Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Reviews

Using human genetics to improve


safety assessment of therapeutics
Keren J. Carss   1, Aimee M. Deaton   2,9, Alberto Del Rio-Espinola3,13, Dorothée Diogo4,
Mark Fielden5,10, Diptee A. Kulkarni6, Jonathan Moggs3, Peter Newham   1,
Matthew R. Nelson7, Frank D. Sistare8,11, Lucas D. Ward   2,9 ✉ and Jing Yuan2,12
Abstract | Human genetics research has discovered thousands of proteins associated with
complex and rare diseases. Genome-wide association studies (GWAS) and studies of Mendelian
disease have resulted in an increased understanding of the role of gene function and regulation
in human conditions. Although the application of human genetics has been explored primarily
as a method to identify potential drug targets and support their relevance to disease in
humans, there is increasing interest in using genetic data to identify potential safety liabilities
of modulating a given target. Human genetic variants can be used as a model to anticipate the
effect of lifelong modulation of therapeutic targets and identify the potential risk for on-target
adverse events. This approach is particularly useful for non-clinical safety evaluation of novel
therapeutics that lack pharmacologically relevant animal models and can contribute to the
intrinsic safety profile of a drug target. This Review illustrates applications of human genetics
to safety studies during drug discovery and development, including assessing the potential
for on- and off-target associated adverse events, carcinogenicity risk assessment, and guiding
translational safety study designs and monitoring strategies. A summary of available human
genetic resources and recommended best practices is provided. The challenges and future
perspectives of translating human genetic information to identify risks for potential drug
effects in preclinical and clinical development are discussed.

Genome-wide association
Human genetic research has discovered thousands of rate of drugs in clinical development due to both lack of
studies proteins associated with complex and rare diseases. efficacy and safety concerns; a survey of 246 terminated
Studies of phenotypes Genome-wide association studies (GWAS)1 and studies of clinical programmes showed that 37% failed owing
performed by testing Mendelian disease2 have resulted in an increased under- to non-clinical or clinical safety issues, versus 18% due to
correlation of genotyped
standing of the role of gene function and regulation in lack of efficacy8.
markers with a phenotype in
a large population of unrelated human conditions, which has been leveraged to discover This challenge is particularly evident for the ther-
individuals. Results are usually new therapeutic targets and provide evidence for ther- apeutic modulation of novel targets that have limited
relatively common coding or apeutic hypotheses of targets already being pursued3,4. biological characterization, where there is a lack of
noncoding variants associated
Retrospective analyses of drugs in clinical development pharmacological activity in animal models owing to
with relatively weak effects
on the phenotype.
demonstrated that drugs with supporting human genetic poor cross-reactivity of drugs to the orthologous disease
data for their intended indication were twice as likely to target, or where animal models have poor translatability
Mendelian disease be approved as drugs that lacked such validation5,6. for human disease or safety. Even when a suitable animal
A typically rare disease caused In addition to identifying new targets for drug effi- model exists, some safety liabilities may emerge only in
by variants in a single gene.
cacy, the study of human genetics also has the potential the chronic animal toxicology studies (typically 6 months
Studies of family members
with these diseases usually to identify risks for drug toxicity and associated adverse in rodents and 9 months in non-rodents, or in 2-year
uncover relatively rare variants events (AEs). Currently, preclinical safety evaluation rodent carcinogenicity studies), or manifest only during
with high penetrance. of pharmaceuticals relies mainly on animal models. developmental periods of exposure, which are often not
Although the use of animals has made huge contribu- studied until later stages of clinical development. As a
tions to safety assessment, there remains uncertainty result, there is significant interest in the application of
✉e-mail: lward@alnylam.com regarding the extent to which preclinical findings can human genetics to improve our understanding of the
https://doi.org/10.1038/ be extrapolated to humans owing to species differences potential safety liabilities associated with modulating
s41573-022-00561-w in biology7. This contributes, in part, to the high failure the activity of intended and unintended drug targets3,9,10.

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 145

0123456789();:
Reviews

Phenome-wide association
One of the first ways that human genetics was applied inform the design of clinical trials, either by excluding
study to drug safety was through pharmacogenetics, the study patients at high risk of the AEs or by enabling specific
A study of a genetic variant of the genetic basis of variation in drug response (Box 1). AE monitoring during the trial. It should be noted that
performed by testing the Although pharmacogenetics is a crucial component in genetic information can be used at all stages, ranging
correlation of the variant with
many phenotypes, typically
improving care and advancing precision medicine, we from target discovery to lead discovery, non-clinical
all of the phenotypes available extend this Review to broader applications of human studies, clinical development, life cycle management and
in a biobank or electronic genetics, making inferences from entire populations post-marketing pharmacovigilance. Accordingly, the
medical system. instead of studying variation only between recipients of starting point for these investigations will vary: in some
a particular drug. cases it will be instigated by consideration of a novel
Genetic variants that increase or decrease protein target based on biology alone; it may be prompted by
expression or function may serve as a model for lifelong findings that arise from testing of a preclinical lead com-
activation or inhibition of a drug target. Such ‘experi- pound in in vitro systems or animals, or from the need to
ments of nature’ can inform the likelihood of safety lia- design such tests; or it may be triggered by AEs observed
bilities following chronic drug exposure, which has been during clinical trials or through post-marketing reports.
illustrated in a recent phenome-wide association study
(PheWAS) for selected drug targets11,12. In addition, a Target validation and on-target safety
retrospective large-scale evaluation recently showed Human genetics, including the study of Mendelian
a correlation between the organ systems affected by disease as well as common and complex traits, can be
genetic variation in drug targets and the organ systems used to predict potential drug-induced target-related
in which adverse effects were observed in clinical trials13. AEs. Examples of clinically approved or failed drugs
Collectively, these results demonstrate that human illustrate the potential application of human genetics
genetic data can be used to help predict the effects to predict the safety of drug targets. A recent study13
associated with therapeutic manipulation of drug targets. evaluated the link between diseases and phenotypes
In this Review, we provide an overview of how associated with genetic variants in drug target genes
human genetic data can provide valuable insights into and AEs reported for the corresponding drugs. Using
potential safety liabilities of modulating drug targets — evidence from both Mendelian genetics and GWAS, the
both intended targets (on-target effects) and unintended authors demonstrated that AEs are more likely to occur
targets (off-target effects), including carcinogenicity risk. in organ systems in which there is genetic evidence that
We provide an overview of available human genetic links the drug target to a phenotype in that organ sys-
resources and illustrate their utility for safety assess- tem. For example, endocrine-related side effects were
ment through several published case studies. Finally, we noted among 12% of drugs for which the target has no
discuss some challenges and limitations of using such association with endocrine-associated phenotypes, while
data, as well as suggested best practices for how human that rate increased to 28% for drugs for which the tar-
genetics data can be used to enhance safety assessment get has associations with endocrine-related phenotypes.
of novel drug targets and modalities. These enrichments were evident even after controlling
for genetic effects related to the intended therapeutic
Application of genetics to drug safety indication and related organs.
A wide variety of techniques can be used to obtain Mendelian disorders, in which rare, highly pene-
human genetics data that can then be applied to the trant loss-of-function (LOF) or gain-of-function (GOF)
drug development process (Fig. 1 and Table 1). Genetic variants in a target gene cause a disease, can point to
evidence can be applied to predict potential AEs during potential AEs of target inhibitors or activators, respec-
various stages of the drug discovery and development tively. Examples include drugs that target diacylglycerol
pipeline, from target selection to clinical development. O-acyltransferase 1 (DGAT1) and sepiapterin reduc-
Human genetics can potentially be leveraged to decide tase (SPR). DGAT1 inhibition has been proposed for
whether to pursue or abandon a programme, to invest in the treatment of type 2 diabetes and obesity14. Clinical
early mechanistic de-risking investigations, to improve trial results reported dose-related gastrointestinal
the design of preclinical safety assessment studies or to AEs, including diarrhoea, which dramatically lim-
ited the therapeutic index15. In 2012, after the results
of the first-in-human study was published, children
Author addresses with DGAT1 deficiency were reported in a family of
1
AstraZeneca, R&D, Cambridge, UK. Ashkenazi Jewish descent16. These children, who carried
2
Amgen, Cambridge, MA, USA. DGAT1 LOF mutations, presented 3 days after birth with
3
Novartis Institutes for BioMedical Research, Basel, Switzerland. severe intractable diarrhoeal disorders. These genetic
4
Takeda, Cambridge, MA, USA. findings provided strong support for the hypothesis that
5
Amgen, Thousand Oaks, MA, USA. the severe gastrointestinal AEs observed in clinical trials
6
GlaxoSmithKline, Collegeville, PA, USA. were target-related and highlight the value of investigat-
7
Deerfield Management Company, L.P., New York, NY, USA.
ing the phenotypes of individuals with such variants at
8
Merck & Co., West Point, PA, USA.
9
Present address: Alnylam Pharmaceuticals, Cambridge, MA, USA. an early stage of consideration of a new drug target.
10
Present address: Kate Therapeutics, San Diego, CA, USA. Similarly, inhibition of SPR, the final enzyme in the
11
Present address: 315 Meadowmont Ln, Chapel Hill, NC, USA. tetrahydrobiopterin (BH4) de novo synthesis pathway,
12
Present address: Pfizer, Cambridge, MA, USA. has been investigated for the treatment of neuropathic
13
Present address: GentiBio Inc., Cambridge, MA, USA. pain. Despite the genetic and non-genetic evidence

146 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

Box 1 | Genetics of patients: the state of pharmacogenetics


Pharmacogenetics, the study of human heritable inter-individual genetic (TPMT)224, and TPMT genetic testing is clinically implemented for dose
variation in drug responses, can add valuable information to the risk individualization225.
assessment of a drug and guide drug therapy decisions. Over the decades, HLA genetic predisposition has been widely studied for immune-
the number of gene variants that have been found to influence drug driven drug toxicities. These investigations are particularly valuable
responses has steadily increased. Some variants were found to make for idiosyncratic toxicities that cannot be predicted preclinically and
drugs toxic; others rendered certain drugs ineffective. These genetic can only be addressed during clinical development. An interesting
variations are most extensively characterized in genes involved in drug example is carbamazepine-induced Stevens–Johnson syndrome (SJS)
absorption, distribution, metabolism or excretion (ADME), such as genes and related toxic epidermal necrolysis (TEN), whereby an association
that encode cytochrome P450 and transporters. Pharmacogenetic with HLA-B*1502 in the Chinese Han population226 led to the FDA
variations can also occur in genes that underlie immunity such as the recommendation for genotyping in Asian populations227 and exclusion
human leukocyte antigen/major histocompatibility complex (HLA/MHC) of patients with HLA-B*1502 from treatment, preventing SJS–TEN cases
locus or in the drug target itself or its pathway. in Taiwanese patients228 (HLA-A*3101 is the relevant allele in European
Understanding the pharmacokinetics of a drug can help anticipate populations229).
adverse events (AEs) that could be caused by variations in drug- Pharmacogenetics has been reviewed extensively, from drug
metabolizing enzymes, and vice versa: discovering associations between discovery230–232, drug efficacy233, drug safety234–238 and regulatory
AEs and patient genetics may reveal unexpected aspects of ADME. perspectives239–241. Currently, variants in around 20 genes provide
Cytochrome P450 enzyme activity is known to be highly variable useful predictions of reactions to 80–100 drugs — about 7% of drugs
owing to genetic variabilities. For example, the antiplatelet agent approved by the FDA242. Although pharmacogenetic discoveries could
clopidogrel requires activation by CYP2C19 to become bioactive. be beneficial to many patients, up-front screening has not been widely
Patients with loss-of-function CYP2C19 alleles (c.681G>A; rs4244285 adopted, and genetic testing is often not used until after a patient
or CYP2C19*2/*2) present with poor metabolization and relatively low has had problems with a medication. This is mainly due to the lack of
CYP2C19 activity, thus higher on-treatment platelet aggregation supporting randomized controlled trials. Clinical use of pharmacogenetic
compared with normal carriers (CYP2C19*1/*1)223. In 2010, the FDA testing has been largely based on observational and retrospective trials
added a boxed warning to the clopidogrel drug label about higher rates conducted in targeted patient populations and for research purposes.
of cardiovascular events in patients undergoing percutaneous coronary Recently, Ubiquitous Pharmacogenomics (U-PGx) has initiated a large,
intervention who are poor metabolizers of CYP2C19 substrates (these randomized controlled trial to study a panel of 13 genes for clinical
cardiovascular events are due to lack of efficacy, so are not strictly AEs by outcomes of 8,000 patients in seven hospitals in seven European
some definitions). countries243. Meanwhile, initiatives on pharmacogenetics passports such
Genetic variants in other enzymes involved in metabolism pathways as guidelines developed by the NIH-funded Clinical Pharmacogenetics
have been implicated in drug AEs. Thiopurine-related haematotoxicity is Implementation Consortium (CPIC) and The Dutch Pharmacogenetics
clinically relevant and therapy-limiting in treatment of patients with Working Group (DPWG) will facilitate the implementation of pharma­
acute lymphoblastic leukaemia (ALL) or inflammatory bowel disease (IBD). cogenetic testing for any upcoming treatments244 or clinical trials using
Severe haematotoxicity in patients under thiopurine therapy has been pharmacogenetics-guided dose, such as for warfarin showing lower risk
associated with polymorphisms in thiopurine S-methyltransferase of major bleeding245.

identifying BH4 synthesis as a key modulator of periph- with the cytotoxic T lymphocyte-associated 4 (CTLA4)
eral pain17, LOF variants in SPR and other genes in the protein26. LRBA deficiency is characterized by immuno­
pathway have been linked to neurological disorders18,19, deficiency and variable autoimmune presentations,
raising potential safety concerns associated with SPR including severe enteritis26–29. Patients with LRBA defi-
inhibitors penetrating the blood–brain barrier. In 2017, ciency show severe decreases in CTLA4 levels, and treat-
Quartet discontinued its SPR inhibitor development ment with abatacept, a CTLA4–IgG1 immunoglobulin
efforts owing to the observation of concerning neuro- fusion drug, can substantially improve enteritis in these
logical effects in toxicology studies, attributed to central patients26,29. These observations demonstrate a func-
nervous system exposure20. tional link between CTLA4 and the gastrointestinal
Mendelian genetics can also point to important cel- symptoms observed in patients with LRBA deficiency,
lular mechanisms that may indirectly predict AEs. One and is consistent with severe, potentially life-threatening,
such example is IL-2 receptor-α (IL-2RA) deficiency, colitis reported in cancer patients treated with ipili-
which causes immunodeficiency with lymphoprolifer- mumab or tremelimumab, two anti-CTLA4 monoclonal
ation and autoimmunity and is characterized by defects antibodies30. Another example is deficiency of erythro-
in T cell regulation21–23. Imbalance between regulatory cyte protein 4.2 (EPB42), which causes spherocytosis31;
T cells and effector T cells in the pancreatic islet has been loss of EPB42 results in depleted CD47 in the erythro-
implicated in breakdown of self-tolerance and develop- cyte membrane, due to their interaction32. Inhibitors of
ment of type 1 diabetes (T1D)24. The immune cell profile CD47 for cancer immunotherapy exhibit anaemia and
of the patients with IL-2RA deficiency, together with its thrombocytopenia as adverse effects due to the role of
reported biological link to T1D, is consistent with CD47 on normal erythrocytes33.
the new-onset diabetes cases reported as a side effect In addition to analyses of rare Mendelian syndromes,
of the IL-2R antibody baziliximab25. analyses of variants segregating in populations provide
Mechanistic information can also come from valuable information for drug safety (Fig.  2). In the
studying the Mendelian genetics of proteins that are past decade, GWAS have identified tens of thousands
known to directly interact with a target of interest. of variants associated with complex traits and enabled
Lipopolysaccharide-responsive and beige-like anchor the investigation of causal gene–phenotype relation-
protein (LRBA) is known to regulate and colocalize ships through integration with functional annotations

