If there’s a dream of what a new drug is supposed to do, it might look something like Kalydeco. In 2012, the new light-blue pill from Vertex Pharmaceuticals rocked the world of cystic fibrosis, a fatal disease that affects 30,000 people in the United State. It’s best known for its attack on the lungs, slowly suffocating its victims while attacking other organs—but when patients got the drug in its experimental phase, some started reporting such enormous improvement in their breathing and energy they were able to take up running, even marathoning.
Kalydeco is emblematic of the promise of new approaches in drug development. Built on a new understanding of how a particular defect in a gene can disrupt the workings of the body, the drug zeroes in on critical proteins inside cells to keep them functioning. “The drug was so good it broke the blind,” says Bernard Munos, a senior fellow at FasterCures, a think tank based at the Milken Institute—meaning the positive trial results were so clear that patients and doctors could easily tell who was receiving the drug and who got a placebo.
At the same time, Kalydeco serves as a cautionary tale. The gene behind cystic fibrosis was identified in the early 1990s, yet it took two decades to bring out a drug that could do anything about it. The cost of treatment with Kalydeco is startlingly high—about $300,000 a year per patient. And most sobering of all, Kalydeco helps about only 5 percent of cystic fibrosis patients. “The results for the patients it helps are absolutely spectacular, but those patients are only a sliver of the population with the disease,” says Munos. If you haven’t heard of Kalydeco—and you probably haven’t—that’s because this drug, which costs the U.S. health care system nearly half a billion dollars per year, currently helps fewer than 2,000 American patients.
The drug-development system that produced Kalydeco is one of proudest achievements of American medicine, and one of our biggest investments as a society. When American leaders talk about “innovation” in health care, they’re largely talking about the development of new pharmaceuticals. Prescription drugs already account for one out of every 10 dollars spent on health care in the U.S., amounting to more than $300 billion a year, and climbing. Drug development enjoys broad political support: The 21st Century Cures Act, one of the last big bipartisan bills to pass Congress, is pumping another $6.3 billion into new research funding, largely to hunt for these badly needed drugs.
But as America looks squarely at the biggest health challenges of the future, there’s reason to worry that the system we’ve built may not be adequate to what’s in front of us. When it comes to the two diseases likely to become our biggest killers— Alzheimer’s and diabetes—death rates are relentlessly ticking up, with few solutions on the horizon. Even more alarming, to many experts who look at America’s long-term health future, is that incentives appear to be leading toward more drugs like Kalydeco, pricey treatments for small groups of patients, rather than toward drugs to tackle major new chronic killers. And if they ever do succeed against these diseases, the costs could be high enough to threaten the solvency of the whole U.S. health care system.
More than 5 million Americans suffer from Alzheimer’s; 1 out of 3 of us will eventually end up with Alzheimer’s or other dementia. Diabetes is a close second on the increasingly worrisome list. The numbers are expected to grow steeply as the population ages—exacting costs not only on patients, but on a health care system and families carrying the burden of long-term chronic care. As you’d expect, researchers, government funders and pharma companies have tried coming up with new drugs to respond to this growing threat, but aren’t finding answers. In a 2017 report from industry-research service EvaluatePharma, of the 20 most promising future drugs approaching the market, none are aimed at Alzheimer’s. The only one that addresses diabetes is a longer-lasting version of an existing drug.
It’s not for lack of trying. Dan Skovronsky, who is taking over as head of research at Eli Lilly, points out that Lilly has invested billions in Alzheimer’s alone with no successful drug to show for it. “Alzheimer’s and diabetes are the most significant and challenging diseases we know,” says Skovronsky. “It could take anywhere from five to 20 years for a breakthrough in either one.”
As experts see it, the existing machinery for developing a drug and bringing it through approval has proven a poor fit for complex, hard-hitting chronic diseases that strike tens of millions of people in ways that play out over decades. What that machinery does well is use cutting-edge biology to attack diseases with narrowly defined mechanisms of damage—diseases that, as with cystic fibrosis, tend to affect a relatively small number of people. The needed insights into Alzheimer’s and diabetes haven’t yet emerged, giving pharma companies little to work with. More than 350 drug candidates have been thrown at Alzheimer’s alone, and none of them has worked well in human trials.
