Understanding why individuals with identical rare genetic mutations end up with wildly different health outcomes has been one of genomics' most vexing puzzles. A landmark meta-analysis now offers the clearest picture yet of how rare chromosomal changes, common genetic variants, sex, age, and medications interact to shape two of medicine's most-tracked traits — height and body mass index — with implications for precision medicine and disease risk prediction.
Drawing on data from 1,447,001 individuals across six biobanks and clinical cohorts, researchers examined recurrent copy number variants (CNVs) — segments of the genome that are deleted or duplicated — alongside polygenic scores, sex, age, and medication use. CNVs generally showed mirror-image, dose-dependent effects: deletions and duplications pushed height and BMI in opposite directions with roughly symmetrical magnitude. However, a notable subset of genomic loci displayed asymmetric dose-responses for adult height, suggesting that biological buffering mechanisms protect against the effect of one allele direction but not the other. Critically, polygenic background and medications combined with CNV effects in patterns largely consistent with simple additivity — the whole equals the sum of its parts — yet loci at chromosomal regions 16p11.2 and 22q11.2 broke this rule, showing context-dependent effects that shifted across developmental stages, physiological states, and sexes. At 22q11.2 specifically, opposing gene contributions within the same deleted or duplicated segment appear to partly cancel each other out, offering a mechanistic explanation for why dosage buffering occurs at certain loci.
This is among the most statistically powered genetic architecture studies ever conducted. The finding that additivity holds at the population aggregate level, while context-dependence is widespread at specific loci, has real consequences for genomic counseling and polygenic risk tools. Current polygenic scores largely assume additive models; this work suggests those models are a useful approximation but will misclassify risk for carriers of specific CNVs in particular physiological contexts. The preprint status means peer review is pending, but the scale of the cohort lends the core findings considerable credibility. This is a confirmatory advance for additivity as a framework, and a paradigm-nudging one for locus-specific exceptions.