Fundamental flaws revealed in popular genetics technique used in over 200,000 studies:
A 2022 Nature paper gives a strongly worded critique of the most commonly used population genetic analysis tool. The paper describes the statistical technique (called PCA – principle component analysis) as deeply flawed, producing contradictory, incorrect and even “absurd” results and conclusions regarding ethnicity, ancestral populations and genetic relationships. Despite this, potentially over 200,000 genetics studies, applying to human, animal and plant genetics have deployed the technique. This new study raises serious questions regarding its use to date, including generating data that has contributed to old beliefs about race and ethnicity, and used to make tales about who and where people come from. The authors conclude that “PCA produces patterns no more historical than Alice in Wonderland and bear no more similarity to geographical maps.” Moreover, they state that the method “can be tuned by the experimenter to yield presumed “correct” results, and “correct” results can be cherry-picked…”.
The authors describe the rise of “dataism” as an ideology coming out of the Big Data era, with proponents believing that “with sufficient data and computing power, the world’s mysteries would reveal themselves. Dataism enthusiasts rarely ask themselves if PCA results are correct but rather how to interpret the results correctly.”
For a summary of the findings, see: https://scitechdaily.com/affecting-up-to-216000-studies-popular-genetic-method-found-to-be-deeply-flawed/
Texas Law Review: ‘America’s Hidden National DNA Database’, 2022
A new article from Texas Law review goes into detail on the topic of the use of DNA databases, both commercial e.g. ancestry tests, and health related, by police for criminal investigations. The rising use of non-forensic DNA databases by police represents a new era of conflation over the use of genetic resources that, to date, have largely treated health- and research-based genetic data as distinct from the forensic data gathered by police.
As explained in the review, genetic screening of newborns is an essential part of public health, allowing for early detection and intervention for treatment of illnesses that impact people in early years of life. The first such screening was introduced in the 1970s in the US for the disorder phentylketonuria. Each state in the US however, has since developed their own screening programs, with some states in 2003 testing for 4 illnesses, while others were testing for 36. From 2005, national recommendations have sought to harmonise the screening programs, now recommending 35 core conditions mandated for testing, along with 26 further secondary conditions being recommended. State laws also vary regarding how long blood samples and related data can be retained for, with some having the right to keep it indefinitely, or have no laws regarding the matter. Test results from current screening programmes are not usually sufficient to identify an individual, but the results of future screening programmes may be (for example, if whole genomes are used). In addition, the stored blood spots themselves could be reanalysed to obtain sufficient genetic information to identify an individual. Thus, storing blood spots beyond the period necessary raises privacy concerns.
Commercial genetic databases are already being used by police for criminal investigations. This use is exemplified in the US by the famous case of the Golden State serial killer who was identified from matching his crime scene DNA to distant relatives who had submitted their DNA to commercial companies, whose databases were subsequently mined by police for the investigation. Unlike forensics databases where it has not, in general, been considered ethical to store ordinary citizens DNA unless linked to serious crime cases, ancestry and clinical data is a goldmine of millions of people’s DNA, providing a valuable resource in the eyes of police.
Clinical and research samples are also attractive to law enforcement due to the quality assurance standards that are not applied to commercial tests, as well as increased cost effectiveness with regard to the analysis required. Moreover, they offer wider scope for analysis in some regard, without relying on genealogical analysis of distant relations only. While some data is not gathered for identification purposes as is the case with forensics, any move to include testing for the same sequences as used by forensics is not beyond possibility according to the review. Moreover, while screening is now limited to specific disease-causing regions of the genome, countries including the US and UK, are now beginning to perform whole genome sequencing of newborns, thus increasing the scope of information available, including to information that would be relevant to law enforcement. The review warns that while whole genome sequencing may not be rolled out to everybody, sequencing of identifiable sequences may be. For example, research identification of biospecimens is already required for certain funded work from the National Institute of Health (US).
While the use of clinical or research data is not currently routinely used by law enforcement anywhere in the world, a recent case was reported in the US where a man was arrested based on police analysis of his child’s newborn screening data, providing a forewarning to what may be possible going forward as databases continue to expand (see Newborn screening program used by police).
