A reality check for personalised medicine: just a

few errors in translation?

Professor Donna DickensonEmeritus Professor of Medical

EthicsUniversity of London

This is not a rhetorical question…

• Is genetic personalised (stratified) medicine a genuine paradigm shift from a ‘one size fits all’ model towards a much more powerful and individualised approach?

• Or are its medical and ethical limitations much more than a few errors in translation?

Benefits of pharmacogenetics

• With genetically individuated treatment, for example in cancer care, patients might be spared ‘megadose’ regime that can produce iatrogenic harm

• Patients who are genetically programmed to respond more quickly to chemotherapy could get lighter doses

• Lifespan might be extended as well

A recent example• Two studies on gliomas (Eckel-Passow et

al., and Cancer Genome Atlas Research Network, NEJM, June 2015) analysed 1380 tumours, breaking patients down into five groups, combinations classified according to three genetic mutations in tumours

• Current histological approach less effective in predicting response to treatment and overall prognosis

Another recent example

• Robinson et al. (Cell, May 2015) developed genetic map of prostate cancer in 150 patients, revealing it as not one disease but several, responsive to different treatments

• Analysed genetic basis for resistance to standard treatment, hormone therapy, as result of specific DNA mutations

(In?)cautious optimism

• At launch in 2010 of NHS initiative to personalise treatment of up to 6000 patients annually by adjusting their therapy to genetic mutations of tumours, Dr Harpal Jumar, chief executive of Cancer Research UK, said: ‘It’s patently obvious that this is going to be the way of the future.’

Example of vemurafenib • 2012: NEJM article reports promising

findings for pharmacogenetic drug used against aggressive melanoma

• Targets genetic mutation found in tumours of about half of patients; of those, only about one-half responded

• The one-quarter who did benefit, however, gained an extra 7 months lifespan compared to conventional treatment

But things aren’t always so clear• At least the figure of one in four who

benefit from this drug is fairly reliable• But for other tests, predictive value is

no better than tossing a coin (Lee and McLeod 2011)

• Paradox of pharmacogenetics: the more specific the drug is to particular genetic subgroups, the harder it becomes to find enough patients to yield statistically significant results

An example of serious limitations• April 2015 study in Science Translational

Medicine of 815 cancer patients: sequencing tumour genomes for clues to genetic changes can misdirect individualised treatment in nearly half of patients, unless first compared to genetic readout of non-cancerous tissue

• Patients might get wrong targeted therapies• Two-thirds of genetic changes identified with

tumour-only sequencing were false positives

Two-thirds? Not ‘just a few errors’• False positive rate might be lessened by

sequencing non-cancerous tissue, but obviously that increases costs

• Undermines argument that personalised medicine is cost-effective because only those patients who would benefit from a therapy will receive it

• Particularly problematic because many pharmacogenetic medicines are expensive

Ethical issues must be addressed

• Often assumed that there are no ethical issues in personalised medicine: that it is unequivocally what would be called A Good Thing in 1066 and All That

• But in my 2013 book Me Medicine vs. We Medicine, I call that assumption into doubt

The wrong treatment for the wrong patient at the wrong time?

• Ethos of pharmacogenetics is ‘the right treatment for the right patient at the right time’

• A high rate of false positives with tumour-only sequencing could result in wrong treatment for ‘right’ patient

• In other cases patients judged ‘wrong’ might be denied treatment altogether

Example of kidney transplants

• Genetic markers can predict probability of immune system rejection (Dion-Labrie et al. 2010)

• Level of expression in haematopoietic cells can forecast incidence of graft versus host syndrome

• But is it ethical to select potential recipients on genetic basis?

Again, not a rhetorical question

• Pro: it is both more efficient and more equitable to use advances in genetic sequencing to select recipients

• Con: apart of the problem of false negatives, what cut-off is equitable if the answer is not black and white?

• Should a patient get a drug even if there is only a 20 percent chance she will respond? 10 percent? One percent?

Resource allocation: ‘tragic choices’• Selecting patients on basis of

genetic likelihood of immune system rejection fits model of allocation by prognosis: resource should go to those with the best chance of recovery

• This criterion can be controversial: it was used to allocate scarce penicillin to STD sufferers during World War Two, not to men wounded in battle

Alternative model of allocation

• Resource should be allocated by diagnosis, to those with most pressing medical need

• We often favour this second criterion: for example, we think that dying patients should receive more time and care than they ‘deserve’ on basis of poor prognosis

Application to kidney transplant case

• Where dialysis is too burdensome or is failing, genetically ‘wrong’ patients needing a kidney transplant will be denied their only hope…

• Even if genetic matching improves overall efficiency…

• And avoids ‘wasting’ scarce resource of kidney: a genuine ethical dilemma

Colorectal cancer example• Testing for KRAS gene associated with

colorectal cancer can partially predict that a ‘wrong’ patient is unlikely to respond to particular therapy, especially if cancer has metastasised (Lee and McLeod 2011)

• Yet sensitivity of KRAS testing is imperfect, so that many patients identified as ‘right’ won’t respond either

Cards on the table…

• As a medical ethicist, I would have concerns about dividing people into treatment/non-treatment camps based only on a genetic test, even if there were no false positives or negatives

• Patients classified as ‘wrong’ might still want to take their chances on treatment if there is no alternative

A reality check in times of austerity• In an era of cost-cutting, it is possible to

envisage a scenario in which genetic testing is used as a rationale for denying some patients a therapy

• One systematic review (You 2011) argued that warfarin management should depend almost entirely on cost-effectiveness

• Prediction that pharmacogenetics will be used mainly to rein in patient demand (Sjogren 2010)

Escalating costs in pharmacogenetics

• Because pharmacogenetic drugs typically treat much smaller populations than old-style blockbusters, drug firms understandably argue that they should cost more

• Examples: Kalydeco (ivacaftor) for cystic fibrosis ($294,000 p.a.)

• Only effective in the four per cent of c.f. patients who have G551D mutation

‘We Medicine’

• In 19th and early 20th centuries, rich and poor alike were vulnerable to contagious diseases such as cholera, typhus and influenza

• Public health measures such as sanitation and vaccination served both social justice and medical progress

‘Me Medicine’• But when individual conditions such

as cancer and cardiovascular disease replaced infectious diseases as main cause of mortality in developed world, illness was individualised, as was medicine’s response

• Personalised medicine techniques, such as pharmacogenetics, take that logic one step further

Will personalised medicine produce a more or less fair society?

• That doesn’t mean personalised medicine is ‘bad’: just that we must locate it in a wider social context

• We can best advance the cause of personalised medicine by being properly sceptical about what it can and can’t do

• …and by insisting on asking the necessary ethical questions