Sorry, we made a mistake
Scientists have denied the link between the HIV-protecting mutation and the risk of death from all causes
Polina Loseva, N+1
The authors of the article on the link between the CCR5Δ32 mutation, which protects against the most common type of HIV, and an increased risk of death from all causes have asked the journal Nature Medicine to withdraw their work, the STAT portal reports. After the publication of the article, a discussion developed around the results of the work, several scientists tried to reproduce these calculations, but came to the conclusion that the mutation does not affect the risk of death – at least among residents of the UK. The authors of the original article believe that the reason for their mistake is that they chose an unsuccessful genetic marker for counting, and their opponents believe that the matter is in the low quality of the work itself.
In early June 2019, researchers from California Xinzhu Wei (Xinzhu Wei) and Rasmus Nielsen (Rasmus Nielsen) published in the journal Nature Medicine the results of their study, in which they analyzed the genomes of more than 400 thousand Britons from the UK Biobank database and compared them with mortality data. Scientists have found that people homozygous for the CCR5Δ32 mutation have an increased risk of death from natural causes.
This relationship is not surprising: homozygotes for recessive mutations often live less than heterozygotes or carriers of "healthy" alleles. However, special attention has been focused on this particular mutation in the last year: it makes a person resistant to infection with the most common type of HIV, and it was her Chinese Jiankui He who tried to introduce the world's first edited children into the genome. Thus, Nielsen and Wei actually stated that He could increase the risk of death of his genetically modified girls – adjusted for the fact that, presumably, the girls were Chinese, and scientists received their data in the population of residents of the UK.
Shortly after the publication of Nielsen 's article and A British epidemiologist from Bristol, Sean Harrison, made a detailed criticism of them. He pointed out many shortcomings in their work: for example, in their model, the authors did not take into account the kinship of people, and it could affect the prevalence of mutation within the sample. They also did not check the quality of the data set: there may be people with an odd number of chromosomes in the sample, and this may be a database error, not a genetic anomaly).
Harrison was also surprised that the authors of the article did not indicate in the main text the confidence interval for their main result – that is, the spread of values within which the risk of death for homozygotes by mutation is located. In his criticism, he also wonders why the reviewers did not ask to add a confidence interval to the main text (it is only in the appendix to the article), and suspects that the authors deliberately hid it, since the spread is large enough and practically cancels the significance of their calculations.
Finally, Harrison tried to repeat the work of Nielsen and Wei. Since the CCR5Δ32 mutation is a deletion, that is, 32 nucleotides are missing in the genome, it is impossible to directly determine it in the DNA sequence. To find out whether a person has it or not, scientists use single–nucleotide polymorphisms - differences in the genome in one nucleotide. There are several such polymorphisms that scientists believe are associated with deletion, and Harrison used one of them as a marker, based on which he reproduced the calculations – and found no connection with mortality.
Nielsen and Wei, in a conversation with STAT, said that they worked together with Harrison to find the reason for the discrepancy in their data. Initially, they used a different single-nucleotide polymorphism and decided that it was the case. They recalculated their data, taking other polymorphisms as a deletion marker, and also found nothing.
Harrison believes that the reason for the error is the negligence of the authors of the work. He also discovered an article that was published in January 2019 and which already shows that the same single-nucleotide polymorphism that Nielsen and Wei used is not associated with an increased risk of death. This means, in his opinion, that this snip does not reflect the presence of a deletion very well, or that the authors of the work were negligent in their calculations.
Sami Nielsen and Wei believe that the error is caused by inaccuracy of UK Biobank data. According to the results of sequencing, it is not always possible to unambiguously indicate the presence of homozygotes or heterozygotes, since it is necessary to determine the frequency of a certain sequence throughout the genome. Scientists believe that because of this inaccuracy, they encountered some genotypes less often than they should have, and therefore their data turned out to be distorted. They told STAT that they tried to check their calculations on other databases, and in at least one they also found an inaccuracy in determining homo- and heterozygosity according to this particular snip.
However, Nielsen and Wei apologized and asked the magazine to withdraw their article. "If a mutation affects mortality," they say, "then the effect is weaker than we wrote."
In December 2018, another controversy began around the study of the risk of death: Italian scientists estimated that it stops growing after 105 years, and Australian Sol Newman suggested explaining this result with statistical distortions.
Portal "Eternal youth" http://vechnayamolodost.ru