Current regulations appear to be far behind the interest shown by law enforcement to access genetic data. The review summarises the current regulatory system state by state. One quarter of states are yet to regulate police access to newborn screening samples and related data, while the rest have varying policies with regard to either the newborn samples or related data, or both. Some states, such as Rhode Island, also make no mention of the retention of samples or related data, let alone regarding police access to it. Where regulations do exist, they range in their scope and substance. The review reports that nearly one third of states likely permit access to either the blood samples or related data by police. Where police are allowed access, the review claims that this has occurred inadvertently, exposing how regulations are not keeping abreast with the evolution of genetic technologies and their applications. Some states however, such as Iowa, have a clear prohibition of police access. Other states regulate either the samples or the related data to varying degrees. Some states even regulate both, but access by law enforcement differs for each.
As summarised in the review, regulatory gaps are clear with regard to newborn screening samples and related data. The review urges that law enforcement be barred from accessing such data in order to preserve trust in essential public health programs.
The charity Progress Educational Trust, hosted an in-depth webinar on the ethical issues and concerns surrounding the sequencing of all new-born babies at birth, and the retention of genomic data across a life time.
When They Warn of Rare Disorders, These Prenatal Tests Are Usually Wrong The New York Times, January 2022
An investigation by the New York Times has found that genetic tests offered to pregnant mothers for detecting rare genetic disorders in their unborn child, are wrong up to 93 % of the time.
In no more than a decade, these genetic tests have moved out of the laboratory to the market place where they now serve more than a third of pregnant women in the USA. But this investigation suggests that such tests may be more beneficial to the companies’ bottom line, than to the health of a new family.
Start-ups and major companies are all involved in selling tests for a range of rare disorders, despite them not being sufficient to rule out the presence of all the disease-causing mutations, nor sufficient to confirm a positive result for any given mutation. Any such test needs to be confirmed with more invasive, painful procedures that involve taking a part of the placenta.
The investigation spoke with several mothers who described the distress after receiving what in the end were ‘false-positive’ results, and having to face difficult decisions about the future of their pregnancy.
Companies describe results as ‘near certain’, “reliable” and “highly accurate,” offering “total confidence” and “peace of mind” for patients. However, these genetic tests are not regulated to ensure clinical validity and utility, and should not be used as a confirmed diagnosis and people using them should seek follow-up testing. However, in many cases this is not happening. Further, healthcare professionals, who are often not equipped to deal with the increasing numbers of new tests on the market, may take results as definitive, adding to the confusion and distress for parents. One example given by geneticists even included a woman who ended her pregnancy after receiving false positive results from one of these genetic tests. Confirmatory tests can only often be performed beyond the point in the pregnancy that an abortion can be sought in the case of a serious genetic disease, limiting the utility of such tests in aiding parental decisions.
A major concern is false positive results, with 93 % of positive results for Prader-Willi and Angelman syndrome being false. The most accurate tests only reached 80 % rates of false positive results for Cri-du-chat syndrome. In general, the accuracy of tests decreases as the rarity of the disease increases and the number of people being screened grows. As such, healthcare policies tend to prioritise tests that can be most accurate, and are actionable in terms of treating diseases, rather than a simple ‘more is better’ approach, that can actually have unintended effects such as those now uncovered by this investigation. Moreover, such genetic disorders can be caused by hundreds of different types of mutations, most of which are not screened for in these tests and thus cannot be ruled out with a negative result, raising serious questions about the validity of mass genetic screening, especially by commercial companies who do not provide the confirmatory and follow up testing, counselling and support that is needed to assist people in making the most suitable choices for their health and that of their baby.
The expansion of commercial companies offering more and more health genetic tests thus requires careful scrutiny to ensure that profit incentives are not coming before actionable improvement in health outcomes.
Research collection of DNA from minoritized communities ‘threatening’ the reputation of genetic research November 2021
A new op-ed in the science journal Nature exposes the exploitative nature of decades of genetic research on those who self-define as Roma people, the largest minority community in Europe.