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 147

0123456789();:
Reviews

Novel therapeutics
O OH
O

Anti-X drugs

Target protein X

a Indication Y Side effect Z

b c d
X–/–

X X
Gene X Gene X Phenotyping individuals Studying evidence for target Guiding secondary
Association studies for with rare biallelic loss of in tumour suppression pharmacology
on-target liabilities function in gene X

Fig. 1 | Overview of the applications of human genetics to drug safety. This scenario illustrates evaluation of the
risk of a novel therapeutic, for example, an inhibitor of protein X. a | Population-scale association techniques, such as
phenome-wide association studies (PheWAS), genome-wide association studies (GWAS) and Mendelian randomization
(MR), can be used to infer putative on-target liabilities; in this example, an inhibitor of protein X may treat indication Y
while potentially causing side effect Z. b | The existence and phenotyping of individuals with very biallelic loss-of-function
variants (in this example, in gene X) can be used to discover liabilities or de-risk a target. c | Evidence of a role for on-target
or off-target effects in cancer from high-throughput sequencing, transcriptomics and perturbation screens can be used
to investigate carcinogenicity risk. d | Information from human genetic associations or syndromes can be used to design
secondary pharmacology screens to contextualize and prioritize off-target liabilities.

and causal inference methods. Retrospective examples indications11. The study found several associations
illustrate the potential of GWAS to inform drug safety. that suggest possible on-target AEs. One example is a
A notable example is the association at the CTLA4 locus functional variant in PNPLA3 (encoding patatin-like
discussed above with several autoimmune diseases phospholipase domain-containing protein 3) that is
that are also reported as common AEs of ipilimumab associated with an increased risk of nonalcoholic fatty
or tremelimumab, including vitiligo 34 and alopecia liver disease (NAFLD, the intended indication) and is
areata35,36. GWAS have the potential to provide valua- also associated with a slightly decreased risk of acne, high
ble information for safety assessment, particularly in cholesterol, gout and gallstones. Hence, therapies for
the ideal scenario where the target gene is clearly estab- NAFLD that act through reducing or inhibiting PNPLA3
lished as the gene driving the GWAS association, and may result in increased risk of these AEs. This study
the functional effects of the causal variant, including also reported association of a partial IFIH1 (encodes
tissue or context-dependent gene expression changes or interferon-induced helicase c domain-containing pro-
GOF or LOF of the protein encoded by the target gene, tein 1) LOF variant with lower risk of several autoim-
are well understood. In particular, when a suitable mune conditions (the intended indications) and with
genetic variant mimicking the effect of the drug has higher risk of asthma. How these observations would
been identified, methods such as PheWAS or Mendelian translate into AEs for a drug that targets these proteins
randomization (MR) can be used to investigate pleio- depends on multiple factors, as we discuss below.
tropic effects and predict potential adverse outcomes, Another recent analysis demonstrated the benefit of
as described below11,34,37–44. integrating additional information to PheWAS, such as
In PheWAS, a variant or group of variants is tested gene expression data and expression quantitative trait loci
for association with a large number of phenotypic end (eQTLs), to predict AEs47. After identifying potential
points. Data are typically derived from electronic health on-target AEs through PheWAS, follow-up analyses
Expression quantitative records (EHRs) or comprehensive questionnaires in including colocalization tests on genetic risk variants48,49
trait loci disease-agnostic cohorts45,46, although, in principle, are required to validate the causal link between the vari-
Variants, typically common and PheWAS can also be performed using full summary ant tested and the associated phenotype (or phenotypes),
noncoding, whose genotype
correlates with expression
statistics from published GWAS. A recent PheWAS which ultimately would be confirmed in randomized
of a gene, usually nearby, interrogated functional variants linked through GWAS clinical trials with the drug product. Exome sequencing
in a given tissue or cell type. with 19 candidate drug targets for common disease data combined with detailed EHRs and quantitative trait

148 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

Endogamous populations
data from the UK Biobank50–52 is proving to be a par- development versus pharmacological inhibition in the
Populations in which ticularly rich resource in this space. Several large-scale adult, incorrect variant selection74, differences in patient
reproduction is restricted PheWAS have recently been published, and the results population between genetic studies and clinical trials,
to relatively small social made available for the community via resources such as concomitant medications (for example, statins) or lim-
groups or extended families,
resulting in a higher degree
the AstraZeneca PheWAS Portal53 and GeneBass. These ited duration of clinical follow-up. These results high-
of consanguinity between databases empower the scientific community, including light the complexity of MR analysis and interpretation.
partners and a higher those without the statistical and computational resources Recently, MR–PheWAS has been proposed to conduct
incidence of homozygosity to perform such studies themselves, to use PheWAS data hypothesis-free MR analyses, which could enable MR
of alleles.
for their own research, including safety assessment of to be more broadly used for hypothesis generation in
Gene constraint
candidate drug targets. safety assessment.
Reduced nucleotide diversity In MR, a genetic variant, or a combination of genetic Investigations of phenotypes associated with LOF
in the coding sequence of a variants, is used as a proxy for exposure to assess the variants in a target gene are increasingly being used as
gene resulting from negative causal effect of an exposure on an outcome (which can evidence for probable safety of pharmaceutical inhibi-
selection on deleterious
variants.
be related to efficacy or safety54–57). MR has been used to tion of that target if the association is statistically robust
investigate unintended effects of drugs, including statins, with no disease relevance. Recent examples where LOF
proprotein convertase subtilisin/kexin-type 9 (PCSK9) variant analyses contributed to gene safety assessment
inhibitors, ezetimibe, IL-1 inhibitors and bempedoic are hydroxyacid oxidase 1 (HAO1), leucine-rich repeat
acid39,41–44. kinase 2 (LRRK2), chemokine CC motif receptor 5
PCSK9 inhibition has been studied extensively (CCR5) and mitogen-activated protein kinase kinase
by MR, because PCSK9 GOF variants cause familial kinase 15 (MAP3K15).
hypercholesterolaemia58–60 and rare-to-common LOF A therapeutic agent that inhibits HAO1 is approved
variants are associated with decreased low-density for the treatment for primary hyperoxaluria type I75. A
lipoprotein (LDL) levels and decreased coronary heart study in the Genes and Health cohort identified a healthy
disease risk61–64. Predicted LOF variants in PCSK9 that adult with predicted complete HAO1 knockout due to
are present in more than 2% of people of African origin a rare homozygous predicted LOF variant, supporting
have no apparent deleterious health consequences61,65. the likely safety of this therapeutic approach76.
This was used as evidence to support the probable safety LRRK2 is a promising therapeutic target for the treat-
of pharmaceutical PCSK9 inhibition. MR studies of ment of Parkinson disease77–79, and heterozygous LOF
PCSK9 and other lipid-lowering drugs have variously variants are not associated with reduced life expectancy
reported associations with Alzheimer disease and type 2 or with any specific disease80, supporting the safety
diabetes42,66–69; however, inhibitors such as evolocumab, profile of LRRK2 inhibition to 50%.
alirocumab and inclisiran are effective and well toler- Similarly, CCR5-Δ32 variants, which prevent func-
ated, with no clinical evidence of these risks70–72. The tional expression of the CCR5 chemokine receptor,
discrepancy between MR analysis and pharmacolog- were identified in HIV-1 infection-resistant people or
ical inhibition was confirmed by the comparison of slow-progressing patients81,82, leading to the develop-
PCSK9 MR–PheWAS and pooled clinical trial results69. ment of CCR5 antagonists for HIV-1 infection83. The
This inconsistency was explained by observations in lack of overt phenotype for homozygous CCR5-Δ32
knockout mice, where downregulation of pancreatic has provided confidence in safety for development of
Pcsk9 expression is linked to β-cell cholesterol overload CCR5 antagonists and genome-edited autologous T cell
and apoptosis, causing impaired glucose metabolism, transplants84.
whereas downregulation of liver or circulating PCSK9 Finally, MAP3K15 inhibition has been proposed as a
protein is linked to plasma LDL-cholesterol (LDL-c) therapeutic opportunity for diabetes85. Using a PheWAS
reduction without altering glucose homeostasis 73. approach, researchers estimated a low safety risk for
Alternative explanations could be loss of PCSK9 in selective therapeutic inhibition of MAP3K15 in humans.
Such examples are likely to become more frequent,
facilitated by the increasing availability of data sets that
Table 1 | Applications of human genetics to drug safety combine genetic and rich phenotypic data86,87. Population
Application Techniques genetics phenomena, such as the increased frequency of
Target identification and GWAS and PheWAS in large populations
alleles that are rare elsewhere owing to founder effects
validation, and evaluation of (such as deCODE in Iceland and FinnGen88,89), and
Study of Mendelian disease
on-target safety increased likelihood of homozygosity of rare alleles in
Phenotyping of individuals with very rare biallelic endogamous populations (such as PROMIS90 and Genes
loss-of-function variants and Health86), will facilitate the identification of individ-
Evaluation of target-mediated Study of somatic mutations, copy number variants uals with heterozygous and homozygous LOF variants
carcinogenicity and gene expression in cancer tissues for safety assessment.
Secondary pharmacology and Genetics-guided prioritization of off-targets included Although the investigation of individual LOF or GOF
off-target toxicity in screening assays variants has demonstrated value to predict target-related
Genetics-informed mechanistic toxicity AEs, gene constraint has not proved conclusively to be
investigations a useful metric to inform safety risk. High constraint
Pharmacogenetics GWAS or candidate-gene studies among patients indicates selective pressure against variants, which is
treated with a particular drug suggestive of functional importance of that sequence.
GWAS, genome-wide association studies; PheWAS, phenome-wide association studies. Genes that are depleted of coding variation, particularly

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 149

0123456789();:
Reviews

a
Cases Controls

Statistical method adjusting


or harmonized for population
stratification and selected • OR =(a/c)/(b/d)
confounders (age, sex, etc.) OR > 1: minor allele increases risk
OR < 1: minor allele decreases risk
• P value

Minor allele count Major allele count


Cases a b
Controls c d

b GWAS c PheWAS
chr1:1234424_A/G Phenotype 1
chr1:9234424_G/C Phenotype 2
Phenotype
chr5:5678998_A/T Phenotype 3
(disease, quantitative Variant
measurement, chr12:17568998_G/A chr19:10463118_G/C Phenotype 4
medication use, etc.)
chr19:10463118_G/C Phenotype 5
chr22:86568998_T/G Phenotype n
Indication Y Side effect Z
16 OR < 1
Association with side effect Z

Association driven OR > 1


by variant X
10
affecting gene G Significance

variant X in gene G
Significance

Association with
8 cut-off
cut-off
6

0
1 2 3 4 5 6 7 8 9 10 12 14 16 18 2022X Phenotype
Chromosome number

d Population Mendelian randomization


Random allocation of alleles
Variant
affecting
Genotype Genotype target gene G
group A group B

Decreased Increased
protein P protein P
levels levels Exposure
(e.g. biomarker,
protein levels)
Higher Lower
rate of rate of Causal
outcome O outcome O inference Confounder

Pharmacological modulation of
protein P encoded by gene G Outcome
may cause side effect O (e.g. disease)

LOF variants, are consistently found to be enriched for ancestry of the population on which they were calcu-
certain classes of genes including those required lated. Nevertheless, their value for disease gene inter-
for essential cellular functions, haploinsufficient genes, pretation and discovery suggests that they may also
mouse knockout genes that result in early lethality and be informative for safety assessment of drug targets —
Haploinsufficient genes genes associated with dominant Mendelian diseases91–93. that more constrained, essential genes may be less safe
Genes in which heterozygous Current constraint metrics mainly capture systemic to inhibit. However, there is currently limited evidence to
loss of function causes a tolerance to variation at early developmental stages, support including gene constraint metrics as a substan-
phenotypic change; that is, the
remaining (haploid) functional
which may limit their utility to model the effect of later, tial part of any framework for using human genomics
copy is insufficient to maintain more tissue-specific inhibition of a gene. Additionally, data in target safety assessment, and the targets of several
the wild-type phenotype. their resolution is affected by the sample size and successful drugs are highly constrained (for example,

