for these inflated values. The confidence interval represents the range within which we can say with confidence that the true value lays. The CI was 0-59% for men and 0-49% for women. This means that it was possible that the true value might have been as low as zero or as high as 59% and 49% for men and women, respectively. This represents a large degree of uncertainty. Such wide confidence intervals are usually related to a lack of statistical power, by virtue of failing to look at a large enough sample size or are due to complex modelling (as in this study). Given that both of these were issues with this study, concordance rates should have been presented instead. Twin concordance rates, as per the study of myopia in children, which is also a complex phenomenon involving genetic and environmental factors, reliably “provide an estimate en masse of the effect of common genetic factors, rare genetic factors, epigenetics, structural genetic factors and to some extent gene-environment interaction factors on trait variance” according to an editorial in the Annals of Eye Science.7 Indeed, as the authors note, in general twin concordance studies provide “an upper bound estimate of heritability,” meaning that twin studies tend to over-estimate the heritable, genetic contribution. Identical twins not only have the same genetic makeup but also share the same uterus, at the same time. Thus, we can say that these low concordance rates reflect the contribution of not only genetic factors to homosexuality, but also in-utero developmental factors as well.
With such large confidence intervals, the claim of a 36-39% genetic contribution becomes highly doubtful. The authors did acknowledge the wide confidence interval in their conclusions, however they did not put forward the twin concordance rates, and the confidence intervals made it neither into the abstract nor into media reports of their paper. One can see therefore that a variety of statistical tests were calculated, which all measured slightly different aspects of the study. Despite this, the figures that were presented on the abstract of the paper and those that made it to the media were those that represented homosexuality as being more genetically determined, despite those values having the weakest evidential basis. To forsake a well known, tried and tested method of calculating the genetic contribution through comparison of concordance rates of identical and non-identical twins, for one that is statistically weak, evinces evidence of methodological ignorance or bias.
These are among the most reliable twin studies to date, as they used large, nationally representative sample sizes and, moreover, did not suffer from the key failing of earlier twin studies, in which recruitment of homosexual twins was done in a self-selecting manner, so that an individual might consider the sexual orientation of his/her twin, prior to participation in the study1. Such studies led to grossly over-inflated concordance rates between identical twins and drew much criticism.
The conclusion of these two largest studies are simple: homosexual individuals are not “born that way”. The Australian twin study showed that if your identical twin experiences homosexual feelings, there is around an 80% chance that you will not. Similarly, the Swedish study showed that if your identical twin has had homosexual partners, there is around an 80% probability that you will only have heterosexual partners. Additionally, if homosexuality was determined genetically, then we should also see greater differences between identical and non-identical twin concordance rates than 5-7%.
Despite such evidence, pronouncements of various genetic and developmental mechanisms of homosexuality are frequent in the media. What are they, and do they stand up to scrutiny?
Neurological and Genetic Mechanisms of Homosexuality Examined
Over the past decades, many purported theories have been suggested and dismissed as the biological “cause” of homosexuality. “Brain difference” arguments for homosexuality rested on the idea that men and women who later became homosexual had more feminine or masculine brains, respectively, at birth, than their heterosexual counterparts, thus leading to homosexual development in adulthood. The reason offered for such purported brain differences is that homosexuals were exposed, as foetuses, to a different level of androgens (male hormones, like testosterone) as compared to their heterosexual counterparts, which caused male homosexual brains to develop as feminine, while lesbian brains developed as more masculine.
Every foetus is originally female. The male is formed through a surge of androgen hormones (like testosterone) produced from the male foetal adrenal glands. This androgen surge results in re-configuration of the developing female genital tract, to form the male genitalia. This also has effects on other organs too, like the brain. During this surge, certain ratios within the body are thought to be determined. Finger length ratios between the index and ring fingers in particular, are thought to relate to prenatal (before birth) hormonal exposure8. Greater androgen exposure is related to smaller finger-length ratios. This is the ratio in size between the index and the ring finger. The idea was that if homosexual men can be shown to have more feminine finger ratios (larger differences between the index and ring finger), maybe that indicates less exposure to androgens during development. Similarly, if lesbian women were exposed to more androgens during development, then maybe that influenced them towards lesbianism, by making them masculine. This new avenue of research was pursued to see whether finger-length ratios between heterosexual and homosexual individuals are different and whether they indicate a difference in androgen exposure during brain development. The results, as we shall soon see, ended up in a mass of contradictions.
One paper looking at 849 men reported a significant difference between gay men and heterosexual men in finger-length ratios9. The results of this paper indicated that gay men had more “feminine” finger-ratios than heterosexual men. The results of this paper were in accordance with the findings of Lippa but were contradicted by the work of Robinson and Manning10, as well as by Rahman and Wilson11, whose work showed the exact opposite – that homosexual men had more masculine finger-length ratios than heterosexual men.
