In the past year, a handful of blockbuster drugs that were expected to generate billions in sales, have turned out to be commercial disappointments due to a vicious debate about drug pricing.

With increasing pressure from a powerful and pugilistic class of payers and middlemen, President Donald Trump has vowed to bring down drug prices. Even for Senator Joe Manchin, whose daughter, Heather Bresch is conflictingly, Mylan CEO, has urged the Congress to address the matter.

"A few years ago, just winning approval meant getting wide access" to patients, said Simos Simeonidis, a biotech analyst at RBC Capital Markets. "That's definitely not the case anymore."

Pharma industry facing a pinch due to drug pricing

On 28 March, the FDA approved a treatment for severe eczema, dupilumab. Wall Street believes that it will rake in more than USD5 billion a year but with the drug priced at USD37,000 a year, it seems unlikely as many patient advocacy groups will demand rebates and reject prescriptions, forcing the manufacturers to recalibrate its expectations of profit.

This leaves the drug makers in a pinch as the expectations of investors are not changing accordingly. Investors demand a return on the hundreds of millions of dollars that drug companies spent to develop and market their products.

Therefore charging too little could put them under fire from shareholders but pricing a drug too high can risk lower rates of prescriptions or refusal from insurers to pay for it. Recurring offenders might even find themselves in front of Congress, accused of price fixing.

The cost of developing new drugs: Time, money and failure

However, developing new drugs is neither cheap nor easy. Many result in successful laboratory tests but in reality, fewer than 12% of all drugs entering clinical trials end up in pharmacies. It also costs approximately USD2.6 billion to market a drug.

The process is mostly trial-by-error, and there are so many possibilities that pharmaceutical researchers use pipetting robots to test a few thousand variants all at once. The best molecules go onto the next stage - animal models or cell cultures - before only a handful go on to bigger animal and human clinical trials - which could still fail.

"Drug development can fail for many reasons," says genetic epidemiologist, Aroon Hingorani. "However, a major reason is the failure to select the correct target for the disease of interest."

All these cost a substantial amount of money and time that pharmaceutical companies have to take into account when pricing drugs. Researchers estimate that about 15% to 20% of the cost of a new drug is due to the discovery phase. To put into perspective, that represents up to a few hundred million dollars and three to six years of work.

AI a solution to identify potential drug molecules

In a bid to reduce cost and failure, researchers have offered computers and artificial intelligence (AI) to narrow down the list of potential drug molecules - saving time and money on the downstream tests.

Many models of algorithms exist to identify genes that code for drug-binding proteins, including a recent model by Hingorani's group.

The group's predictive model added new layers of complexity that combined genetic information with protein structure data and known interactions, resulting in 4,500 potential drug targets - doubling prior estimates for how much of the human genome is considered "druggable".

When two clinicians filtered the data, they found 144 existing drugs that could be repurposed for other diseases. These computational approaches also promise to cut the discovery phase into a few months and save tens of thousands of dollars.

And when it comes to drug discovery, time is money.

Incorporation of AI into the drug pipeline requires evidence

But there is still no drug on the market currently that started out with an AI system. Critics are also skeptical of these purely computational approaches - many seeking solid evidence.

Hingorani's company, twoXAR is working to collect that evidence. In collaboration with the Asian Liver Centre at Stanford, Hingorani's software identified 10 possible treatments, which Samuel So, the director of the liver centre tested out.

From the list, one of the treatments is undergoing human trials and another, which is the only existing FDA-approved treatment for a cancer took five years to develop - which twoXAR and Stanford managed to develop in four months.

This solution might just be what the pharmaceutical industry is looking for to curb their high failure rate. After all, small gains on time could be worth billions of dollars, and save human lives. MIMS

Read more:
Patient-led research may be the future of medical studies
Pharmaceutical companies refuse calls for transparency in price hikes
New drug is effective in lowering LDL levels, but raises questions about drug pricing