150 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

◀ Fig. 2 | Population genetics methods overview. Population genetics refers to panels are selected by industry on the basis of biolog-
rare-to-common variants found in the general population, which may be tested for ical links between these proteins and adverse drug
predisposition to a disease or a trait. a | Genetic association studies use statistical methods reactions99. Human genetics can help to predict pro-
to test whether the frequency of a variant differs between individuals with different teins liable to perturb vital organ systems and repre-
phenotypes. The phenotype may be binary, comparing cases against controls, or a sents a potentially valuable but underused resource for
measurable trait with a continuous variation; in the illustrated example, the frequency of
selecting off-target proteins against which to screen. An
the variant alleles is compared between cases and controls. Potential confounding factors
that can (partially) explain the (lack of) association, including but not limited to differences
examplar is the potassium voltage-gated channel sub-
in ancestry, age or sex, should be carefully considered. The association analysis provides family H member 2 (hERG) (encoded by KCNH2)101.
an effect size, which estimates the direction and magnitude of the association of the allele This is a known mechanism through which drugs
tested, and a P value indicating the significance of the association. The effect size can be can cause QT prolongation and increase risk of cardiac
expressed as a β-coefficient (for a quantitative trait) or as an odds ratio (OR) (for a binary arrhythmias102 or seizures and is predicted by genetics,
phenotype). A corrected significance cut-off is often warranted to account for the as KCNH2 LOF variants cause long QT syndrome103.
number of tests performed in the analysis and to differentiate true-positive from Further support comes from the observation that
false-positive associations. In the context of safety assessment, considerations patients with certain non-pathogenic variants in KCNH2
about false-negative associations, including statistical power, may also be important. are more sensitive to the arrhythmogenic effects of phar-
b | In a genome-wide association study (GWAS), millions of variants are tested for
macological hERG channel inhibition104,105. Although
association with a phenotype. c | In a phenome-wide association study (PheWAS), one
variant (or multiple aggregated rare variants) can be tested for association with a large
some other proteins commonly included on secondary
spectrum of phenotypes; in the illustrated example, the tested variant associates with pharmacology screens have human genetic support —
decreased risk of disease Y, but also associates with an increased risk of a possible side for example, genetic evidence implicates the sodium
effect phenotype Z. d | When a significant robust association between a variant and voltage-gated channel α subunit 5, Nav1.5 (encoded by
a phenotype is identified, Mendelian randomization methods can be applied to infer SCN5A) in various cardiac arrhythmias and gastrointes-
the causal mechanism. Mendelian randomization tests whether an association between tinal symptoms106 — genetics may be useful to identify
a variant and a phenotype, often a disease, can be directly attributed to an effect of additional proteins that could more broadly inform sec-
the variant on an intermediate, quantitative, phenotype. The intermediate phenotype ondary pharmacology screens100,107. Importantly, as with
can be a biomarker or protein measurement, for example. Mendelian randomization on-target safety, the strength of genetic evidence must be
mimics a randomized control trial and takes advantage of the non-modifiable nature
critically evaluated.
of germ­line genetic variants, preventing reverse causality (left panel). However, it relies
on an assumption that the variant mechanistically affects the outcome only through the
An additional consideration for off-target screening
same mechanism as the drug, and not through other pathways (red crosses, right panel). is deciding which phenotypes are of the greatest safety
concern. ICH guidelines (S7A) highlight the cardiovas-
cular, respiratory and central nervous systems as vital
3-hydroxy-3-methylglutaryl-coenzyme A reductase organ systems and consider them the most important
(HMGCR) and statins)47,94,95. ones to be assessed in safety pharmacology studies.
Lastly, selective targeting of variant alleles, such as Therefore, genes with high-impact variants that affect
pathogenic germline or somatic GOF variants, can these systems should be prioritized for screening. A
prevent liabilities linked to loss of wild-type allele, as recent systematic assessment of Mendelian disease genes
shown by the small interfering RNA (siRNA) TD101 identified several proteins that could be added to current
for pachyonychia congenita96, two antisense oligonucleo­ off-target screens based on genetic evidence implicating
tides for Huntington disease97 and the EGFR inhibitor them in nervous system and cardiovascular disorders107
osimertinib in lung cancer98. (Table 2). Genetics should be considered for use in com-
bination with the more traditional pharmacological
Off-target drug screening approaches to guide selection of secondary pharmacol-
A key consideration for drug safety is not only the biol- ogy screens, and increasing amounts of genetic data will
ogy of the intended target, but also interactions with likely highlight more proteins that should be considered
‘off-targets’, or secondary pharmacology. The process of for off-target screening.
evaluating secondary pharmacology can be less straight- Lastly, the secondary pharmacology experimental
forward, because unlike the primary target, the identity design should take into account the drug modality.
QT prolongation of the secondary targets and the dynamics of their inter- A typical screen for binding or enzymatic inhibition
A delay in ventricular actions with the drug are not easy to anticipate before by a small molecule is inappropriate for screening the
repolarization during the collecting experimental data. Genetics can be used to off-target effects expected from technologies such as
cardiac cycle, visible by understand the biological effects of these off-target inter- RNA interference (RNAi) or proteolysis targeting chime-
electrocardiography and
a common drug-induced
actions both prospectively, to guide counter-screening ras (PROTACs), and the universe of potential off-targets is
adverse event. during drug development, and retrospectively, to different for these technologies108.
help identify off-targets responsible for preclinical or In addition to broad assessments of drug specificity,
PROTACs clinical toxicities. a counter screen against a protein related to the drug
Heterobifunctional small
Small-molecule drug candidates are commonly target is usually used during the development of both
molecules composed of
two ligand binding moieties assessed for their ability to bind or modulate various small- and large-molecule therapeutics. Several his-
connected via a linker. off-target proteins using in vitro secondary pharmacol- torical examples are informative (Table 3). One notable
One moiety binds to the target ogy assays99,100. These secondary pharmacology panels example in which interactions with a protein related to
protein, while the second are not mandated by regulatory authorities but, ideally, the drug target caused preclinical toxicity is that of the
moiety engages with cell E3
ubiquitin ligase complexes to
they should consist of proteins in which inadvertent β-secretase 1 (BACE1) inhibitors. Ocular toxicity caused
induce proteolysis of the target modulation would cause serious safety issues. Many by these inhibitors mirrored symptoms seen in neuronal
protein. ‘anti-targets’ on secondary pharmacology screening ceroid lipofuscinosis — a retinal disease caused by LOF

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 151

0123456789();:
Reviews

Table 2 | Selected proteins that could be added to current off-target screens


Protein Gene Mutation-associated OMIM Category
phenotype identifier
Cathepsin F CTSF Ceroid lipofuscinosis 615362 Nervous system
Cathepsin C CTSC Papillon–Lefevre syndrome, 602365 Immune system
Haim–Munk syndrome
Glutamate metabotropic receptor 1 GRM1 Spinocerebellar ataxia 614831 Nervous system
Hyperpolarization-activated cyclic HCN4 Sick sinus syndrome, Brugada 163800, Cardiovascular system
nucleotide gated potassium channel 4 syndrome 613123
High-affinity choline transporter SLC5A7 Neuronopathy 158580 Nervous system
(solute carrier 5 member 7)
Thromboxane A synthase 1 TBXAS1 Ghosal haematodiaphyseal 231095 Cardiovascular system
syndrome
Transient receptor potential cation TRPM4 Progressive familial heart 604559 Cardiovascular system
channel subfamily M member 4 block
Selection is based on a systematic analysis of Mendelian disease genes107. OMIM, Online Mendelian Inheritance in Man.

variants in CTSD, which encodes cathepsin D, an aspar- sequence-based off-target effects were subsequently
tyl protease related to BACE1 (refs.109,110). Modulation designed out during lead optimization. If off-target
of cathepsin D by BACE1 inhibitors was subsequently sequences are bioinformatically predicted in silico or
confirmed experimentally111. We suggest that the genet- experimentally observed for clinical candidate thera-
ics of proteins related to the drug target be examined peutic nucleic acids, the human genetic variation and
closely and, if any are implicated in phenotypes of seri- phenotypic association of these off-targets should be
ous safety concern, selectivity of molecules over these further evaluated. In support of this approach, draft
proteins should be assessed using relevant assays. regulatory guidance on developing antisense oligo­
Besides its utility for designing prospective off-target nucleotide and siRNA therapeutics to treat hepatitis B123
screens, genetics can be used retrospectively to identify states that information on human genetic variation be
off-target interactions mediating drug-related AEs100. considered for each off-target match that is predicted
For example, patients taking the Janus kinase 2 (JAK2) in silico. Lastly, for genome editing modalities, genetic
inhibitor fedratinib exhibited neurological symptoms background can alter off-target risk124 and its impact
similar to those of patients with the genetic disor- should be evaluated according to the latest FDA draft
der Wernicke’s encephalopathy; a similar syndrome, guidelines125.
thiamine-responsive encephalopathy type 2, is caused by As novel modalities (beyond small molecules with
variants in the thiamine transporter 2 gene solute carrier protein targets) are increasingly employed in drug
family 19 member 3 (SLC19A3)112,113. This led to the dis- development, a genetics-driven approach to off-target
covery that fedratinib inadvertently inhibited thiamine screening allows companies to go beyond traditional
transporter 2 (ref.114). ‘druggable’ proteins, but this will require considera-
Another potential example is that of thalidomide, tion of secondary pharmacology that may be unique to
which has well-documented teratogenic effects. Recent the modality.
studies have linked thalidomide-induced developmen-
tal toxicity to degradation of the transcription factor Evaluating carcinogenicity risk
spalt-like transcription factor 4 (SALL4)115–117. LOF The carcinogenicity risk of novel therapeutics is espe-
variants in SALL4 cause limb shortening, congenital cially difficult to evaluate experimentally because tum-
heart disease and ear and eye abnormalities118,119 pheno­ origenesis can be a lengthy process and may be evident
copying birth defects caused by thalidomide120. When only after chronic exposure. Valuable insights can often
a drug-associated AE is thought to be mediated by be gained through assessment of potential cancer pheno­
off-target activities, proteins with genetic evidence of types associated with genetic variation (for example, rare
involvement in that phenotype could be identified and or common germline variation or somatic mutations) or
prioritized for experimental follow-up. genetic modification of drug targets in animal models.
Seed hybridization Human genetics resources can also be used to sup- However, molecular, cellular and pathophysio­logical
Interaction of a short sequence port evaluation of potential or observed off-target differences between animal models and humans con-
at the 5′ end of a micro RNA effects associated with drug modalities other than found the extrapolation of preclinical carcinogenic-
(miRNA) or a small interfer­
small molecules, such as therapeutic nucleic acids. ity findings to humans. Thus, there is an increasing
ing RNA (siRNA) with the
3′ untranslated region of a For example, RNAi-mediated seed hybridization-based interest in systematically evaluating the potential for
mRNA, resulting in targeted off-target effects were found to be a major driver of human genetic associations of drug target genes with
degradation. Although this is hepatotoxicity observed with N-acetylgalactosamine tumorigenicity phenotypes using GWAS and Mendelian
the natural behaviour of (GalNAc)-conjugated siRNAs in rodent toxicity inherited cancer syndrome data, and by lever­aging
endogenous miRNA, it can
be an unintentional off-target
screens121. Similarly, hybridization-driven off-target databases of recurrent somatic mutations and epige-
activity of exogenous siRNA effects of antisense oligonucleotides have been shown netic and expression profiles in human tumour tissues
therapeutics. to contribute to hepatotoxicity of these drugs122. These and cell lines126,127.

152 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

Cancer driver genes


For example, human genetic data have provided an For the genes not yet established as cancer driver genes
The genes in a tumour in which opportunity to assess the potential carcinogenesis risks or with no obvious link to cancer through Mendelian
somatic mutations have been associated with therapeutic modulation of transform- cancer syndromes and GWAS, the carcinogenicity
positively selected, facilitating ing growth factor-β (TGFβ) signalling. TGFβ cytokines risk could be assessed by correlating their functional
tumour growth.
likely play a dual role in tumorigenesis depending on disruption with cancer longitudinal outcome data or
the context, which may be tumour-suppressive or with cell growth or survival synthetic lethal screens, for
pro-tumorigenic depending on the phase of tumour example, using RNAi or gene editing approaches142–144.
growth128,129. Recent in vitro data in cancer cell lines For example, a somatic mutation in a putative cancer
shows that perturbation of TGFβ through inactiva- driver gene would be expected to be associated with
tion or inhibition of its downstream effectors includ- poor survival outcome or more aggressive disease and
ing TGFβ receptor type 2 (TGFBR2) could help to would respond to a drug targeting the driver change,
overcome a strongly inhibitory tumour microenviron- resulting in tumour shrinkage and improvement in
ment and improve antitumour T cell responses130,131. survival145. Similarly, synthetic lethality as demon-
However, several studies highlight an association strated by cell death or a dramatic decrease in cell via-
between mutations in TGFBR1 and TGFBR2 and car- bility can be expected between a driver somatic change
cinogenic outcomes. Rare, heterozygous LOF mutations (hotspot somatic mutation, copy number change or over­
in TGFBR1 are reported in Ferguson–Smith disease, an expression) in a potential oncogene and its knockout or
autosomal dominant disorder characterized by devel- knock-down in synthetic lethal screens. Human genetic
opment of multiple self-healing squamous epitheli- association studies for the activation of oncogenes or
oma, an invasive skin tumour132. In line with this, an inhibition of tumour suppressors are also anticipated to
increased incidence of keratoacanthoma (a low-grade be leveraged for the evaluation of oncogenic potential
epithelial skin tumour) and skin squamous cell car- associated with genome engineering-based therapeutics
cinoma (SCC) have been reported in recent clinical (for example, CRISPR–Cas9) in which unintended and
trials of oncology drugs that inhibit TGFβ signalling nonrandom off-target editing is observed146,147.
such as the multi-kinase inhibitor sorafenib 133–135. The COSMIC database represents a valuable resource
Sequencing of two of the sorafenib-induced neo- to identify cancer driver genes from a wealth of tumour
plasms identified two somatic missense mutations in sequencing data 148. However, the inherent genetic
TGFBR1 (ref.136). These results demonstrate the value instability of tumours can make it difficult to identify
of human genetics to inform potential carcinogenic causative as opposed to correlative mutations. Recent
risks when scrutinizing data provided by in  vitro advances in human genome sequencing technologies
cell models130,131. such as error-corrected next generation sequencing
Genomic features of cancer (for example, frequent and access to large population-based comprehensive
somatic mutations, copy number changes, gene fusions phenotype–genotype databases such as UK Biobank,
or expression changes in tumour versus normal) can 23andMe and deCODE genetics149 may provide addi-
provide a causal link between target genes and car- tional opportunities to identify associations between
cinogenicity. Some well-known examples include germline or somatic mutations and carcinogenic risk
somatic activating mutations in BRAF observed in in non-tumorigenic samples. Such studies could serve
up to 60% of melanomas134,137, internal tandem dupli- to identify novel associations between GOF or LOF
cation of FLT3 (FLT3-ITD; approximately 25% of all mutations (either in the germline or somatically) and
AML cases)138, BCR–ABL gene fusion in CML139 and tumorigenicity.
TP53 LOF mutations in numerous cancers140. In addi-
tion, AE outcome data from patients treated with drugs Complexities and challenges
that target these frequent somatic genetic changes The various examples discussed above highlight the
have provided evidence for additional mechanisms potential power of human genetics to inform drug
for carcinogenicity; for example, the paradoxical acti- safety. Despite these utilities, there are significant chal-
vation of CRAF gives rise to skin SCC and other skin lenges that limit translation of genetic insights into drug
malignancies in melanoma patients treated with BRAF safety risk assessment. First, making causal inferences
inhibitors141. from genetic data is challenging and requires a deep
appreciation of statistical caveats. Second, once a causal
association has been strongly implicated between a geno­
Table 3 | Selected off-target effects for which human genetics has been type and a phenotype, there are additional limitations in
informative retrospectively
translating a genetic association to drug development
Drug Primary Off target Adverse event Genetic disease insights. These limitations fall into three broad areas:
target associated with pharmacological translation, gene isoform expression
off target
patterns and disease context.
LY2811376, BACE1 CTSD Retinal degradation Neuronal ceroid
AMG-8718 lipofuscinosis Pharmacological translation
Fedratinib JAK2 SLC19A3 Encephalopathy Encephalopathy A key limitation of translating genetic information to
Thalidomide Many SALL4 Limb deformities Duane-radial ray safety assessment is the fact that genetic variation may
syndrome or may not accurately mimic the effects of a drug. The
BACE1, β-secretase 1; CTSD, cathepsin D; JAK2, Janus kinase 2; SALL4, spalt-like transcription type of genetic variation (GOF versus LOF, heterozygous
factor 4; SLC19A3, solute carrier family 19 member 3. versus homozygous, coding versus regulatory, germline

