The Economist 14Mar2020

12 Technology Quarterly |Personalised medicine The EconomistMarch 14th 2020

2

1

for an sirnacalled inclisiran that interferes with

the expression of the gene PCSK9—thus stopping

the pesky protein from being made in the first

place. Inclisiran needs to be injected only twice a

year, rather than once a month, as antibodies do.

New biological insights, new ways of analys-

ing patients and their disease and new forms of

drug are thus opening up a wide range of thera-

peutic possibilities. Unfortunately, that does not equate to a range

of new profitable opportunities.

Thanks in part to ever better diagnosis, there are now 7,000

conditions recognised as “rare diseases” in America, meaning that

the number of potential patients is less than 200,000. More than

90% of these diseases have no approved treatment. These are the

diseases that personalised, precision medicine most often goes

after. Nearly 60% of the personalised medicines approved by the

fda in 2018 were for rare diseases.

That might be fine, were the number of diseases stable. But pre-

cision in diagnosis is increasingly turning what used to be single

diseases into sets of similar-looking ones brought about by dis-

tinctly different mechanisms, and thus needing different treat-

ment. And new diseases are still being discovered. Medical pro-

gress could, in short, produce more new diseases than new drugs,

increasing unmet need.

Some of it will, eventually, be met. For one thing, there are gov-

ernment incentives in America and Europe for the development of

drugs for rare diseases. And, especially in America, drugs for rare

diseases have long been able to command premium prices. Were

this not the case, Novartis would not have paid $8.7bn last year to

buy Avexis, a small biotech firm, thereby acquiring Zolgensma, a

gene therapy for spinal muscular atrophy (sma). Most people with

smalack a working copy of a gene, SMN1, which the nerve cells that

control the body’s muscles need to survive. Zolgensma uses an

empty virus-like particle that recognises nerve cells to deliver

working copies of the gene to where it is needed. Priced at $2.1m

per patient, it is the most expensive drug ever brought to market.

That dubious accolade might not last long. BioMarin, another bio-

tech firm, is considering charging as much as $3m for a forthcom-

ing gene therapy for haemophilia.

Drug firms say such treatments are economically worthwhile

over the lifetime of the patient. Four-fifths of children with the

worst form of smadie before they are four. If, as is hoped, Zol-

gensma is a lasting cure, then its high cost should be set against a

half-century or more of life. About 200 patients had been treated in

America by the end of 2019.

But if some treatments for rare diseases may turn a profit, not

all will. There are some 6,000 children with smain America. There

are fewer than ten with Jansen’s disease. When

Dr Nizar asked companies to help develop a

treatment for it, she says she was told “your dis-

ease is not impactful”. She wrote down the neg-

ative responses to motivate herself: “Every day I

need to remind myself that this is bullshit”.

A world in which markets shrink, drug de-

velopment gets costlier and new unmet needs

are ceaselessly discovered is a long way from the utopian future

envisaged by the governments and charities that paid for the se-

quencing of all those genomes and the establishment of the

world’s biobanks. As Peter Bach, director of the Centre for Health

Policy and Outcomes, an academic centre in New York, puts it with

a degree of understatement: if the world needs to spend as much to

develop a drug for 2,000 people as it used to spend developing one

for 100,000, the population-level returns from medical research

are sharply diminishing.

Moore is less

And it is not as if the costs of drug development have been con-

stant. They have gone up. What Jack Scannell, a consultant and for-

mer pharmaceutical analyst at ubs, a bank, has dubbed Eroom’s

law—Eroom being Moore, backwards—shows the number of drugs

developed for a given amount of r&dspending has fallen inexora-

bly, even as the amount of biological research skyrocketed. Each

generation assumes that advances in science will make drugs easi-

er to discover; each generation duly advances science; each gener-

ation learns it was wrong.

For evidence, look at the way the arrival of genomics in the

1990s lowered productivity in drug discovery. A paper inNature Re-

views Drug Discoveryby Sarah Duggers from Columbia University

and colleagues argues that it brought a wealth of new leads that

were difficult to prioritise. Spending rose to accommodate this

boom; attrition rates for drugs in development subsequently rose

because the candidates were not, in general, all that good.

Today, enthused by their big-science experience with the ge-

nome and enabled by new tools, biomedical researchers are work-

ing on exhaustive studies of all sorts of other ‘omes, including pro-

teomes—all the proteins in a cell or body; microbiomes—the

non-pathogenic bacteria living in the mouth, gut, skin and such;

metabolomes—snapshots of all the small molecules being built up

and broken down in the body; and connectomes, which list all the

links in a nervous system. The patterns they find will doubtless

produce new discoveries. But they will not necessarily, in the short

term, produce the sort of clear mechanistic understanding which

helps create great new drugs. As Dr Scannell puts it: “We have

treated the diseases with good experimental models. What’s left

are diseases where experiments don’t replicate people.” Data alone

canot solve the problem.

Daphne Koller, boss of Insitro, a biotech company based in San

Francisco, shares Dr Scannell’s scepticism about the way drug dis-

covery has been done. A lot of candidate drugs fail, she says, be-

cause they aim for targets that are not actually relevant to the biol-

ogy of the condition involved. Instead researchers make decisions

based on accepted rules of thumb, gut instincts or a “ridiculous

mouse model” that has nothing to do with what is actually going

on in the relevant human disease—even if it makes a mouse look

poorly in a similar sort of way.

But she also thinks that is changing. Among the things preci-

sion biology has improved over the past five to 10 years have been

the scientists’ own tools. Gene-editing technologies allow genes to

be changed in various ways, including letter by letter; single-cell

analysis allows the results to be looked at as they unfold. These

edited cells may be much more predictive of the effects of drugs

than previous surrogates. Organoids—self-organised, three-di-

mensional tissue cultures grown from human stem cells—offer

simplified but replicable versions of the brain, pancreas, lung and

Eroom’s law

United States, number of new molecules approved* per $1bn global R&D spending

Inflation adjusted

100

10

1

0.1

Logscale

1950 60 70 80 90 2000 10 17

Source: Scannell et al. (2012), with additional

post-2012 data by Scannell et al. *By US Food and Drug Administration (FDA)

FDA clears backlog following PDUFA

regulations plus small bolus of HIV drugs

First wave of

biotechnology-

derived therapies Increase in

‘orphans’ plus

‘targeted’

cancer drugs

FDA tightens regulations

post-thalidomide

Zolgensma is the most

expensive drug ever

brought to market.