AS DRUGS companies increasingly focus on how different groups of patients
respond to a particular medication, a study has shown for the first time that
grouping people by race is not the best way to do this.
鈥淚t鈥檚 a great step forward to show that race labels are not needed as a
guide,鈥� says Rochelle Long, programme director of the Pharmacogenetics Research
Network at the US National Institutes of Health.
Many modern trials look at whether particular ethnic groups might be more or
less sensitive to the drug being tested. Companies can now try to get approval
for a drug even if it only works for a particular group rather than the
population in general. In March this year, the Food and Drug Administration said
it would approve a heart drug called BiDil specifically for Afro-Americans if
the results of a late-stage trial are positive.
Advertisement
But the latest study, led by David Goldstein of University College London,
shows that looking at people鈥檚 DNA is a better way to identify groups that
respond differently to treatments than race. Goldstein鈥檚 team looked at 354
individuals from eight populations from around the world. When they compared
certain chromosomal regions known to be good at reflecting people鈥檚 genetic
ancestry, they found the individuals fell into four clusters.
Goldstein鈥檚 team then looked at variations in six drug-metabolising enzymes
known to affect the way people respond to drugs. Goldstein used either race
(black, Caucasian or Asian) or the four-cluster classification to see which was
better at sorting out people according to their enzyme variations. Sequence
similarity was a much better predictor, while clustering people by race could
have masked important differences in drug response. The study will appear in a
future issue of Nature Genetics.
The researchers also confirmed previous findings that racial labels do not
always correlate with genetic relatedness. For example, 62 per cent of
Ethiopians fell into the same category as Ashkenazi Jews, Armenians and
Norwegians. Clustering them in a 鈥渂lack鈥� group with Afro-Caribbeans would not
reflect their genetic make-up.
But drugs studies that used race as predictors were not ill-intentioned, Long
says. 鈥淧rior to having the right methods of analysis, people used what they
saw.鈥� And one obvious characteristic is race.
鈥淓veryone knew that skin pigment is a lousy predictor,鈥� says Howard McLeod at
the Washington University School of Medicine. This is the first time that anyone
has backed up that hunch with hard evidence.
But while the four-cluster method is an improvement and could be used in
clinical trials immediately, it鈥檚 not a perfect guide to the way people will
respond to drugs. Ultimately, says McLeod, pharmacogenetics should focus on
finding ways to speed up and lower the cost of technologies for sequencing each
person鈥檚 genome. That will be the best way to determine how someone will respond
to a drug, he says.