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 153

0123456789();:
Reviews

versus somatic) will govern the connection between the suggested to be due to cell proliferation being driven
gene and phenotype studied. Likewise, the modality, tis- by a scaffolding behaviour of TMEM16A (via EGFR
sue targeting and/or distribution and dosing of a drug complex formation) rather than its role as a calcium
(small molecule versus antibody versus therapeutic pro- channel161,162. A similar discrepancy between PROTACs
tein) will influence the extent and duration of on- and and other modalities has been shown for inhibition of
off-target modulations. SMARCA2/4 (ref.163).
Thus, there are several major limitations with respect Systematic drug target gene homology assessments
to interpretation of genetic associations with AEs. In the during early development that take into account human
first instance, pharmaceuticals might not result in com- genetic variation of a drug target within the intended
plete inhibition or antagonism of their target over the patient population and genetic variation in animal
entire course of treatment or lifetime, which potentially breeds or strains support drug design and optimal spe-
would be required to ‘phenocopy’, or mirror the effect of, cies selection for both pharmacology and toxicology
LOF genotypes. Therefore, it may be possible to safely assessments. When the mechanism of a drug is not mod-
drug a target where the target is partially modulated ulating the activity of a specific protein, but instead tar-
(for example, 50–75%) without mimicking the genetic geted killing of a cell type, silencing of a mRNA molecule
phenotype. For example, serum lipoprotein a (Lp(a)) or editing of genomic DNA itself, different principles are
level is associated positively with cardiovascular risk and likely to apply.
inversely with risk of type 2 diabetes150,151. Molar con- For example, in the case of a drug such as a bispecific
centration was identified as the attribute of Lp(a) that T cell engager (BiTE) that targets a certain type of cell
affects risk of cardiovascular diseases but T2D risk was for killing based on surface marker expression, genetic
associated only with low molar concentration of Lp(a) associations with that cell surface marker are likely not
(<3.5 nM). Thus, pharmacological reduction of serum as relevant to safety evaluation as the expression pat-
Lp(a) concentration in the individuals with the greatest tern of the marker across cell types, which will be much
concentration down to the population median of 14 nM more informative. The impact of human genetic varia-
is predicted to decrease coronary artery disease risk tion also needs to be carefully considered for the safety
without increasing T2D risk152. Accordingly, heterozy- assessment of cell and gene therapies, including genome
gous LOF variants may represent an appropriate model editing specificity (for example, variants creating novel
to better inform the drug risk–benefit ratio. Although or higher potency off-target loci). Genetic variation (for
genetic effects on gene dosage can be helpful for risk example, in DNA repair pathways) might influence viral
assessment of therapeutics, there is a lack of precision as vector integration site patterns, although this has yet to
well as dynamic variation that hampers interpretation. be reported.
Similarly, genetic effects may take months or
years to manifest (for example, RORγ candidia- Gene expression patterns
sis and mycobacteriosis 153 or MERTK and retinal Cross-tissue expression patterns are an important medi-
phenotypes154,155); translating this lifelong genotype– ator for translating genetic associations to drug safety.
phenotype association into drug modulation of a target For example, if the genetic variant affects an isoform that
gene into a therapeutic window represents a significant is not expressed in a given tissue, it cannot be expected to
challenge. be predictive of pharmacological effects on that tissue. In
It should also be recognized that proteins are fre- addition, the characteristics of the therapeutic agent
quently modular in nature, composed of multiple in question such as mechanism of action, dose, chem-
domains that enable broad functionality of the protein, ical properties and route of administration will affect
such as kinase domains, scaffold functions, co-activators, whether a genetic association is relevant to drug safety.
co-repressor binding sites, subcellular and extracellular For example, if a drug is targeted specifically to tissue
localization motifs, and post-translational modification in the liver and cannot cross the blood–brain barrier,
sites156,157. Genetic variants that result in complete loss a neurological genetic association that is known to be
of expression or early truncation, or splice site changes, mediated solely by behaviour of the target protein in the
mimic RNA downregulation or protein degradation, brain will not be relevant to evaluation of the drug.
whereas damaging missense variants may mimic func-
tional inhibition or antagonized protein while keep- Disease context
ing other functions intact (for example, scaffolding). Some disease states may be promoted as a consequence
Therefore, the phenotypes associated with these different of excessive and/or sustained expression of genetic net-
variants might be significantly different. works (for example, sustained cytokine expression in
For instance, high expression and amplification of the autoimmune diseases such as rheumatoid arthritis164).
TMEM16A gene (encodes anoctamin 1 chloride chan- In such a setting pharmacological regulation results in
nel) has been demonstrated in several cancers, reach- normalization, that is, a disease-relevant pathway may
ing 36% in oesophageal cancer158,159. Its blockade with be operating at multiple folds of normal levels, and
specific functional inhibitors of TMEM16A-mediated drug treatment inhibits this to the basal level, which
chloride currents did not alter cell proliferation in would not necessarily equate to the effects of the drug
tumour cell lines, whereas PROTACs that decreased or genetic variant in a normal non-pathogenic context.
TMEM16A protein levels showed dose-dependent inhi- One of the major advantages of Mendelian genetics
bition of cell proliferation160. This difference in efficacy in translational sciences is the established link between
between PROTACs and a functional inhibitor has been the variants and the gene function (with the caveat that

154 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

the level of confidence in this link relies on the quality and interpretation, to avoid pitfalls. Critically evaluat-
of the supporting literature). However, many Mendelian ing the validity of these data is essential; for example,
disorders are characterized by complex and sometimes candidate-gene association studies in the literature have
variable presentations with many clinical features, which a high false-positive rate174. An understanding of sta-
makes it challenging to translate into predictable AEs. tistical pitfalls is important, especially when interpret-
Despite these limitations, a recent retrospective study ing the results of GWAS or PheWAS. For example, the
demonstrates that, across drug targets with reported P value threshold for a significant result should consider
disease-causing variants, a decreased risk for AEs that the number of both independent variants and pheno-
manifest in a given organ system is observed if the organ types tested. Prior hypotheses about variants believed to
system is not affected in the Mendelian disorder13. functionally mimic a therapeutic intervention and the
Despite the increasing role of human genetics in phenotypes likely to be affected can be used to limit
target validation, choosing a target free of genetically the search space and increase power, although caution
predicted liabilities is still not a guarantee that AEs will is recommended here. Below is a workflow suggesting
not be observed clinically, owing to the aforementioned how genetic data could be considered in a typical early
caveats. For this reason, human genetics should be target evaluation.
thought of as a tool for hazard identification, but can
never be expected to provide a complete picture of the Determine the breadth of the search
potential risks of a novel therapeutic agent. If specific variants functionally linked to the gene are
associated with a phenotype relevant to the therapeutic
Diversity and ethical considerations indication, this strengthens the plausibility of the rele-
Most human genetic research has been performed in vance of other phenotype associations to drug safety. If
populations of European ancestry, which raises con- there is no association with the therapeutic indication,
cerns for the use of genetics for drug development165–167. the reasons for a lack of association should be critically
This bias in the data makes GWAS most powered to considered with respect to the therapeutic hypothesis.
enumerate risk alleles in populations of European Are there no well-powered genetic data for the indi-
ancestry, resulting in comparatively poorer AE risk cation of interest? Does the available variation affect
prediction in other populations. As a matter of equity the gene only in tissue contexts that are not relevant
but also to increase power for target validation to ben- to the indication?
efit drug development in general, it is imperative for
genetic research to prioritize expanding data collection Investigate germline genetic evidence for other
in populations of non-European ancestry168. phenotypes
In the USA, diversifying participation in genetic studies Mendelian genetics. Search the Mendelian gene litera-
and expanding the health benefits of genetic informa- ture, for example, Online Mendelian Inheritance in Man
tion has been a stated goal of the All Of Us biobanking (OMIM)175. The highest-confidence gene–phenotype
project led by the NIH169. Expanding study popula- relationships are listed in OMIM as relationships where
tions will also involve performing research in low- and “the molecular basis of the disorder is known.” Critically
middle-income countries, which prompts ethical con- evaluate the validity of these relationships using mod-
cerns around protecting the interests of research parti­ ern best practices176 and resources such as ClinGen177
cipants and local scientists and benefit sharing170,171. and ClinVar178. Be especially critical of ‘candidate-gene’
In Africa, both the publicly funded initiative H3Africa172 or ‘candidate-SNP’ publications. Advantages of using
and the private venture 54Gene173 have centred local Mendelian diseases are that the causal gene is usually
researchers, institutions and economic development in clear and effects are usually strong; disadvantages are
their genetic studies. that sample sizes of affected individuals are usually low.
Some integrative databases such as DisGeNet (https://
Resources and suggested best practices www.disgenet.org/dbinfo) combine Mendelian and
A wide range of publicly available databases collecting complex associations. Scrutinize the Mendelian litera-
human genetics data can be used in evaluation of the ture for the phenotypic spectrum involved in the syn-
safety of intended targets or off-targets. In this section, drome. If phenotypes are secondary to developmental
we aim to provide a list of these resources, highlight alterations, they may be less informative to drug varia-
common pitfalls to be avoided in the interpretation of tion. Estimate the power to detect or exclude additional
the data, and suggest a framework of best practices to effects by looking at the sample size, penetrance and
follow when using human genetics and genomics data variable expressivity of the syndrome.
in early safety assessment of a target gene (Fig. 3).
A large array of human genomics databases are GWAS and rare-variant association studies. Search the
available with information to aid variant interpretation, GWAS literature, including the NHGRI-EBI GWAS
including population sequencing results, variation tol- catalogue in the ‘Reported Genes’ and ‘Mapped Genes’
erance scores, gene-level information including gene– columns179. Be especially critical of associations that do
disease associations, and databases relating to GWAS not reach genome-wide significance (P < 5 × 10−8) or that
and PheWAS results and their interpretation. Selected did not replicate in a larger study in the catalogue of the
examples of each of these are described in Table 4. same phenotype. Advantages of using common SNP
Given the breadth and depth of human genetics data association study results include the large sample size of
available, statistical rigour is required during interrogation individuals carrying the variant, allowing high sensitivity

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 155

0123456789();:
Reviews

a Mendelian genetics
• Direction of effect: LOF or GOF? Relevance to therapeutic hypothesis
Disease-causing (inhibition vs agonism/activation)?
variants implicated in • Clinical features and organ systems affected?
Mendelian disorders • Dominant or recessive inheritance? If recessive, medical information
available for heterozygous carriers?
• Functional assessment (molecular and cellular effects) of variants?
• Disorder due to developmental malformations: if, yes relevance?

Human biallelic LOF data


• Medical history: comprehensive longitudinal clinical information
Individuals with rare available?
biallelic LOF variants • Clinical assessment: can medical observations be attributed to the
variants?
• Functional assessment (molecular and cellular effects) of variants?

Genetic association studies


Target Association in region GWAS
gene in or near gene • Statistical significance/replication?
• Direction of association (increased/decreased risk)?
• Phenotypes and affected organ systems?
• Causality assessment: association driven by variant affecting gene
function/expression?

PheWAS
• Association with other phenotypes?
• Statistical significance/replication?
Variant associated with • Direction of association (increased/decreased risk)?
indication • Phenotypes and affected organ system?
• Causality assessment/regional association context: associations driven
by variant?
Also consider • Association independent of indication association?
• Direct interacting
proteins Rare variant association studies
• Relevant proteins in • Statistical significance? Number of carriers? Replication?
pathway Association driven by • Phenotypes and affected organ systems?
• Protein family members rare variants • Effect of associated variants (partial LOF, GOF)?
• Direction of association (increased/decreased risk)?
• Causality assessment: association driven by rare variants affecting gene?

b
PheWAS of variant V: association with
phenotype P Identification of most likely causal variants
15
GWAS Yes Bayesian Set of most likely
−log10 P value

PheWAS variant summary fine-mapping causal variants


10 statistics No LD-based proxy SNP1 SNP2 SNP3 SNP4 SNP5
available Lead variant set definition Lead
5

Annotation of most likely causal variants


0
74.55 74.60 74.65 74.70 74.75 Disease-relevant pLOF/non-synonymous Non-coding/epigenetic
Chromosome 5 (Mb) functional assay variant annotations
Tissue 1 Tissue 2
GWAS regional association with phenotype P Gene 1 Gene 1 Gene 1
Gene 2 x Gene 2 Gene 2 x
15 PheWAS variant 100 Gene 3 Gene 3 Gene 3
Lead
variant 80
−log10 P value

10 60 Tissues relevance?
40 Interrogation of eQTL and pQTL studies
5
20
GWAS
0 0 summary Conditional Supporting eQTL/
Yes analysis and Posterior
74.55 74.60 74.65 74.70 74.75 pQTL evidence
statistics probability
Chromosome 5 (Mb) Significant available colocalization
eQTL/pQTL Yes anlysis
Gene 1 Gene 3 in tissue/cell Tissues
Gene 2 eQTL/pQTL 1 2 3 4
type in region summary GWAS
No LD Gene 1
statistics lead
available variant Gene 2 x
Gene 3
eQTL
lead
variant Tissue relevance?