Among women, again a contradictory pattern emerges. Looking at 1,235 women, Lippa9 found no significant difference in finger-length ratios between homosexual and heterosexual women. This conclusion was also reached by Van Anders and Hampson12, while others studies showed opposite results13,14.
Such discrepancies are thought to be explainable largely by differences in ethnicity as detailed by Manning in 200215, who showed that such finger-length ratios vary widely between different races due to genetic differences, unrelated to hormonal exposure. This confounding by ethnicity largely invalidated the findings of finger-length ratio differences, especially when we consider that those studies that looked only at individuals from the same ethnicity, found that homosexual men in fact had smaller finger-length ratios, indicating a more masculine development in utero, than heterosexual men10. This observation is likely to be the correct view, though not for the reason one would imagine as is elaborated in the accompanying article entitled “Sexual Orientation Determined in the Womb?” Nevertheless, it demonstrates that the idea that homosexual men and women have more feminine or male brains, respectively, is unfounded.
Further evidence against the brain development argument comes from comparison with such conditions as Congenital adrenal hyperplasia (CAH). This is a condition that results in the exposure of the female brain to levels of androgen hormones akin to those experienced by the male foetus. If the hormonal theory of sexual-orientation is true, we should expect to see marked differences in brain scans between non-CAH and CAH females. This however, is not the case, as shown by work from researchers in Sweden16, who demonstrated using PET scanning that CAH women and non-CAH women have indistinguishable patterns of brain activity and connection in those very areas which show the greatest differences between men and women. This team looked at the ‘limbic system’, an area of the brain co-ordinating emotion, behaviour, motivation and long-term memory. Their work showed that there was no difference in type or amount of activity in the limbic system between CAH women and non-CAH women, while there was a significant difference between women and men, in both type and degree of activity. If exposure to over 10x the amount of androgens prenatally does not alter gender-sensitive brain areas in CAH women, small differences in prenatal hormones, within the normal range of a developing foetus, would not alter brain structure and function.
Alongside looking for evidence of prenatal hormonal effects, researchers also started looking for differences in brain structures between homosexual and heterosexual individuals. In 1991, Simon LeVay argued that differences in the size of certain cell-clusters of the brain’s hypothalamus were responsible for homosexual behaviour. LeVay claimed that an area of the hypothalamus, known as INAH-3, was smaller in homosexual men17 than in heterosexual men. INAH-3 is smaller in women than men, and so LeVay claimed that homosexual brains are “more feminine”, using this as an indication that homosexuality is hard-wired.
Though his work attracted huge attention in the eyes of the public through an effective media campaign, among scientists, his work has been rightly critiqued, most notably by Anne-Fausto Sterling in her work “Myths of Gender”18. LeVay looked at 19 homosexual men who died of AIDS and compared their brains, post-mortem, with 13 men whose sexual orientation he did not know, thus making his conclusions invalid. It was also pointed out that AIDS is known to cause cerebral and cortical atrophy (shrinking of the brain size) making the smaller INAH-3 size, entirely explicable. When others sought to replicate LeVay’s experiment, their results did not match.19 LeVay himself later admitted that the results do not allow one to decide if the size of the INAH-3 in an individual is the cause or consequence of that individual’s sexual orientation17.
Further attempts at establishing a neurological basis for homosexuality have been attempted, however, all of their conclusions have similarly failed to be replicated20,21. In the case of the brain region called the “anterior commissure”, attempts to replicate the results found the exact opposite of the original studies22.
Evidence from DNA
With the development of genetic analysis techniques, scientists increasingly turned towards studying DNA to seek the genetic basis of homosexuality. In 1993, behavioural geneticist Dean Hamer claimed to have identified regions of human DNA that differed between homosexual and heterosexual men and which he claimed influenced the sexual orientation of men23.
Hamer began with the premise that if there were a genetic influence on male homosexuality, it should be inherited on the X-chromosome, since sons inherit an X-chromosome only from their mother, thus explaining his observation that homosexual men are more likely to have homosexual brothers. Hamer did not address the fact that homosexual men could be more likely to have homosexual brothers as a result of being raised in the same household and not due to genetic causes. Hamer’s work predated the mapping of the human genome. As such, he did not look at specific genes, but used “DNA Linkage Analysis” of the X-chromosome that brothers share, to see which regions of the chromosome were highly expressed in both individuals, taking this as an indication as to the high expression of genes in those regions. He then posited that such genes may be drivers of the higher incidence of male homosexuality. Hamer’s study found that a region of DNA on the X-chromosome known as Xq28 was marked out in 33 of 40 (83%) pairs of Caucasian, homosexual brothers. The public quickly took notice.