Roma people have experienced discrimination for hundreds of years in Europe, including during the Nazi era. Discrimination remains, with people suffering from social exclusion, including that enacted by the state. For example, Slovakia was segregating children in schools as recently as 1990s, a practice that was recently defended by the government in 2015 as being based on “genetically determined disorders associated with ‘inbreeding’”. Now this analysis shows how DNA research not only reflects such discrimination, but can also contribute to driving discrimination and mistreatment of already marginalised communities. Such prejudices expose the dangerously close relationship between genetics and eugenics concepts.
The analysis, conducted by a group of scientists on over 450 research papers, assessed how DNA was obtained and researched. The research spanned papers published from 1921 to 2021, many of which were published in the last 30 years. DNA has been collected since the 1990s, as well as blood samples taken as far back as the 1970’s from which DNA can be taken.
Some if the issues discovered include a lack of consent for taking DNA (including from prisoners), which was then shared in public databases without the individual’s knowledge. Data from many studies has also been shared for secondary use, such as for medical research without knowledge of participants. Yet, the researchers were not been able to find an example of research that has been conducted in a truly cooperative way — such as involving members of the community or efforts to improve the community’s access to health services, including therapies that might already be available. Moreover, even if data is anonymised, isolated communities tend to be more vulnerable to de-anonymisation, which may further exacerbate over-policing of these communities. With Roma people subjected to increased police surveillance and DNA testing, genetics research is benefiting from the overrepresentation of Roma peoples’ DNA in police databases.
Research articles were also found to be commonly littered with derogatory and insulting terms, as well racist assumptions. Terms such as “inbred” were used, or defining the community in a manner that goes against how people themselves identify, for example defining the populations to exclude those of mixed ancestry or those the researchers deem to be non-Roma, despite many Roma people themselves not considering themselves to have separate ancestry to Europeans.
Such genetics studies threaten the human right to DNA privacy, increase potential for state surveillance as well as further biased and incorrect assumptions that can fuel discrimination. The researchers thus call for any research to be conducted by or with communities themselves as well as with scholars who understand the political and societal contexts facing these populations.
This publication is vital in highlighting the common exploitative nature of genetics research, where social and state discrimination is merely reflected in the work that is produced. With the sensitivity of personal genetics information, and the potential to fuel prejudice with concepts of scientific racism, the ethics of genetics research is an urgent issue that needs to be thoroughly addressed. This research is bringing light to this important subject.
20 years after the human genome was first sequenced, dangerous gene myths abound Philip Ball The Guardian, 9th June 2021
Sequencing the human genome was hailed by many as decoding humanity’s “instruction manual”, with genes supposedly carrying all the information needed to determine particular traits, and mediate health and disease. Instead, some 20 years later, genetics has since turned this deterministic view of genetics on its head.
As dissected by Phillip Ball in The Guardian article, this deterministic “instruction book” image is precisely the fallacy that genomics has overturned, as our understanding of genetics grows. Yet such false claims are yet to be addressed by genomic researchers.
The human genome project has consistently promoted and sustained a misleading view of genes.
Failure to address the falsehoods of genetic determinism is laying the ground for the latest era of consumer tests that claim to reveal where we come from and what makes us, us.
It is also facilitating the latest resurgence in racist and eugenics politics.
Most traits have a genetic component that is immensely complicated, with perhaps hundreds of genes involved, as well as other environmental, and simply random factors.
While there have been some successes in genetic medicine, it has not transformed the field; successes in gene therapy are rare, and personalised medicine tailored to individual genetics has not materialised.
Use of SNP chips to detect rare pathogenic variants: retrospective, population based diagnostic evaluation BMJ (2021) 372 (Open access)
A study from the University of Exeter College of Medicine and Health, UK has found that common genetic screening methods used in commercial DNA kits and research projects are “extremely poor” at correctly detecting rare genetic variants linked to diseases such as breast and ovarian cancer. For an individual person, they conclude that “such screening methods are more likely to be wrong than right”.