156 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

◀ Fig. 3 | Proposed framework for incorporating human germline genetics in target Search for other population-scale association studies
safety reviews. a | Human genetics provides a powerful toolkit to complement not included in the GWAS catalogue, such as reports of
target safety reviews and enables investigators to explore the role of target and individual results from GWAS where the genome-wide
off-target genes in a large spectrum of drug safety-relevant phenotypes in humans, hits were not reported, exome-sequencing studies,
using hypothesis-free or hypothesis-driven approaches. Comprehensive germline rare-variant association study databases such as the
genetic assessment includes interrogation of genome-wide association studies (GWAS),
AstraZeneca PheWAS Portal 53 and GeneBass, and
phenome-wide association studies (PheWAS), rare variant associations, Mendelian
genetics and exploration of individuals with very rare high-impact variants. Important
custom array designs. Be mindful of the statistical
considerations are required to establish the relevance of the genetic findings, including thresholds used.
robustness (quality of the data, report or analysis, statistical significance, replication, If a putative functional variant is available (based on
statistical power), direction of effect and gene causal assessment. b | To interpret association with the indication phenotype, or based
association results from GWAS or PheWAS, a set of downstream analyses are warranted on other strong functional evidence), search for it in GWAS
to establish the causal link between the target or off-target gene and the phenotype and PheWAS databases, for example, PhenoScanner186,
identified in the association analysis. Commonly used methods include identification GWAS ATLAS187, BIG188, the PheWAS Catalogue189 and
and annotation of the most likely causal variant for the association, and interrogation of GRASP190. Apply a P value threshold (that is, multiple
expression quantitative trait loci (eQTL) or protein quantitative trait loci (pQTL) to testing correction) appropriate to the number of pheno-
investigate whether the phenotype association can be attributed to a role of the GWAS
types being tested. If there is a prior concern from pre-
causal variant on the target or off-target gene expression or protein levels in relevant
tissue or cell type. LD, linkage disequilibrium; LOF, loss of function; SNP, single nucleotide
clinical or clinical evidence, this can narrow the search
polymorphism. space and lower the significance threshold, relative to
searching phenome-wide. If a pleiotropic association
is identified, determine the likelihood that the associ-
to modest effects on phenotype; and the sharing of ation is driven by the same causal variant or gene (true
common variation worldwide, allowing the possibility pleiotropy) using the molecular QTL and LD databases
of replication in multiple populations. Disadvantages described above. If the variant is also associated with a
are the challenges in causal gene assignment owing to phenotype related to the indication, use colocalization
the regulatory rather than coding mechanisms usually methods to confirm a shared mechanism between the
responsible and confounding owing to linkage disequi- indication and the pleiotropic association.
librium (LD). Determine level of confidence that the
target gene, rather than another nearby gene, is mecha- Evaluate carcinogenicity risk
nistically responsible for the association signal from the To understand whether the gene may be an oncogene
common variant, as detailed in the following list. or tumour suppressor, search for the gene in databases
• Inspect the regional pattern of LD by looking at nearby of somatic variation and expression in cancer, such as
GWAS results in the browser and in LD databases, the COSMIC Cancer Gene Census148, cBioPortal/The
for example, LDLink180 to ensure that the association Cancer Genome Atlas (TCGA)191 or IntOGen192.
is not a ‘shadow’ — for example, a signal driven by
LD with a much stronger causal association nearby. Evaluate LOF data
Determine whether there are multiple independent If the drug mechanism involves decreasing the function
signals. Look for genes on the haplotype block. or expression of a protein, inspect patterns of LOF var-
• Look for candidate causal variants: identify the credi- iation in population cohorts such as the gnomAD por-
ble set of potential causal variants, using fine-mapping tal and the Bravo portal to TOPMED data193, FinnGen,
approaches if GWAS summary data are available, or Genes and Health and Pakistan Genomic Resource,
by searching for variants in high LD if no summary applying the principles outlined by Minikel et al.95 to
statistics data are available. Search for annotations that carefully inspect the pattern of LOFs across exons and
link variants in credible set to genes (protein-coding their quality scores. If phenotype data are available for
variant, or variants affecting regulatory elements) heterozygous or complete knockouts identified through
using browsers, for example, OpenTargets Genetics, biobank studies, examine these individuals’ phenotype
RegulomeDB181 and HaploReg182. information. If enough carriers of a specific variant are
• Look for molecular QTLs: protein QTL (pQTL) evi- available, perform a candidate-variant PheWAS using an
dence, available in the GWAS browser, and eQTL algorithm optimized for rare variants such as SAIGE194. If
evidence in browsers such as the GTEx Portal183 and multiple rare LOF variants are identified throughout the
eQTLGen184. If an additional association signal has gene, perform a rare-variant association study using an
been found at the locus from these sources, perform aggregation method such as SKAT195, SAIGE-GENE196,
colocalization analysis to confirm a shared mecha- or collapsing analysis197.
nism between the disease association and QTL signal,
using a method such as summary-based Mendelian Future perspectives and conclusions
randomization (SMR)185 or coloc48. Consider the rel- Human genetics studies along with clinical sample bio-
evance of the QTL tissue or cell type to the GWAS banks have opened many new avenues of research in
phenotype. drug discovery and development. Indeed, in the com-
• Formally test MR assumptions using an algorithm ing years vast amounts of genotype–phenotype data are
such as MR-Egger74. expected to be generated from biobank repositories as a
• Look for experimental validation in the literature, result of sequencing millions of exomes and genomes.
paying close attention to whether alternative causal These data will undoubtedly reveal novel human genetic
genes identified above were tested. variants. In addition, populations of non-European

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 157

0123456789();:
Reviews

Table 4 | Selected human genomics resources that might inform target safety assessment
Resource Description Website Refs.
1000 Genomes Population sequencing data (n = 2,504, WES and WGS) https://www.genome. 205

gov/27528684/1000-genomes-project
100,000 Genomes Consortium of genotype–phenotype interactions from UK https://www.genomicsengland.co.uk/ 206

Project population about-genomics-england/the-100000-


genomes-project/
AZ PheWAS Portal Repository of gene–phenotype associations generated using https://azphewas.com/ 53

exome-sequencing and phenotype data from the UK Biobank


BIG Database of GWAS results (brain imaging and other traits from https://big.stats.ox.ac.uk 188

UK Biobank)
BioGRID Database of protein, chemical and genetic interactions https://thebiogrid.org/ 207

Bravo TOPMed Population sequencing data https://bravo.sph.umich.edu/freeze5/hg38/ 193

ClinGen Curated database of clinically relevant variants https://www.clinicalgenome.org/ 177

ClinVar Database of clinically relevant variants https://www.ncbi.nlm.nih.gov/clinvar/ 178

COSMIC Cancer Gene Database of genes implicated in cancer https://cancer.sanger.ac.uk/census 148

Census
Decipher Database of clinically relevant variants https://decipher.sanger.ac.uk/ 208

DisGeNet Database of genes and variants associated with human https://www.disgenet.org/ 209

diseases
East London Genes & Population sequencing data (only list of pLOF variants, not http://www.genesandhealth.org/ 210

Health phenotypes, are freely available)


eQTLGen Database of blood eQTLs https://www.eqtlgen.org/ 184

FinnGen Consortium of genotype–phenotype interactions from Finnish https://www.finngen.fi/en/access_results 211

population
GeneATLAS Database of genotype–phenotype associations from UK http://geneatlas.roslin.ed.ac.uk/ 212

Biobank
GeneBass Repository of gene–phenotype associations generated using https://genebass.org/
exome-sequencing and phenotype data from the UK Biobank
gnomAD Population sequencing data and LOF intolerance scores http://gnomad.broadinstitute.org/ 92

GRASP Curated database of GWAS results https://grasp.nhlbi.nih.gov/Overview.aspx 190

GTEx Portal Database of gene expression and eQTLs across many tissues https://www.gtexportal.org/home/ 213

GWAS ATLAS Database of GWAS summary statistics https://atlas.ctglab.nl 187

GWAS Catalog Curated database of GWAS results https://www.ebi.ac.uk/gwas/ 214

HaploReg Tool to find noncoding regulatory features that overlap given http://compbio.mit.edu/HaploReg/ 182

SNP and LD SNPs


HGMD Database of clinically relevant variants (not freely available https://www.hgmd.cf.ac.uk/ac/index.php 215

to commercial organizations)
LDlink Calculations of linkage disequilibrium between https://ldlink.nci.nih.gov/ 180

1000 Genomes SNPs


MGI PheWeb Database of GWAS results from the Michigan Genomics https://pheweb.sph.umich.edu/
Initiative and UK Biobank
Missense Tolerance Ratio Regional variation tolerance scores http://mtr-viewer.mdhs.unimelb.edu.au 216

OMIM Curated database of genetic disease https://omim.org/ 217

Open Targets Database of genetic information in the context of drug https://www.opentargets.org/ 218

target validation
Orion Noncoding variation tolerance scores http://www.genomic-orion.org/ 219

PhenoScanner Database of genotype–phenotype associations http://www.phenoscanner.medschl.cam.ac.uk/ 220

PheWAS Catalogue Database of PheWAS results from BioVU https://phewascatalog.org/ 189

RegulomeDB Tool to find noncoding regulatory features that overlap https://www.regulomedb.org/ 181

given SNP
RVIS Gene-level variation tolerance score https://genic-intolerance.org/ 91

The cBioPortal for Curated somatic mutation database https://www.cbioportal.org/ 191,221

Cancer Genomics
UniProt Database of protein information https://www.uniprot.org/ 222

eQTL, expression quantitative trait loci; GWAS, genome-wide association studies; LD, linkage disequilibrium; LOF, loss of function; OMIM, Online Mendelian
Inheritance in Man; PheWAS, phenome-wide association studies; SNP, single nucleotide polymorphism; WES, whole exome sequencing; WGS, whole genome
sequencing.

158 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

ancestry are currently very under-represented in most concerns should be considered in terms of informing
human genetics studies, resulting in reduced under- preclinical and clinical development, patient risk–
standing of the range and functional consequences of benefit and patient management strategies remains to be
genetic variation within these populations198. Efforts by further defined. Drug targets and off-targets for which
the human genetics community will continue to redress predicted safety signals are identified by human genetic
this population disparity. Furthermore, the expansion data could be included in secondary pharmacology
of rare variant discovery into the noncoding portions screening or inform the strategy and design of early pre-
of the genome will provide an even greater opportunity clinical safety screens or clinical monitoring strategies.
for the identification of candidate protective alleles for They may also provide a strategy of generating testable
various disease states. hypotheses that can inform a proactive approach to fur-
Large-scale deeply phenotyped biobanks are being ther assessing drug safety at early stages of development.
created that contain arrays of biomarkers and catalogue Furthermore, human genetic data from biobanks can
various disease states and subtypes. The development of provide biological plausibility or implausibility for the
new genotyping technologies and computational meth- relevance of drug AEs reported from clinical develop-
odologies such as deep mutational scanning (DMS) and ment and pharmacovigilance, and provide evidence for
the priority index (Pi) pipeline199 may provide mecha- post-marketing regulatory review. Unanticipated safety
nistic insights, potentially predict the functional or clin- signals that occur during later phase drug development,
ical impact of these variants of uncertain significance or that are reported from pharmacovigilance, may be
and prioritize the genes and drug targets responsible for further investigated for possible genetic associations to
GWAS signals200. support patient stratification and selection.
Recent advances in human epigenome mapping have The use of human genetics in support of drug safety
provided insights into the importance of human non- assessment is gaining significant momentum and is com-
coding DNA regulatory elements such as enhancers for plementary to informatics evaluations of known and
controlling cell-type-specific genome functions201,202. putative target safety concerns9. However, genetics-based
Most human disease-associated genetic variation evidence from emerging MR and biobank studies will
maps to the regulatory genome, and many GWAS var- need to be better integrated with various preclinical and
iants directly affect transcription factor occupancy201. clinical safety data resources. Pre-competitive sharing
Thus, increased attention is being given to integrating of genetic and drug safety data, national initiatives for
cell- and tissue-specific epigenomic profiles, including biobanks, and industry and academic partnerships
large-scale epigenome-wide association studies (EWAS), (for example, OpenTargets) will be important stimuli for
for the characterization of genotype–phenotype relation­ further evaluation of broader drug safety assessment appli-
ships203,204. The unprecedented access to genome-centric cations. We outline here some suggested best practices for
phenotypic data, biobanks and computational tools is the rigorous interpretation of potential drug target-related
enabling the identification of new drug target candi- genotype–phenotype associations that should help guide
dates and providing deeper insights into disease aetiol- future discussions on the regulatory value of human
ogy, including multifactorial or complex diseases with genetics-based safety assessment data, potentially facili-
multiple common genetic variants. tated by a cross-sector working group including industry,
Human genetics is becoming an increasingly impor- academic and regulatory stakeholders.
tant approach for early target de-risking during drug
development. How these genetically defined safety Published online 19 October 2022

1. Visscher, P. M. et al. 10 years of GWAS discovery: 9. Roberts, R. A. Understanding drug targets: no 17. Tegeder, I. et al. GTP cyclohydrolase and
biology, function, and translation. Am. J. Hum. Genet. such thing as bad news. Drug Discov. Today 23, tetrahydrobiopterin regulate pain sensitivity and
101, 5–22 (2017). 1925–1928 (2018). persistence. Nat. Med. 12, 1269–1277 (2006).
2. Chong, J. X. et al. The genetic basis of Mendelian 10. Okada, Y. et al. Genetics of rheumatoid arthritis 18. Bonafe, L., Thony, B., Penzien, J. M., Czarnecki, B.
phenotypes: discoveries, challenges, and contributes to biology and drug discovery. Nature & Blau, N. Mutations in the sepiapterin reductase
opportunities. Am. J. Hum. Genet. 97, 199–215 506, 376–381 (2014). gene cause a novel tetrahydrobiopterin-dependent
(2015). 11. Diogo, D. et al. Phenome-wide association studies monoamine-neurotransmitter deficiency without
3. Plenge, R. M., Scolnick, E. M. & Altshuler, D. across large population cohorts support drug hyperphenylalaninemia. Am. J. Hum. Genet. 69,
Validating therapeutic targets through human target validation. Nat. Commun. 9, 4285 269–277 (2001).
genetics. Nat. Rev. Drug Discov. 12, 581–594 (2013). (2018). 19. Thony, B. & Blau, N. Mutations in the
4. Kamb, A., Harper, S. & Stefansson, K. Human genetics 12. Jerome, R. N. et al. Leveraging human genetics BH4-metabolizing genes GTP cyclohydrolase I,
as a foundation for innovative drug development. to identify safety signals prior to drug marketing 6-pyruvoyl-tetrahydropterin synthase, sepiapterin
Nat. Biotechnol. 31, 975–978 (2013). approval and clinical use. Drug Saf. 43, 567–582 reductase, carbinolamine-4a-dehydratase, and
5. Nelson, M. R. et al. The support of human genetic (2020). dihydropteridine reductase. Hum. Mutat. 27,
evidence for approved drug indications. Nat. Genet. 13. Nguyen, P. A., Born, D. A., Deaton, A. M., Nioi, P. 870–878 (2006).
47, 856–860 (2015). & Ward, L. D. Phenotypes associated with genes 20. Booth, B. Painful Truth: The Successful Failure Of
6. King, E. A., Davis, J. W. & Degner, J. F. Are drug encoding drug targets are predictive of clinical trial A Biotech Startup. Forbes (17 November 2017);
targets with genetic support twice as likely to be side effects. Nat. Commun. 10, 1579 (2019). https://www.forbes.com/sites/brucebooth/2017/
approved? Revised estimates of the impact of genetic 14. Cao, J. et al. Targeting acyl-CoA:diacylglycerol 11/17/painful-truth-successful-failure-of-a-biotech-
support for drug mechanisms on the probability of acyltransferase 1 (DGAT1) with small molecule startup
drug approval. PLoS Genet. 15, e1008489 (2019). inhibitors for the treatment of metabolic diseases. 21. Sharfe, N., Dadi, H. K., Shahar, M. & Roifman, C. M.
7. Monticello, T. M. et al. Current nonclinical testing J. Biol. Chem. 286, 41838–41851 (2011). Human immune disorder arising from mutation
paradigm enables safe entry to first-in-human 15. Denison, H. et al. Diacylglycerol acyltransferase 1 of the alpha chain of the interleukin-2 receptor.
clinical trials: the IQ consortium nonclinical to clinical inhibition with AZD7687 alters lipid handling and Proc. Natl Acad. Sci. USA 94, 3168–3171 (1997).
translational database. Toxicol. Appl. Pharmacol. hormone secretion in the gut with intolerable side 22. Caudy, A. A., Reddy, S. T., Chatila, T., Atkinson, J. P.
334, 100–109 (2017). effects: a randomized clinical trial. Diabetes Obes. & Verbsky, J. W. CD25 deficiency causes an immune
8. Waring, M. J. et al. An analysis of the attrition Metab. 16, 334–343 (2014). dysregulation, polyendocrinopathy, enteropathy,
of drug candidates from four major pharmaceutical 16. Haas, J. T. et al. DGAT1 mutation is linked to a X-linked-like syndrome, and defective IL-10 expression
companies. Nat. Rev. Drug Discov. 14, 475–486 congenital diarrheal disorder. J. Clin. Invest. 122, from CD4 lymphocytes. J. Allergy Clin. Immunol. 119,
(2015). 4680–4684 (2012). 482–487 (2007).