Hamer soon replicated his findings and achieved a positive, but weaker correlation24. This time however, the sample size was smaller: 32 pairs of homosexual brothers. In this study, 67% of homosexual brothers (as compared to 83% in his previous study) shared the Xq28 region; however, he compared this to 11 heterosexual pairs of brothers, of whom only 22% shared this DNA region. Hamer’s two studies seemed to point to a genetic predisposition to homosexuality.
It was soon realised however that Hamer’s results were overstated; a third study sought to replicate Hamer’s work under very similar conditions, only using Canadian homosexual males rather than Italian homosexual males.25 This team, led by George Rice, differed also insofar as it used a considerably larger sample size: 52 pairs of homosexual brothers. Rice’s results showed that only 55% of homosexual brothers shared the Xq28 region, indicating that 45% of homosexual brothers do not share this region. The result of 55% was a far cry from Hamer’s results of 83% or 67%. When Hamer’s data was combined with Rice’s, the results emerged that Xq28 was not exclusive to male homosexuals.26
The issue of Xq28 was largely put to rest in 2014; with linkage scan of the full genome for sexual orientation in 908 individuals from 384 families, all homosexual brothers. This study was useful for one main reason: the Xq28 region was found to not be significantly linked to homosexuality27, the likelihood ratio being below the accepted significance threshold of 3.0, and significantly below the recommended significance threshold of 3.3 for such studies28. Despite the low LOD score for Xq28, which reduced further when multipoint LOD scoring was used, below the relevant significant threshold (and despite manipulating the linkage model to account for the “fraternal birth order effect” – shown to be a feature of nurture, not nature) the authors still described it as a significant result “especially in the context of past studies”.
These results were much in keeping with a previous genome wide scan which also showed Hamer’s Xq28 association to evaporate with larger sample sizes29. The nullifying effect of the larger sample size was attributed to the fact that homosexuality is likely more complicated than just the Xq28 region, and that Hamer’s data was distorted by use of small and unrepresentative samples.
The 2014 paper, in addition to disproving Xq28, highlighted a new region of DNA on Chromosome 8 as significantly shared between homosexual brothers. This however, is suspect. This is mainly because the study did not make use of a control group of heterosexual brothers so as to see whether these gene regions were simply associated with others factors in common between brothers, homosexual or not. Indeed, the linkages could be related to any number of common factors, such as ethnicity.
One can see from the above history, that the process of seeking a genetic basis for homosexuality has been a very messy one. Weak associations have been found, analysed, disproven, before new associations are found as the merry-go-round continues. Part of the reason is the methods that are utilised. Apart from not using control groups, as in the 2014 study, other more intricate flaws are clear. Firstly, none of these studies are applicable to female homosexuality. Secondly, the genetic linkage method used is entirely unsuited to studying homosexuality. This is because linkage studies can give inaccurate results when analysing associations between genes and behaviours30. Linkage studies using likelihood ratios (as in the above studies), require:
…precise genetic models, including penetrance, disease gene frequency, and the clear classification of individuals as unaffected or affected. Similarly, the presence of phenocopies can drastically affect the lod (likelihood) score and the calculated location of the…gene.30
This means that the technique used in the above studies is unreliable in the context of analysing homosexuality, as homosexuality has no precise genetic model; the penetrance of the genes – referring to what degree a gene finds expression – that may or may not be involved in its development are totally unknown; and individuals are not easily classified as homosexual or heterosexual since, as detailed in “Can Sexual Orientation Change in Adulthood?”, a significant percentage of homosexual individuals change sexual orientation during their lifetimes.
Thus, it is not surprising in the least that seeking the genetic basis for homosexuality has not produced any convincing results. After over thirty years of genetic research, the evidence has shown that genes may influence the personality and predisposition of an individual, making them more susceptible to environmental influences in favour of homosexuality, however, there is no deterministic genetic factor that makes a person “born gay”. As Hamer himself stated, our genes do not make us do it31. He also made it clear that there will never be a test that will say for certain whether a child will be gay. We know that for certain31. Indeed, as we have already shown through identical twin studies – despite identical genes, similar in-utero environments, and shared upbringings, the identical twin of a homosexual is around 80% likely to be heterosexual.
Despite the mounting evidence that homosexuality is largely influenced by non-genetic components, a new and emerging field of genetic study has entered the fray: epigenetics.
Can Epigenetics Explain Homosexuality?
Epigenetics is an emerging field of genetic research, which has received widespread attention in recent years. Epigenetics can be understood as a field of research that analyses not the sequence of the genetic code, but how the code is expressed. It looks at how genes can be turned “on” or “off” by various cellular and environmental factors. The reason a skin cell is different to a brain cell or a kidney cell is because, despite all containing the same DNA, each one has different genes turned “on” or “off”.