The researchers recommend that such screening methods, called ‘SNP chips’, are not relied upon by doctors, and instead, that such test results be validated with standard diagnostic methods. The use of ‘SNP chip’ data has implications for healthcare, as people are being increasingly screened for rare disease variants via consumer tests, for which they seek medical advice and medical interventions. Some research projects may also return results from SNP chips to research participants, or are considering doing so in future. If people are given incorrect results based on a ‘SNP chip’ alone, this could be very serious. For example, many women have surgery to remove their breasts and/or ovaries if they believe they have a mutation in the BRCA1 or BRCA2 gene that gives them a high risk of developing breast or ovarian cancer.
Direct-to-consumer tests usually do not sequence the whole genome. Instead, they typically use ‘SNP-chip’ genotyping methods, which check for the presence or absence of specific small genetic variants, e.g. single nucleotide polymorphisms (SNPs) or small deletions or insertions. These methods were originally developed to look at common genetic variation in human populations, but are now being increasingly developed to detect rare genetic variants in people, despite the difficulties in accurately screening for rare variants with such techniques. This is because with very rare variants where they may be only a single or handful of carriers, they become difficult to distinguish over the background experimental noise of reference data sets from the wider population.
In order to perform a systematic evaluation of the accuracy of SNP chip methods in detecting rare genetic variants, researchers compared more accurate sequencing data of almost 50,000 DNA samples from the UK Biobank project with ‘SNP chip’ data. They also analysed SNP chip data sets from 21 direct-to-consumer test participants.
Of the 21 data sets from consumer tests, 100 % of all known disease-causing or risk enhancing single nucleotide polymorphisms (SNPs) were incorrect when compared with the actual sequence data. 74 % of small known disease-causing deletions or insertions were also incorrect. Moreover, 20 of the 21 people investigated had at least one false positive result for a rare disease variant when compared with sequencing – meaning that these people were told that they had a problem that they do not have.
Of the 49,908 UK biobank participants data, similarly low levels of correct genotyping were observed. This part of the study looked at rare mutations in the BRCA1 and BRCA2 genes, that increase a person’s risk of breast and/or ovarian cancer. The researchers compared the results of sequencing with two different SNP chips. Across both SNP chips, 425 BRCA variants thought to increase cancer risk were detected in 889 UK Biobank participants. Of these, just 17 variants in 37 participants were present in the sequencing data, and the others were false positives – meaning they identified a problem that did not exist. Even the most common true positive (present in 10 participants) had conflicting and uncertain interpretations. A further 43 BRCA variants associated with increased cancer risk were present in the sequencing data of 70 participants but were not detected by either SNP chip, meaning that these people’s increased risk of cancer would not have been identified. The authors concluded that the performance of both chips for genotyping BRCA variants was very poor.
How to design a national genomic project—a systematic review of active projects Human Genomics (2021) 15:20 (open access)
This study provides a summary of various genomic research projects being undertaken in countries across the globe. They identify 41 national projects that are currently active. The major aims of these projects vary, but they also share some common goals including attempting to determine the genetic variation that exists in healthy individuals, searching for disease-susceptibility variants, increasing the required infrastructure for future research and translation to the clinic, as well as enabling the idea of personalised medicine.
The study raises some challenges of these projects, including in understanding genetic variation in the healthy population, due to the difficulties in defining who is healthy at any given time point, with implications determining which variants increase risk of disease.
Projects vary in scope, with Estonia for example, planning to sequence over 30 % of the population, while others are focusing on many fewer people. The design of projects was also observed to be influenced by whether or not projects are privately or publicly funded, resulting in different priorities with regard to what diseases are assessed, and the scope of assessment. Differences in data sharing are also apparent, with some making data available to commercial companies, while others do not. Most projects are however, committed to some form of data sharing with researchers and the public.
Opportunistic genomic screening. Recommendations of the European Society of Human Genetics. European Journal of Human Genetics (2021) 29:365–377 (open access)
The European Society of Human Genetics has published recommendations on the use of genomic screening to look for genetic variants that are unrelated to the original health problem being assessed in a patient. This additional form of screening, called ‘opportunistic genomic screening’, is being suggested for roll out as part of genomics services that are in the midst of expansion in various countries, including in the US, France and the UK. This means that screening for completely unrelated disorders to the health problem being assessed for, may form part of routine healthcare approaches.