Scratch a little deeper, and the puzzles of Alzheimer’s and diabetes begin to look as much like a policy challenge as a scientific one. With today’s incentives, the pharma industry has built an attractive business model on narrowly targeted drugs like Kalydeco. Bigger, murkier and more complex diseases like diabetes and Alzheimer’s require a much larger investment with little confidence that it will result in correspondingly lucrative products—or even any products at all.
Thankfully for society, this problem of mismatched incentives may be a solvable one. Government funders, with cooperation from academia and the pharma industry, are now trying to inject money into basic research and forge new types of alliances that can give pharma companies better science to work with and to lower the industry’s risk in trying. It’s not yet clear the new efforts will be enough to kickstart a new generation of promising discoveries—or how soon even the fastest-moving new drug would even arrive. But they may at least point the way out of what is slowly but surely becoming a national health nightmare.
IN THE PAST, the drug industry has swung at some big societal targets and connected. Life expectancy for people with HIV, a death sentence 30 years ago, is now close to ordinary. Heart disease and cancer are still the top two killers, but the risks they pose are shrinking. Medicine has cut the death rate from heart disease by about two-thirds since the 1970s, thanks in part to drugs that, among other benefits, lower cholesterol and blood pressure. Progress against cancer, too, has been impressive and ongoing, largely because of advances in chemotherapy increasingly driven by genomic insight. The ranks of cancer survivors have doubled over the past 25 years, and today two-thirds of those diagnosed with cancer can expect to live at least another five years.
For Alzheimer’s and diabetes, the trends go the other way. Current estimates indicate 44 percent of people between the ages of 75 and 84 in the United States have Alzheimer’s, and the disease costs the country $236 billion. The number of people with Alzheimer’s is projected to triple by 2050 to more than 15 million—at which point, if current trends hold, more than 100 million people will be unpaid caregivers for Alzheimer’s patients and annual costs will soar to more than $700 billion. The number of Americans with full-blown diabetes is projected to increase by half by 2030, to more than 55 million, by which time nearly 400,000 people will be dying yearly from the disease, and annual costs will have risen to more than $600 billion.
There is no good way to estimate the costs of such devastation on suffering and quality of life, but it’s hard to see how the damage wouldn’t be overwhelming to society. The emergence of effective treatments could derail that grim journey, but today only 1.2 percent of current large clinical trials are testing a drug aimed at Alzheimer’s or other form of dementia. Fully a quarter of all trials, on the other hand, are testing potential cancer drugs. That’s because cancer drug development has a workable model: Relying on basic research to identify specific genes and proteins related to specific tumor types of a certain kind of cancer, coming up with drugs that interfere with the actions of those genes or proteins, and then testing them on patients who have that specific type of tumor—even though such patients might make up only a fraction of the population of patients who have that kind of cancer.
The cost of the final stage of human testing—typically the most expensive part of drug development—for this sort of focused cancer treatment might typically run about $20 million. A similar trial to assess the effectiveness of an Alzheimer’s drug on cognitive function, or of a diabetes drug on long-term health, can run to 10 times as much and more. That’s because it takes a much larger number of patients who stay in the trial over a much longer period of time to get a clear answer. In other words, the economics of drug development tends to work against the pharma industry continuing to throw money at the very diseases that hurt the most people. “It costs between $1.4 and $2.6 billion to bring a drug through clinical trials to FDA approval,” says Philip Smith, deputy director of the Division of Diabetes, Endocrinology and Metabolic Diseases at the National Institutes of Health. “Not only has the failure rate of candidates gone through the roof, but it’s happening in later-stage human testing”—that is, the largest and most expensive trials. That’s true for drugs for most diseases, but Smith notes the problem is much worse for the more complex diseases.