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 159

0123456789();:
Reviews

23. Goudy, K. et al. Human IL2RA null mutation mediates advance precision medicine. Annu. Rev. Genomics hypercholesterolemia or atherosclerosis. J. Am. Coll.
immunodeficiency with lymphoproliferation and Hum. Genet. 17, 353–373 (2016). Cardiol. 77, 1182–1193 (2021).
autoimmunity. Clin. Immunol. 146, 248–261 (2013). 47. Duffy, A. et al. Tissue-specific genetic features inform 72. Leiter, L. A. et al. Alirocumab safety in people with
24. Tang, Q. et al. Central role of defective interleukin-2 prediction of drug side effects in clinical trials. Sci. Adv. and without diabetes mellitus: pooled data from
production in the triggering of islet autoimmune 6, eabb6242 (2020). 14 ODYSSEY trials. Diabet. Med. 35, 1742–1751
destruction. Immunity 28, 687–697 (2008). 48. Giambartolomei, C. et al. Bayesian test for (2018).
25. Prasad, N. et al. Is basiliximab induction, a novel risk colocalisation between pairs of genetic association 73. Da Dalt, L. et al. PCSK9 deficiency reduces insulin
factor for new onset diabetes after transplantation for studies using summary statistics. PLoS Genet. 10, secretion and promotes glucose intolerance: the role
living donor renal allograft recipients? Nephrology 19, e1004383 (2014). of the low-density lipoprotein receptor. Eur. Heart J.
244–250 (2014). 49. Evans, D. M. & Davey Smith, G. Mendelian 40, 357–368 (2019).
26. Lo, B. et al. Patients with LRBA deficiency show randomization: new applications in the coming age 74. Bowden, J., Davey Smith, G. & Burgess, S.
CTLA4 loss and immune dysregulation responsive to of hypothesis-free causality. Annu. Rev. Genomics Mendelian randomization with invalid instruments:
abatacept therapy. Science 349, 436–440 (2015). Hum. Genet. 16, 327–350 (2015). effect estimation and bias detection through Egger
27. Lopez-Herrera, G. et al. Deleterious mutations in 50. Szustakowski, J. D. et al. Advancing human genetics regression. Int. J. Epidemiol. 44, 512–525 (2015).
LRBA are associated with a syndrome of immune research and drug discovery through exome 75. Garrelfs, S. F. et al. Lumasiran, an RNAi therapeutic
deficiency and autoimmunity. Am. J. Hum. Genet. 90, sequencing of the UK Biobank. Nat. Genet. 53, for primary hyperoxaluria type 1. N. Engl. J. Med.
986–1001 (2012). 942–948 (2021). 384, 1216–1226 (2021).
28. Alangari, A. et al. LPS-responsive beige-like anchor 51. Bycroft, C. et al. The UK Biobank resource with 76. McGregor, T. L. et al. Characterising a healthy adult
(LRBA) gene mutation in a family with inflammatory deep phenotyping and genomic data. Nature 562, with a rare HAO1 knockout to support a therapeutic
bowel disease and combined immunodeficiency. 203–209 (2018). strategy for primary hyperoxaluria. eLife 9, e54363
J. Allergy Clin. Immunol. 130, 481–488.e482 (2012). 52. Backman, J. D. et al. Exome sequencing and analysis (2020).
29. Charbonnier, L. M. et al. Regulatory T-cell deficiency of 454,787 UK Biobank participants. Nature 599, 77. Paisan-Ruiz, C. et al. Cloning of the gene containing
and immune dysregulation, polyendocrinopathy, 628–634 (2021). mutations that cause PARK8-linked Parkinson’s
enteropathy, X-linked-like disorder caused by loss-of- 53. Wang, Q. et al. Rare variant contribution to human disease. Neuron 44, 595–600 (2004).
function mutations in LRBA. J. Allergy Clin. Immunol. disease in 281,104 UK Biobank exomes. Nature 597, 78. Zimprich, A. et al. Mutations in LRRK2 cause
135, 217–227 (2015). 527–532 (2021). autosomal-dominant parkinsonism with pleomorphic
30. Bertrand, A., Kostine, M., Barnetche, T., Truchetet, M. E. 54. Millard, L. A. et al. MR-PheWAS: hypothesis pathology. Neuron 44, 601–607 (2004).
& Schaeverbeke, T. Immune related adverse events prioritization among potential causal effects of body 79. Simon-Sanchez, J. et al. Genome-wide association
associated with anti-CTLA-4 antibodies: systematic mass index on many outcomes, using Mendelian study reveals genetic risk underlying Parkinson’s
review and meta-analysis. BMC Med. 13, 211 (2015). randomization. Sci. Rep. 5, 16645 (2015). disease. Nat. Genet. 41, 1308–1312 (2009).
31. Bouhassira, E. E. et al. An alanine-to-threonine 55. Gill, D. et al. Associations of genetically determined 80. Whiffin, N. et al. The effect of LRRK2 loss-of-function
substitution in protein 4.2 cDNA is associated iron status across the phenome: a Mendelian variants in humans. Nat. Med. 26, 869–877 (2020).
with a Japanese form of hereditary hemolytic anemia randomization study. PLoS Med. 16, e1002833 81. Samson, M. et al. Resistance to HIV-1 infection
(protein 4.2NIPPON). Blood 79, 1846–1854 (1992). (2019). in caucasian individuals bearing mutant alleles of
32. Bruce, L. J. et al. Absence of CD47 in protein 56. Zheng, J. et al. Phenome-wide Mendelian randomization the CCR-5 chemokine receptor gene. Nature 382,
4.2-deficient hereditary spherocytosis in man: an mapping the influence of the plasma proteome on 722–725 (1996).
interaction between the Rh complex and the band 3 complex diseases. Nat. Genet. 52, 1122–1131 (2020). 82. Liu, R. et al. Homozygous defect in HIV-1 coreceptor
complex. Blood 100, 1878–1885 (2002). 57. Smith, G. D. & Ebrahim, S. ‘Mendelian accounts for resistance of some multiply-exposed
33. Jiang, Z., Sun, H., Yu, J., Tian, W. & Song, Y. Targeting randomization’: can genetic epidemiology contribute individuals to HIV-1 infection. Cell 86, 367–377
CD47 for cancer immunotherapy. J. Hematol. Oncol. to understanding environmental determinants of (1996).
14, 180 (2021). disease? Int. J. Epidemiol. 32, 1–22 (2003). 83. Emmelkamp, J. M. & Rockstroh, J. K. CCR5
34. Jin, Y. et al. Genome-wide association studies of 58. Abifadel, M. et al. Mutations in PCSK9 cause antagonists: comparison of efficacy, side effects,
autoimmune vitiligo identify 23 new risk loci and autosomal dominant hypercholesterolemia. pharmacokinetics and interactions–review of the
highlight key pathways and regulatory variants. Nat. Genet. 34, 154–156 (2003). literature. Eur. J. Med. Res. 12, 409–417 (2007).
Nat. Genet. 48, 1418–1424 (2016). 59. Timms, K. M. et al. A mutation in PCSK9 causing 84. Tebas, P. et al. Gene editing of CCR5 in autologous
35. Petukhova, L. et al. Genome-wide association study autosomal-dominant hypercholesterolemia in a Utah CD4 T cells of persons infected with HIV. N. Engl.
in alopecia areata implicates both innate and adaptive pedigree. Hum. Genet. 114, 349–353 (2004). J. Med. 370, 901–910 (2014).
immunity. Nature 466, 113–117 (2010). 60. Di Taranto, M. D. et al. Identification and in vitro 85. Nag, A. et al. Human genetic evidence supports
36. Betz, R. C. et al. Genome-wide meta-analysis in characterization of two new PCSK9 gain of MAP3K15 inhibition as a therapeutic strategy
alopecia areata resolves HLA associations and reveals function variants found in patients with familial for diabetes. medRxiv https://doi.org/10.1101/
two new susceptibility loci. Nat. Commun. 6, 5966 hypercholesterolemia. Sci. Rep. 7, 15282 (2017). 2021.11.14.21266328 (2021).
(2015). 61. Cohen, J. et al. Low LDL cholesterol in individuals 86. Narasimhan, V. M. et al. Health and population
37. Choi, L. et al. Evaluating statistical approaches to of African descent resulting from frequent nonsense effects of rare gene knockouts in adult humans
leverage large clinical datasets for uncovering mutations in PCSK9. Nat. Genet. 37, 161–165 with related parents. Science 352, 474–477
therapeutic and adverse medication effects. (2005). (2016).
Bioinformatics 34, 2988–2996 (2018). 62. Cohen, J. C., Boerwinkle, E., Mosley, T. H. Jr. & 87. Van Hout, C. V. et al. Exome sequencing and
38. Rao, A. S. et al. Large-scale phenome-wide association Hobbs, H. H. Sequence variations in PCSK9, low LDL, characterization of 49,960 individuals in the UK
study of PCSK9 variants demonstrates protection and protection against coronary heart disease. Biobank. Nature 586, 749–756 (2020).
against ischemic stroke. Circ. Genom. Precis. Med. 11, N. Engl. J. Med. 354, 1264–1272 (2006). 88. Lim, E. T. et al. Distribution and medical impact
e002162 (2018). 63. Kathiresan, S. et al. Six new loci associated with of loss-of-function variants in the Finnish founder
39. Ference, B. A. et al. Mendelian randomization study blood low-density lipoprotein cholesterol, high-density population. PLoS Genet. 10, e1004494 (2014).
of ACLY and cardiovascular disease. N. Engl. J. Med. lipoprotein cholesterol or triglycerides in humans. 89. Tanigawa, Y. et al. Rare protein-altering variants in
380, 1033–1042 (2019). Nat. Genet. 40, 189–197 (2008). ANGPTL7 lower intraocular pressure and protect
40. Interleukin-6 Receptor Mendelian Randomisation 64. Willer, C. J. et al. Newly identified loci that influence against glaucoma. PLoS Genet. 16, e1008682
Analysis (IL6R MR) Consortium The interleukin-6 lipid concentrations and risk of coronary artery (2020).
receptor as a target for prevention of coronary heart disease. Nat. Genet. 40, 161–169 (2008). 90. Saleheen, D. et al. Human knockouts and phenotypic
disease: a mendelian randomisation analysis. Lancet 65. Hooper, A. J., Marais, A. D., Tanyanyiwa, D. M. analysis in a cohort with a high rate of consanguinity.
379, 1214–1224 (2012). & Burnett, J. R. The C679X mutation in PCSK9 is Nature 544, 235–239 (2017).
41. Walker, V. M., Davey Smith, G., Davies, N. M. & present and lowers blood cholesterol in a Southern 91. Petrovski, S., Wang, Q., Heinzen, E. L., Allen, A. S.
Martin, R. M. Mendelian randomization: a novel African population. Atherosclerosis 193, 445–448 & Goldstein, D. B. Genic intolerance to functional
approach for the prediction of adverse drug events (2007). variation and the interpretation of personal genomes.
and drug repurposing opportunities. Int. J. Epidemiol. 66. Schmidt, A. F. et al. PCSK9 genetic variants and risk PLoS Genet. 9, e1003709 (2013).
46, 2078–2089 (2017). of type 2 diabetes: a Mendelian randomisation study. 92. Karczewski, K. J. et al. The mutational constraint
42. Swerdlow, D. I. et al. HMG-coenzyme A reductase Lancet Diabetes Endocrinol. 5, 97–105 (2017). spectrum quantified from variation in 141,456
inhibition, type 2 diabetes, and bodyweight: evidence 67. Williams, D. M., Finan, C., Schmidt, A. F., Burgess, S. humans. Nature 581, 434–443 (2020).
from genetic analysis and randomised trials. Lancet & Hingorani, A. D. Lipid lowering and Alzheimer 93. Petrovski, S. et al. The intolerance of regulatory
385, 351–361 (2015). disease risk: a Mendelian randomization study. sequence to genetic variation predicts gene dosage
43. Alghamdi, J. et al. Risk of neuropsychiatric adverse Ann. Neurol. 87, 30–39 (2020). sensitivity. PLoS Genet. 11, e1005492 (2015).
effects of lipid-lowering drugs: a Mendelian 68. Benn, M., Nordestgaard, B. G., Frikke-Schmidt, R. 94. Begum, T., Ghosh, T. C. & Basak, S. Systematic
randomization study. Int. J. Neuropsychopharmacol. & Tybjaerg-Hansen, A. Low LDL cholesterol, PCSK9 analyses and prediction of human drug side effect
21, 1067–1075 (2018). and HMGCR genetic variation, and risk of Alzheimer’s associated proteins from the perspective of protein
44. Interleukin 1 Genetics Consortium. Cardiometabolic disease and Parkinson’s disease: Mendelian evolution. Genome Biol. Evol. 9, 337–350 (2017).
effects of genetic upregulation of the interleukin 1 randomisation study. BMJ 357, j1648 (2017). 95. Minikel, E. V. et al. Evaluating drug targets through
receptor antagonist: a Mendelian randomisation 69. Schmidt, A. F. et al. Phenome-wide association human loss-of-function genetic variation. Nature 581,
analysis. Lancet Diabetes Endocrinol. 3, 243–253 analysis of LDL-cholesterol lowering genetic 459–464 (2020).
(2015). variants in PCSK9. BMC Cardiovasc. Disord. 19, 96. Trochet, D., Prudhon, B., Vassilopoulos, S. &
45. Bush, W. S., Oetjens, M. T. & Crawford, D. C. 240 (2019). Bitoun, M. Therapy for dominant inherited diseases by
Unravelling the human genome-phenome 70. Sabatine, M. S. et al. Efficacy and safety of allele-specific RNA interference: successes and pitfalls.
relationship using phenome-wide association studies. evolocumab in reducing lipids and cardiovascular Curr. Gene Ther. 15, 503–510 (2015).
Nat. Rev. Genet. 17, 129–145 (2016). events. N. Engl. J. Med. 372, 1500–1509 (2015). 97. Rook, M. E. & Southwell, A. L. Antisense
46. Denny, J. C., Bastarache, L. & Roden, D. M. 71. Wright, R. S. et al. Pooled patient-level analysis oligonucleotide therapy: from design to the huntington
Phenome-wide association studies as a tool to of inclisiran trials in patients with familial disease clinic. BioDrugs 36, 105–119 (2022).