The Society argues that a cautious approach should be applied to any form of opportunistic screening, suggesting that pilot programs could be used to compare any potential benefits to other, more cost effective alternatives such as cascade testing, where family members of someone carrying a disease-causing mutation are also offered screening.
Their cautious approach is based on a number of ethical and health complexities raised by opportunistic genomic screening. Moving from the original genetic test for a specified health problem, to screening programs as part of an ‘active search’ for other unrelated genetic variants comes with both risks and benefits, and thus careful analysis is needed to ensure it is a proportional intervention. While genetic variants may be found that could be mitigated by medical interventions, risks can also arise from insufficient evidence regarding the health impact of a genetic variant. The interpretation of variants in unaffected people is hindered by the lack of validation of such tests at the population level when a family history of disease is lacking. The society thus recommends any screening to prioritise only well known, and highly penetrant genetic risk factors. Further, additional psychological burdens would be associated with an unfavourable test result, which would also raise costs in the form of required genetic counselling, if such services were to become routine. Identifying genetic ‘risk’ factors also has the potential for turning everyone who gets opportunistic screening into a “patient-in-waiting”, affecting people’s ability to get a job or a form of insurance, especially in nations with privatised healthcare systems.
Proposals to “opt-out” of opportunistic screening, rather than “opting in”, would also challenge the human right to fully informed consent over the access and handling of personal genetic data. Such a procedure would go against the norm for patient screening programs, which usually require the full and explicit consent of those being offered the tests. People should have also the right to decline information, or restrict the types of testing being performed. It is possible that someone may want to know about certain genetic variants but not others. Consent issues are further exacerbated by the increasing merging of health and research programs that may violate respect for autonomy. For example, if someone was offered a combined research and healthcare screening package, they may feel coerced into offering access for research in order to gain healthcare information. The society recommends that informed consent should be central to any such screening programs. A wider debate is also recommended, to prevent patients being reduced to being objects of well-intentioned medical deliberations and interventions.
Finally, with regard to children, the society recommends a highly cautious approach that restricts screening to childhood diseases that are fully ‘actionable’, i.e. can be properly treated or prevented whilst still in childhood. Proposals to routinely perform pre-natal sequencing, or sequencing of new-borns, requires urgent debate with regard to such programs.
LawSeq. Mapping and Shaping the Law of Genomics: https://lawseq.umn.edu
Position Statement on Direct to Consumer Genomic Testing by the UK Royal College of General Practitioners (2019)
The Royal College of General Practitioners (RCGP) and the British Society for Genetic Medicine (BSGM) recommend that health professionals should exercise caution when asked to offer, or provide, clinical expertise about the results of Direct to Consumer (DTC) genomic or genetic testing.
The analytical validity, sensitivity and clinical utility of such testing may be much lower than is popularly perceived. For certain types of DTC results, there is a very high chance of false positive or false negative results. This means that patients should be offered the NHS care which would otherwise have been offered (e.g. family history and risk assessment, healthy lifestyle advice, or referral to specialist care)1 regardless of their DTC result.
Should UK primary care be an early adopter of genomic medicine? British Journal of Medical Practice (2019) 69 (684): 330-331:
A new editorial discusses the rise in genomic testing, in the context of the UK’s recent projects to expand its role in healthcare settings, though the issues raised are broadly applicable to other equivalent national genomics programs.
“Increasingly, genomic data are being used for the diagnosis and treatment of disease, and in due course could be used for the prevention of disease.1 Following multiple endorsements of genomic medicine in 2018, UK Secretary of State for Health and Social Care, Matt Hancock, announced in January 2019 the development of a direct-to-consumer service for whole-genome sequencing, with provision for ‘customers’ to donate their data for research purposes.2 We present a dissensus — arguments why UK primary care should, and why it should not, be an early adopter of this technology, in order to understand its ethical aspects.”