For example, a process called “methylation” turns “off” many genes. This involves the adding of a chemical subunit – a “methyl group” – to the gene. The important thing to remember is that human behaviour itself can result in epigenetic changes. For example, one epigenetic analysis32 revealed that smoking increases one’s risk of cancer through modifying one’s gene expression. This is thought to occur by turning “on” pro-cancerous genes and turning “off” DNA repair genes.
Though epigenetic changes occur over the course of one’s life, it has been realised that it is possible to inherit epigenetic modifications from one’s parents.33 The majority of epigenetic markers on the DNA of the sperm or ovum are erased at fertilisation; however, many epigenetic changes escape this process, especially in the maternal DNA.34 In other words: you don’t just inherit genes from your parents; you can also inherit which genes are turned “on” or “off”. This is extraordinarily important, as it may mean that one’s lifestyle and life events may subsequently influence the genetic features of future generations35.
Tuck Ngun, a postdoc from the University of California, presented results at the American Society of Human Genetics 2015 conference, claiming that he and his team had found five epigenetic modifications to DNA that are correlated with homosexuality. The press, taking merely a few snippets of information, ran with headlines such as “Have They Found the Gay Gene?” in the London’s Metro, without even considering the results as a whole. In the conference however, the results were regarded with considerably less enthusiasm. Here’s why.
Ngun’s team took 37 pairs of identical male twins, one of whom was heterosexual and one of whom was homosexual, as well as 10 pairs of homosexually identical twins, and analysed their epigenetic markers. He created a computer algorithm to predict sexual orientation, based on 6000 of these epigenetic markers and found that it predicted sexuality correctly 67% of the time. Besides the extremely small sample sizes he used, a second problem, as pointed out to him by Ed Yong’s piece, No, Scientists Have Not Found the ‘Gay Gene’ of The Atlantic36, is that the algorithm was developed from the data that was used to test how accurate the algorithm was! Ngun admitted as much himself in a post on his personal blog37. He wrote:
All models (from the very first to the final one) were built using just the training data. Only after we had created the model did we test their performance on the test data (the algorithm didn’t ‘see’ these during model creation). If performance was unsatisfactory, we remade the model by selecting a different set of predictors/features/data based on information from the training set and then re-evaluating on the test set.
In other words, the study was rigged from the start to find positive epigenetic associations to homosexuality. Ed Yong of The Atlantic explained that:
If you use this strategy, chances are you will find a positive result through random chance alone. Chances are some combination of methylation marks out of the original 6,000 will be significantly linked to sexual orientation, whether they genuinely affect sexual orientation or not. This is a well-known statistical problem that can be at least partly countered by running what’s called a ‘correction for multiple testing’. The team didn’t do that.
Columbia University statistician Andrew Gelman commented that38: Once you go back like that, you’ve violated the “test set” principle…Sure, you only did one test on your data. But had the data been different, you would’ve done a different test (because your remaking of the model, as described in the above quote, would’ve been different). Ed Yong of The Atlantic, summed up the study as follows: So, ultimately, what we have is an underpowered fishing expedition that used inappropriate statistics and that snagged results which may be false positives. The failure of the algorithm as a reliable tool to analyse the relationship between epigenetics and homosexuality, was also explained by William M. Briggs, statistician of online fame. He explained how the method used was unable to provide any meaningful conclusions39. Sara Reardon, writing for Nature News40, highlighted a similar theme when she writes that, associations found in small studies are prone to evaporate when tested in larger groups.
Besides the small sample size and the faulty method, there was a third problem, which firmly demonstrates the error of Ngun’s study as a whole. Researchers in 2005 found that identical twins inherit not only identical genes from their parents, but also identical epigenomes too41. In other words, they inherit not only the same genes, but also the same “on” or “off” buttons with those genes. It is as twins get older and go through their own individual life events, that their epigenetic “on” or “off” buttons change. In the words of the researchers: We found that, although twins are epigenetically indistinguishable during the early years of life, older monozygous (identical) twins exhibited remarkable differences. They attributed increasing differences with age to differences in lifestyles. Thus, any epigenetic differences between identical twins that account for one being homosexual and the other being heterosexual, as in the case of 37 of 47 of Ngun’s samples, must have been un-inherited differences, accumulated after birth as a result of life-experiences, because twins are epigenetically indistinguishable during the early years of life.
This firmly demonstrates that homosexuality is not epigenetically driven, given the fact that around 85-90% of identical twins of homosexual individuals, are heterosexual, as detailed earlier.
We have reviewed evidence from twin studies, hormonal studies, neurological studies, genetic and epigenetic studies and found that there is no inherited or in utero developmental factor which determines a person will be homosexual. An individual may inherit certain features, which may influence their response to life experiences (covered in other articles), however there seems to be no predetermination of homosexuality, genetic or otherwise. Nobody is “born gay”.