160 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:
Reviews

98. Nagasaka, M. et al. Beyond osimertinib: the Drugs for Treatment: Guidance for Industry, https:// 148. Forbes, S. A. et al. COSMIC: somatic cancer genetics
development of third-generation EGFR tyrosine www.fda.gov/media/117977/download (2022). at high-resolution. Nucleic Acids Res. 45, D777–D783
kinase inhibitors for advanced EGFR+ NSCLC. 124. Scott, D. A. & Zhang, F. Implications of human genetic (2017).
J. Thorac. Oncol. 16, 740–763 (2021). variation in CRISPR-based therapeutic genome 149. Abul-Husn, N. S. & Kenny, E. E. Personalized medicine
99. Bowes, J. et al. Reducing safety-related drug editing. Nat. Med. 23, 1095–1101 (2017). and the power of electronic health records. Cell 177,
attrition: the use of in vitro pharmacological profiling. 125. Guidance Document: Human Gene Therapy Products 58–69 (2019).
Nat. Rev. Drug Discov. 11, 909–922 (2012). Incorporating Human Genome Editing (US Food and 150. Expert Panel on Detection, Evaluation, and Treatment
100. Whitebread, S. et al. Secondary pharmacology: Drug Administration, 2022); https://www.fda.gov/ of High Blood Cholesterol in Adults. Executive
screening and interpretation of off-target activities - regulatory-information/search-fda-guidance-documents/ summary of the third report of the national cholesterol
focus on translation. Drug Discov. Today 21, human-gene-therapy-products-incorporating-human- education program (NCEP) Expert panel on detection,
1232–1242 (2016). genome-editing evaluation, and treatment of high blood cholesterol
101. ICH. Guidance for Industry: S7A Safety Pharmacology 126. Moggs, J. G., MacLachlan, T., Martus, H. J. & in adults (Adult Treatment Panel III). JAMA 285,
Studies for Human Pharmaceuticals (2001). Bentley, P. Derisking drug-induced carcinogenicity 2486–2497 (2001).
102. Hamon, J. et al. In vitro safety pharmacology profiling: for novel therapeutics. Trends Cancer 2, 398–408 151. Mora, S. et al. Lipoprotein(a) and risk of type 2
what else beyond hERG? Future Med. Chem. 1, (2016). diabetes. Clin. Chem. 56, 1252–1260 (2010).
645–665 (2009). 127. Fielden, M. R. et al. Modernizing human cancer risk 152. Gudbjartsson, D. F. et al. Lipoprotein(a) concentration
103. Curran, M. E. et al. A molecular basis for cardiac assessment of therapeutics. Trends Pharmacol. Sci. and risks of cardiovascular disease and diabetes.
arrhythmia: HERG mutations cause long QT syndrome. 39, 232–247 (2018). J. Am. Coll. Cardiol. 74, 2982–2994 (2019).
Cell 80, 795–803 (1995). 128. Dumont, N. & Arteaga, C. L. The tumor 153. Okada, S. et al. Impairment of immunity to Candida
104. Kannankeril, P., Roden, D. M. & Darbar, D. microenvironment: a potential arbitrator of the and Mycobacterium in humans with bi-allelic
Drug-induced long QT syndrome. Pharmacol. Rev. tumor suppressive and promoting actions of TGFbeta. RORC mutations. Science 349, 606–613 (2015).
62, 760–781 (2010). Differentiation 70, 574–582 (2002). 154. Gal, A. et al. Mutations in MERTK, the human
105. Paulussen, A. D. et al. Genetic variations of KCNQ1, 129. Caja, F. & Vannucci, L. TGFbeta: a player on multiple orthologue of the RCS rat retinal dystrophy gene,
KCNH2, SCN5A, KCNE1, and KCNE2 in drug-induced fronts in the tumor microenvironment. J. Immunotoxicol. cause retinitis pigmentosa. Nat. Genet. 26, 270–271
long QT syndrome patients. J. Mol. Med. 82, 12, 300–307 (2015). (2000).
182–188 (2004). 130. Qin, T. et al. A novel highly potent trivalent TGF-beta 155. Sayama, A. et al. UNC569-induced morphological
106. Chen, Q. et al. Genetic basis and molecular receptor trap inhibits early-stage tumorigenesis changes in pigment epithelia and photoreceptor
mechanism for idiopathic ventricular fibrillation. and tumor cell invasion in murine Pten-deficient cells in the retina through MerTK inhibition in mice.
Nature 392, 293–296 (1998). prostate glands. Oncotarget 7, 86087–86102 Toxicol. Pathol. 46, 193–201 (2018).
107. Deaton, A. M. et al. Rationalizing secondary (2016). 156. Koonin, E. V., Wolf, Y. I. & Karev, G. P. The structure
pharmacology screening using human genetic 131. Grenga, I. et al. Anti-PD-L1/TGFbetaR2 (M7824) of the protein universe and genome evolution. Nature
and pharmacological evidence. Toxicol. Sci. 167, fusion protein induces immunogenic modulation of 420, 218–223 (2002).
593–603 (2019). human urothelial carcinoma cell lines, rendering them 157. Ekman, D., Bjorklund, A. K., Frey-Skott, J. &
108. Liu, X. et al. A proteomic platform to identify off-target more susceptible to immune-mediated recognition Elofsson, A. Multi-domain proteins in the three
proteins associated with therapeutic modalities that and lysis. Urol. Oncol. 36, 93.e1–93.e11 (2018). kingdoms of life: orphan domains and other unassigned
induce protein degradation or gene silencing. Sci. Rep. 132. Goudie, D. R. et al. Multiple self-healing squamous regions. J. Mol. Biol. 348, 231–243 (2005).
11, 15856 (2021). epithelioma is caused by a disease-specific spectrum 158. Wang, H. et al. Cell-specific mechanisms of TMEM16A
109. Siintola, E. et al. Cathepsin D deficiency underlies of mutations in TGFBR1. Nat. Genet. 43, 365–369 Ca2+-activated chloride channel in cancer. Mol. Cancer
congenital human neuronal ceroid-lipofuscinosis. (2011). 16, 152 (2017).
Brain 129, 1438–1445 (2006). 133. Lacouture, M. E. et al. Cutaneous keratoacanthomas/ 159. Crottes, D. & Jan, L. Y. The multifaceted role of
110. Gisolfi, C. V., Summers, R. W., Schedl, H. P. & squamous cell carcinomas associated with TMEM16A in cancer. Cell Calcium 82, 102050 (2019).
Bleiler, T. L. Intestinal water absorption from select neutralization of transforming growth factor beta by 160. Bill, A. et al. Small molecule-facilitated degradation of
carbohydrate solutions in humans. J. Appl. Physiol. the monoclonal antibody fresolimumab (GC1008). ANO1 protein: a new targeting approach for anticancer
73, 2142–2150 (1992). Cancer Immunol. Immunother. 64, 437–446 (2015). therapeutics. J. Biol. Chem. 289, 11029–11041 (2014).
111. Zuhl, A. M. et al. Chemoproteomic profiling reveals 134. Strauss, J. et al. Phase I trial of M7824 161. Bill, A. et al. ANO1/TMEM16A interacts with EGFR
that cathepsin D off-target activity drives ocular (MSB0011359C), a bifunctional fusion protein and correlates with sensitivity to EGFR-targeting
toxicity of beta-secretase inhibitors. Nat. Commun. targeting PD-L1 and TGFbeta, in advanced solid therapy in head and neck cancer. Oncotarget 6,
7, 13042 (2016). tumors. Clin. Cancer Res. 24, 1287–1295 (2018). 9173–9188 (2015).
112. Debs, R. et al. Biotin-responsive basal ganglia disease 135. Arnault, J. P. et al. Keratoacanthomas and squamous 162. Bill, A. & Alex Gaither, L. The mechanistic role of the
in ethnic Europeans with novel SLC19A3 mutations. cell carcinomas in patients receiving sorafenib. calcium-activated chloride channel ANO1 in tumor
Arch. Neurol. 67, 126–130 (2010). J. Clin. Oncol. 27, e59–e61 (2009). growth and signaling. Adv. Exp. Med. Biol. 966,
113. Kono, S. et al. Mutations in a thiamine-transporter 136. Arnault, J. P. et al. Skin tumors induced by 1–14 (2017).
gene and Wernicke’s-like encephalopathy. N. Engl. sorafenib; paradoxic RAS-RAF pathway activation 163. Farnaby, W. et al. BAF complex vulnerabilities in
J. Med. 360, 1792–1794 (2009). and oncogenic mutations of HRAS, TP53, and cancer demonstrated via structure-based PROTAC
114. Zhang, Q. et al. The Janus kinase 2 inhibitor fedratinib TGFBR1. Clin. Cancer Res. 18, 263–272 (2012). design. Nat. Chem. Biol. 15, 672–680 (2019).
inhibits thiamine uptake: a putative mechanism for 137. Carlino, M. S. et al. Correlation of BRAF and 164. Burska, A., Boissinot, M. & Ponchel, F. Cytokines as
the onset of Wernicke’s encephalopathy. Drug. Metab. NRAS mutation status with outcome, site of biomarkers in rheumatoid arthritis. Mediators Inflamm.
Dispos. 42, 1656–1662 (2014). distant metastasis and response to chemotherapy in 2014, 545493 (2014).
115. Donovan, K. A. et al. Thalidomide promotes metastatic melanoma. Br. J. Cancer 111, 292–299 165. Sirugo, G., Williams, S. M. & Tishkoff, S. A. The
degradation of SALL4, a transcription factor (2014). missing diversity in human genetic studies. Cell 177,
implicated in Duane radial ray syndrome. eLife 7, 138. Daver, N., Schlenk, R. F., Russell, N. H. & Levis, M. J. 26–31 (2019).
e38430 (2018). Targeting FLT3 mutations in AML: review of current 166. Popejoy, A. B. & Fullerton, S. M. Genomics is failing
116. Matyskiela, M. E. et al. SALL4 mediates teratogenicity knowledge and evidence. Leukemia 33, 299–312 on diversity. Nature 538, 161–164 (2016).
as a thalidomide-dependent cereblon substrate. (2019). 167. Fatumo, S. et al. A roadmap to increase diversity in
Nat. Chem. Biol. 14, 981–987 (2018). 139. Goldman, J. M. Chronic myeloid leukemia: a historical genomic studies. Nat. Med. 28, 243–250 (2022).
117. Belair, D. G. et al. Thalidomide inhibits human perspective. Semin. Hematol. 47, 302–311 (2010). 168. Hindorff, L. A. et al. Prioritizing diversity in human
iPSC mesendoderm differentiation by modulating 140. Muller, P. A. & Vousden, K. H. p53 mutations in genomics research. Nat. Rev. Genet. 19, 175–185
CRBN-dependent degradation of SALL4. Sci. Rep. 10, cancer. Nat. Cell Biol. 15, 2–8 (2013). (2018).
2864 (2020). 141. Cox, A. D. & Der, C. J. The raf inhibitor paradox: 169. All of Us Research Program Investigators. et al. The
118. Kohlhase, J. et al. Okihiro syndrome is caused by unexpected consequences of targeted drugs. “All of Us” Research Program. N. Engl. J. Med. 381,
SALL4 mutations. Hum. Mol. Genet. 11, 2979–2987 Cancer Cell 17, 221–223 (2010). 668–676 (2019).
(2002). 142. McDonald, E. R. III et al. Project DRIVE: a 170. de Vries, J. et al. Ethical issues in human genomics
119. Kohlhase, J. et al. Mutations at the SALL4 locus compendium of cancer dependencies and synthetic research in developing countries. BMC Med. Ethics
on chromosome 20 result in a range of clinically lethal relationships uncovered by large-scale, deep 12, 5 (2011).
overlapping phenotypes, including Okihiro syndrome, RNAi screening. Cell 170, 577–592.e10 (2017). 171. Munung, N. S. & de Vries, J. Benefit sharing
Holt-Oram syndrome, acro-renal-ocular syndrome, 143. de Weck, A. et al. Correction of copy number induced for human genomics research: awareness and
and patients previously reported to represent false positives in CRISPR screens. PLoS Comput. Biol. expectations of genomics researchers in Sub-Saharan
thalidomide embryopathy. J. Med. Genet. 40, 14, e1006279 (2018). Africa. Ethics Hum. Res. 42, 14–20 (2020).
473–478 (2003). 144. Rauscher, B., Heigwer, F., Breinig, M., Winter, J. 172. Pennisi, E. Genomes arising. Science 371, 556–559
120. Vargesson, N. Thalidomide-induced teratogenesis: & Boutros, M. GenomeCRISPR – a database (2021).
history and mechanisms. Birth Defects Res. C. for high-throughput CRISPR/Cas9 screens. 173. Maxmen, A. The next chapter for African genomics.
Embryo Today 105, 140–156 (2015). Nucleic Acids Res. 45, D679–D686 (2017). Nature 578, 350–354 (2020).
121. Janas, M. M. et al. Selection of GalNAc-conjugated 145. Flaherty, K. T. et al. Inhibition of mutated, activated 174. Munafo, M. R. & Gage, S. H. Improving the reliability
siRNAs with limited off-target-driven rat BRAF in metastatic melanoma. N. Engl. J. Med. 363, and reporting of genetic association studies.
hepatotoxicity. Nat. Commun. 9, 723 (2018). 809–819 (2010). Drug Alcohol Depend. 132, 411–413 (2013).
122. Burel, S. A. et al. Hepatotoxicity of high affinity 146. Fu, Y. et al. High-frequency off-target mutagenesis 175. Amberger, J. S. & Hamosh, A. Searching online
gapmer antisense oligonucleotides is mediated by induced by CRISPR-Cas nucleases in human cells. Mendelian inheritance in man (OMIM):
RNase H1 dependent promiscuous reduction of very Nat. Biotechnol. 31, 822–826 (2013). a knowledgebase of human genes and genetic
long pre-mRNA transcripts. Nucleic Acids Res. 44, 147. Pattanayak, V. et al. High-throughput profiling of phenotypes. Curr. Protoc. Bioinformatics 58,
2093–2109 (2016). off-target DNA cleavage reveals RNA-programmed 1.2.1–1.2.12 (2017).
123. US Department of Health and Human Services. Cas9 nuclease specificity. Nat. Biotechnol. 31, 176. Richards, S. et al. Standards and guidelines for the
Chronic Hepatitis B Virus Infection: Developing 839–843 (2013). interpretation of sequence variants: a joint consensus

