Polygenes and\u00a0peas<\/p>\n
To appreciate just how important height has been in genetics, it\u2019s worth taking a step back to the turn of the twentieth century, when studies of height were at the center of fierce debates about how inheritance\u00a0works.<\/p>\n
On one side, Hirschhorn says, there were scientists studying people with severe alterations in stature who found patterns running in families. They argued that height behaved like Gregor Mendel\u2019s garden peas<\/a>: just as the peas could be tall or dwarfed, height could be passed down in one of two forms. But when other researchers plotted graphs with the heights of many people, they found that human height didn\u2019t behave like the peas. Instead, it followed a bell curve, with few people being very tall or short and many people falling somewhere\u00a0in-between.<\/p>\n This image from a 1914 article<\/a> in the Journal of Heredity\u00a0by geneticist Albert Blakeslee shows students at the Connecticut Agricultural College grouped according to height. It illustrates how height tends to form a \u201cnormal curve,\u201d with most students grouped near the middle and a few on either tail end. Image: Internet Archive<\/p>\n Both camps used studies of height to argue they\u2019d figured out how human genetics works. \u201cAnd both were right,\u201d says Hirschhorn. \u201cOr both were wrong, depending how you look at\u00a0it.\u201d<\/p>\n In 1918, British statistician Ronald Fisher settled the debate<\/a>. He argued that multiple genes contribute to variation in height, each of which follow Mendelian rules of inheritance. \u201cEach one of these genes makes you a little taller, a little shorter, and they all sort of add together,\u201d says Hirschhorn. \u201cIt works like Mendel, but it actually ends up resulting in a normal curve in the real\u00a0world.\u201d<\/p>\n Using the example of height, Fisher had come up with the idea of a polygenic trait: a characteristic shaped by the cumulative effects of many changes in the genetic code that vary between individuals. It\u2019s a concept that applies to countless human characteristics, from skin color and weight to the likelihood of developing cancer or heart disease<\/a>.<\/p>\n \u201cHeight is the classic polygenic trait,\u201d Hirschhorn says. \u201cBecause it\u2019s so easily measured, we can get very large datasets. We can get further in understanding how polygenic traits work, and that presumably gives us lessons we can apply to other polygenic traits and\u00a0diseases.\u201d<\/p>\n A tall\u00a0order<\/p>\n Hirschhorn originally got interested in height for personal reasons. Not because he\u2019s unusually tall or short \u2014 \u201cI am of average height, although I think I\u2019ve shrunk,\u201d he quips \u2014 but because of his daily experience in the clinic. Short stature is one of the most common reasons parents bring children to see a pediatric endocrinologist. During his fellowship in the field in the 1990s, Hirschhorn sometimes saw children who were growing more slowly than their peers \u2014 but a quick glimpse at the parents would indicate the slow growth was likely due to harmless genetic\u00a0disposition.<\/p>\n \u201cI used to tell the parents, there\u2019s a lot of genes that you carry, and you\u2019ve passed some of the shorter ones to your child, but we don\u2019t know what those genes are,\u201d he recalls. After repeating \u201cwe don\u2019t know\u201d dozens of times, Hirschhorn realized it was something he might just be able to figure\u00a0out.<\/p>\n Although eighty-plus years had passed since Fisher\u2019s influential paper, scientists hadn\u2019t made much progress in pinpointing the common genetic variants that shape height. They knew that about 80 to 90 percent of height is shaped by genetics, with environmental factors playing a smaller role. And by studying family histories, they\u2019d identified hundreds of monogenic traits: single, rare genetic variants that can have large effects on height. But since most people\u2019s stature is instead shaped by small variations in thousands of genes, just tracking family history wasn\u2019t enough to identify the common variants\u00a0involved.<\/p>\n Thanks to advances in DNA sequencing in the early 2000s, though, a new tool was on the horizon: the genome-wide association study, or GWAS. This technique draws from scans of the genomes of a large number of people, comparing those with a given trait to those who don\u2019t have it and identifying genetic markers that occur more frequently in one group versus the other. Hirschhorn was an early adopter of the tool, applying it to\u00a0height.<\/p>\n A model\u00a0trait<\/p>\n Height is mostly determined by genetics, with environmental factors like nutrition playing a smaller role. About half of the genetic influence comes from polygenic changes in common DNA variants (shown in the blue circle). The rest comes from from polygenic rare variants (yellow circle) and single-gene mutations (red circle). Some genes are linked to only one type of variation, while others overlap across two or even all three categories. Figure: Bicknell LS et al., The Genetic Basis of Human Height, Nature Review Genetics, 26, 604-619 (2025), Springer Nature<\/p>\n In 2009, Hirschhorn organized the Genetic Investigation of Anthropometric Traits (GIANT) consortium, an international collaboration of scientists who pool data to conduct genome-wide association studies related to height. \u201cFirst we found one gene variant linked to height. Then we found ten. Then there were a couple hundred,\u201d says Hirschhorn, who still chairs the GIANT consortium today. The more genomic data he and collaborators could access, and the more they refined the tools used to analyze it, the more variants they uncovered. By 2022, they were able to pinpoint more than 12,000 variants<\/a> reliably associated with\u00a0height.<\/p>\n Along the way, they gleaned lessons and best practices that would go on to influence the study of other polygenic traits. While some researchers had predicted that increasing the sample size beyond a few hundred thousand individuals would yield diminishing returns, Hirschhorn and colleagues demonstrated that increasing sample size into the millions directly increased the power of GWAS to detect important variants. By the time their 2022 study was published, which included genomic data from more than five million people, they had mapped all of the regions of the genome that contained common variants influencing height. And they had reached saturation: a point at which increasing the sample size no longer significantly improved accuracy in predicting height. They had shown what is achievable for understanding polygenic traits when sample sizes are big\u00a0enough.<\/p>\n Because a disproportionate number of the available samples came from individuals of European ancestry, though, they found that their ability to predict height was much less reliable for individuals from other backgrounds. Their work has also highlighted the importance of continued investment in genome-wide association studies that include diverse, non-European populations<\/a>.<\/p>\n The biology of\u00a0growth<\/p>\n Another insight Hirschhorn and colleagues revealed is that the heritability of a particular polygenic trait can be concentrated within a specific region of the genomic code, like a kind of\u00a0hotspot.<\/p>\n As an endocrinologist focused on hormone-related conditions, Hirschhorn originally expected most of the variants influencing height to be those that control and regulate growth hormone. To his surprise, though, most of the height-related variants discovered are actually expressed in a different genomic region, one that controls the growth plate: a soft area of cartilage near the ends of children\u2019s bones where cells divide to form new\u00a0bone.<\/p>\n \u201cNature is basically telling us the biology of growth,\u201d Hirschhorn says. \u201cIt\u2019s saying, \u2018Hey, it\u2019s the growth plate \u2014 that\u2019s where most of the action\u00a0is.\u2019\u201d<\/p>\n<\/p>\n
![\"chart\"\/](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2025\/08\/41576_2025_834_Fig3_HTML.jpeg\")
<\/p>\n