NaTure RevIeWS | DRuG DISCOveRy volume 22 | February 2023 | 161

0123456789();:
Reviews

recommendation of the American College of Medical 204. Lappalainen, T. & Greally, J. M. Associating cellular 231. Roses, A. D. Pharmacogenetics and drug
Genetics and Genomics and the Association for epigenetic models with human phenotypes. development: the path to safer and more effective
Molecular Pathology. Genet. Med. 17, 405–424 (2015). Nat. Rev. Genet. 18, 441–451 (2017). drugs. Nat. Rev. Genet. 5, 645–656 (2004).
177. Rehm, H. L. et al. ClinGen – the clinical genome 205. 1000 Genomes Project Consortium. A global 232. Roses, A. D. Pharmacogenetics in drug discovery
resource. N. Engl. J. Med. 372, 2235–2242 (2015). reference for human genetic variation. Nature 526, and development: a translational perspective.
178. Landrum, M. J. et al. ClinVar: improving access to 68–74 (2015). Nat. Rev. Drug Discov. 7, 807–817 (2008).
variant interpretations and supporting evidence. 206. Turro, E. et al. Whole-genome sequencing of patients 233. Nelson, M. R. et al. The genetics of drug efficacy:
Nucleic Acids Res. 46, D1062–D1067 (2018). with rare diseases in a national health system. Nature opportunities and challenges. Nat. Rev. Genet. 17,
179. Buniello, A. et al. The NHGRI-EBI GWAS catalog 583, 96–102 (2020). 197–206 (2016).
of published genome-wide association studies, 207. Chatr-Aryamontri, A. et al. The BioGRID interaction 234. Wei, C. Y., Lee, M. T. & Chen, Y. T. Pharmacogenomics
targeted arrays and summary statistics 2019. database: 2017 update. Nucleic Acids Res. 45, of adverse drug reactions: implementing personalized
Nucleic Acids Res. 47, D1005–D1012 (2019). D369–D379 (2017). medicine. Hum. Mol. Genet. 21, R58–R65 (2012).
180. Machiela, M. J. & Chanock, S. J. LDassoc: an online 208. Firth, H. V. et al. DECIPHER: database of chromosomal 235. Alfirevic, A. & Pirmohamed, M. Adverse drug
tool for interactively exploring genome-wide association imbalance and phenotype in humans using ensembl reactions and pharmacogenomics: recent advances.
study results and prioritizing variants for functional resources. Am. J. Hum. Genet. 84, 524–533 (2009). Per. Med. 5, 11–23 (2008).
investigation. Bioinformatics 34, 887–889 (2018). 209. Pinero, J. et al. The DisGeNET knowledge platform for 236. Collins, S. L., Carr, D. F. & Pirmohamed, M. Advances
181. Boyle, A. P. et al. Annotation of functional variation in disease genomics: 2019 update. Nucleic Acids Res. in the pharmacogenomics of adverse drug reactions.
personal genomes using RegulomeDB. Genome Res. 48, D845–D855 (2020). Drug. Saf. 39, 15–27 (2016).
22, 1790–1797 (2012). 210. Finer, S. et al. Cohort profile: East London Genes 237. Cook, J. C., Wu, H., Aleo, M. D. & Adkins, K. Principles
182. Ward, L. D. & Kellis, M. HaploReg: a resource for & Health (ELGH), a community-based population of precision medicine and its application in toxicology.
exploring chromatin states, conservation, and genomics and health study in British Bangladeshi and J. Toxicol. Sci. 43, 565–577 (2018).
regulatory motif alterations within sets of genetically British Pakistani people. Int. J. Epidemiol. 49, 20–21i 238. Cacabelos, R., Cacabelos, N. & Carril, J. C. The role
linked variants. Nucleic Acids Res. 40, D930–D934 (2020). of pharmacogenomics in adverse drug reactions.
(2012). 211. Locke, A. E. et al. Exome sequencing of Finnish Expert. Rev. Clin. Pharmacol. 12, 407–442 (2019).
183. GTEx Consortium. The GTEx Consortium atlas of isolates enhances rare-variant association power. 239. Lesko, L. J. & Woodcock, J. Translation of
genetic regulatory effects across human tissues. Nature 572, 323–328 (2019). pharmacogenomics and pharmacogenetics:
Science 369, 1318–1330 (2020). 212. Canela-Xandri, O., Rawlik, K. & Tenesa, A. An atlas a regulatory perspective. Nat. Rev. Drug Discov. 3,
184. Võsa, U. et al. Unraveling the polygenic architecture of genetic associations in UK Biobank. Nat. Genet. 50, 763–769 (2004).
of complex traits using blood eQTL metaanalysis. 1593–1599 (2018). 240. Maliepaard, M. et al. Pharmacogenetics in the evaluation
bioRxiv https://doi.org/10.1101/447367 (2018). 213. Consortium, G. T. et al. Genetic effects on gene of new drugs: a multiregional regulatory perspective.
185. Zhu, Z. et al. Integration of summary data from GWAS expression across human tissues. Nature 550, Nat. Rev. Drug Discov. 12, 103–115 (2013).
and eQTL studies predicts complex trait gene targets. 204–213 (2017). 241. Ehmann, F. et al. Pharmacogenomic information in
Nat. Genet. 48, 481–487 (2016). 214. MacArthur, J. et al. The new NHGRI-EBI Catalog of drug labels: European Medicines Agency perspective.
186. Staley, J. R. et al. PhenoScanner: a database of human published genome-wide association studies (GWAS Pharmacogenomics J. 15, 201–210 (2015).
genotype-phenotype associations. Bioinformatics 32, Catalog). Nucleic Acids Res. 45, D896–D901 (2017). 242. Relling, M. V. & Evans, W. E. Pharmacogenomics
3207–3209 (2016). 215. Stenson, P. D. et al. The human gene mutation in the clinic. Nature 526, 343–350 (2015).
187. Watanabe, K. et al. A global overview of pleiotropy database: towards a comprehensive repository of 243. Cecchin, E., Roncato, R., Guchelaar, H. J., Toffoli, G.
and genetic architecture in complex traits. Nat. Genet. inherited mutation data for medical research, genetic & Ubiquitous Pharmacogenomics, C. Ubiquitous
51, 1339–1348 (2019). diagnosis and next-generation sequencing studies. pharmacogenomics (U-PGx): the time for
188. Elliott, L. T. et al. Genome-wide association studies Hum. Genet. 136, 665–677 (2017). implementation is now. An Horizon2020 program
of brain imaging phenotypes in UK Biobank. Nature 216. Traynelis, J. et al. Optimizing genomic medicine in to drive pharmacogenomics into clinical practice.
562, 210–216 (2018). epilepsy through a gene-customized approach to Curr. Pharm. Biotechnol. 18, 204–209 (2017).
189. Denny, J. C. et al. Systematic comparison of missense variant interpretation. Genome Res. 27, 244. van der Wouden, C. H. et al. Development of the
phenome-wide association study of electronic medical 1715–1729 (2017). PGx-passport: a panel of actionable germline genetic
record data and genome-wide association study data. 217. Amberger, J. S., Bocchini, C. A., Schiettecatte, F., variants for pre-emptive pharmacogenetic testing.
Nat. Biotechnol. 31, 1102–1110 (2013). Scott, A. F. & Hamosh, A. OMIM.org: online Clin. Pharmacol. Ther. 106, 866–873 (2019).
190. Leslie, R., O’Donnell, C. J. & Johnson, A. D. GRASP: Mendelian inheritance in man (OMIM(R)), an online 245. Yang, T. et al. Genotype-guided dosing versus
analysis of genotype-phenotype results from 1390 catalog of human genes and genetic disorders. conventional dosing of warfarin: a meta-analysis of 15
genome-wide association studies and corresponding Nucleic Acids Res. 43, D789–D798 (2015). randomized controlled trials. J. Clin. Pharm. Ther. 44,
open access database. Bioinformatics 30, i185–i194 218. Koscielny, G. et al. Open Targets: a platform for 197–208 (2019).
(2014). therapeutic target identification and validation.
191. Cerami, E. et al. The cBio cancer genomics portal: an Nucleic Acids Res. 45, D985–D994 (2017). Author contributions
open platform for exploring multidimensional cancer 219. Gussow, A. B. et al. Orion: detecting regions of the The authors contributed equally to all aspects of the article.
genomics data. Cancer Discov. 2, 401–404 (2012). human non-coding genome that are intolerant to
192. Gonzalez-Perez, A. et al. IntOGen-mutations identifies variation using population genetics. PLoS ONE 12, Competing interests
cancer drivers across tumor types. Nat. Methods 10, e0181604 (2017). A.M.D., L.D.W., M.F. and J.Y. are former employees of
1081–1082 (2013). 220. Kamat, M. A. et al. PhenoScanner V2: an expanded Amgen. K.J.C. and P.N. are employees of AstraZeneca.
193. Taliun, D. et al. Sequencing of 53,831 diverse tool for searching human genotype-phenotype A.D.R.E. is a former employee of Novartis Institutes for
genomes from the NHLBI TOPMed Program. Nature associations. Bioinformatics 35, 4851–4853 (2019). BioMedical Research and is currently employed by GentiBio,
590, 290–299 (2021). 221. Gao, J. et al. Integrative analysis of complex cancer Inc. J.M. is an employee of Novartis Institutes for BioMedical
194. Zhou, W. et al. Efficiently controlling for case-control genomics and clinical profiles using the cBioPortal. Research. D.D. is an employee of Takeda Development Center
imbalance and sample relatedness in large-scale genetic Sci. Signal. 6, pl1 (2013). America. D.A.K. is an employee of GlaxoSmithKline. M.R.N. is
association studies. Nat. Genet. 50, 1335–1341 (2018). 222. The UniProt Consortium. UniProt: the universal protein an employee of Deerfield Management Company, L.P. F.D.S.
195. Wu, M. C. et al. Rare-variant association testing for knowledgebase. Nucleic Acids Res. 46, 2699 (2018). is a former employee and currently a part-time contractor of
sequencing data with the sequence kernel association 223. Scott, S. A. et al. Clinical pharmacogenetics Merck Sharp & Dohme Corp., a subsidiary of Merck & Co.,
test. Am. J. Hum. Genet. 89, 82–93 (2011). implementation consortium guidelines for CYP2C19 Inc., Kenliworth, NJ, USA. A.M.D. and L.D.W. are employees
196. Zhou, W. et al. Scalable generalized linear mixed model genotype and clopidogrel therapy: 2013 update. and stockholders of Alnylam Pharmaceuticals. J.Y. is an
for region-based association tests in large biobanks Clin. Pharmacol. Ther. 94, 317–323 (2013). employee of Pfizer.
and cohorts. Nat. Genet. 52, 634–639 (2020). 224. Weinshilboum, R. M. & Sladek, S. L. Mercaptopurine
197. Povysil, G. et al. Rare-variant collapsing analyses pharmacogenetics: monogenic inheritance of Peer review information
for complex traits: guidelines and applications. erythrocyte thiopurine methyltransferase activity. Nature Reviews Drug Discovery thanks Kathleen Meyer,
Nat. Rev. Genet. 20, 747–759 (2019). Am. J. Hum. Genet. 32, 651–662 (1980). Munir Pirmohamed and the other, anonymous, reviewer for
198. Martin, A. R. et al. Clinical use of current polygenic 225. CPIC® Guideline for Thiopurines and TPMT and their contribution to the peer review of this work.
risk scores may exacerbate health disparities. NUDT15 (Clinical Pharmacogenetics Implementation
Nat. Genet. 51, 584–591 (2019). Consortium, 2018); https://cpicpgx.org/guidelines/ Publisher’s note
199. Fang, H. et al. A genetics-led approach defines the guideline-for-thiopurines-and-tpmt/ Springer Nature remains neutral with regard to jurisdictional
drug target landscape of 30 immune-related traits. 226. Chung, W. H. et al. Medical genetics: a marker for claims in published maps and institutional affiliations.
Nat. Genet. 51, 1082–1091 (2019). Stevens–Johnson syndrome. Nature 428, 486 (2004).
200. Newberry, R. W., Leong, J. T., Chow, E. D., 227. Ferrell, P. B. Jr & McLeod, H. L. Carbamazepine,
Springer Nature or its licensor holds exclusive rights to this
Kampmann, M. & DeGrado, W. F. Deep mutational HLA-B*1502 and risk of Stevens–Johnson
article under a publishing agreement with the author(s) or
scanning reveals the structural basis for alpha-synuclein syndrome and toxic epidermal necrolysis: US
other rightsholder(s); author self-archiving of the accepted
activity. Nat. Chem. Biol. 16, 653–659 (2020). FDA recommendations. Pharmacogenomics 9,
manuscript version of this article is solely governed by the
201. Maurano, M. T. et al. Systematic localization of 1543–1546 (2008).
terms of such publishing agreement and applicable law.
common disease-associated variation in regulatory 228. Chen, P. et al. Carbamazepine-induced toxic effects
DNA. Science 337, 1190–1195 (2012). and HLA-B*1502 screening in Taiwan. N. Engl. J. Med.
202. Gupta, R. M. et al. A genetic variant associated 364, 1126–1133 (2011).
Related links
with five vascular diseases is a distal regulator 229. McCormack, M. et al. HLA-A*3101 and
Genes and Health: http://www.genesandhealth.org/research
of endothelin-1 gene expression. Cell 170, carbamazepine-induced hypersensitivity reactions in
Pakistan Genomic Resource: https://www.cncdpk.com/page/
522–533.e515 (2017). Europeans. N. Engl. J. Med. 364, 1134–1143 (2011).
Pakistan-Genomic-Resource-(PGR)
203. Flanagan, J. M. Epigenome-wide association studies 230. Lindpaintner, K. The impact of pharmacogenetics
(EWAS): past, present, and future. Methods Mol. Biol. and pharmacogenomics on drug discovery. Nat. Rev.
1238, 51–63 (2015). Drug Discov. 1, 463–469 (2002). © Springer Nature Limited 2022

162 | February 2023 | volume 22 www.nature.com/nrd

0123456789();:

You might also like