One ray of hope is the notion that the Kalydeco model—more narrowly targeted drugs for a small segment of people—could be applied to big diseases like Alzheimer’s. Drug companies are starting to think in terms of “stratifying” patient populations even for widespread diseases, based on identifying different versions of the diseases or how they progress in different patients. That way, drug candidates and trials can be focused on a more clearly defined problem with a potentially clearer target, leading to smaller clinical trials. This is the story behind much recent progress against cancer, which medicine no longer attacks as one disease, but as a vast array of tumor types and patient groups.
Can Alzheimer’s and diabetes be sliced in ways that make it easier to find promising drugs? The answer seems to be yes. “There are probably gene mutations for these diseases that would make a drug more likely to work on a smaller number of patients that have the mutation,” says Paula Bates, a researcher at the Institute for Molecular Diversity and Drug Design at the University of Louisville, who works closely with the pharma industry. “Some of the cancer immunotherapies didn’t seem to work on patients for a long time, until someone found a set of patients that one of the therapies did work on.”
But there’s a drawback to applying the patient-stratification approach to Alzheimer’s and diabetes. Where Kalydeco’s developers were able to aim at a single, well-understood gene—and cancer researchers can target very narrow cell processes in a subtype of tumor—Alzheimer’s appears to emerge from the crushingly complex interactions of entire networks of genes and environmental factors, few components of which have been clearly identified, and that probably vary from patient to patient. These factors suggest that any one drug that finally does prove helpful might well work on only an even tinier percentage of patients than Kalydeco. “We’d all like to come up with a drug that works on everyone,” says Matthias von Herrath, who heads diabetes research centers at pharmaceutical company Novo Nordisk and the La Jolla Institute. “That just may not be possible.”
Some are already talking about Alzheimer’s as a bundle of diseases rather than a single distinct problem. “Alzheimer’s is a mix of pathologies,” says Serge Gauthier, who directs Alzheimer’s research at the Centre for Studies in Aging at McGill University in Montreal. “We need to try treating it not with a single drug, but with different drugs that target the specific disease activities in the patient’s brain.” Trials that will test Alzheimer’s drug candidates on the 10 percent of patients who have a particular pair of genes that seem linked to a particular variation of the disease are already in the planning stage, Gauthier says. If such narrower patient-population tests produce encouraging results, he adds, drug companies are likely to take a fresh look at some of the previously tested candidates who failed in bigger groups of patients, and consider a new trial on more carefully filtered groups of individuals.
The same thinking is taking over strategies for diabetes drug development for which the focus has always been on the most readily observable component of the disease: controlling blood glucose levels. “We’ve treated diabetes as if it’s been a single disease of glucose levels,” says the NIH’s Smith. “Now it’s apparent to us that it’s a complex collection of potential disease pathways that all happen to share a final common pathway of glucose. Treating these different pathways with the same approach is illogical.”
Smith’s NIH group is now looking for evidence that some patients may have specific genetic risks for one of up to four different versions of diabetes. “That could help the rate of success for drug development go way up,” says Smith. And in that way, the group may be serving as a role model for how government and academic researchers can help better position pharma companies to follow the new stratified-patient track in the search for effective Alzheimer’s and diabetes drugs.
A WIDE RANGE of experts say the new effort to find drugs that provide better results to subsets of Alzheimer’s and diabetes patients is likely to produce results, even if the results are still years away. But facing the challenge of these two diseases with new drugs isn’t just about coming up with something that works. It’s equally about money. “A drug that isn’t accessible to patients because of costs is no better to those patients than a drug that hasn’t been discovered,” says Marc-André Gagnon, a pharmaceutical-policy researcher at Carleton University in Ontario, Canada.
As the pharma industry’s research model has shifted from finding drugs to treat 10 million people to drugs that treat 10,000, its pricing model has shifted as well—as the $300,000-a-year cost of Kalydeco attests. U.S. policy has exacerbated the cost problem, says Gagnon, by providing pharmaceutical companies with grants, tax breaks and extra patent protection for coming up with drugs that address disorders affecting relatively small numbers of patients, or so-called orphan drugs. Combined with the new tools that make it easier to target drugs at subsets of patients based on genetic differences, and an ability to charge enormous prices for the resulting drugs, the orphan-drug incentives have helped nudge pharma companies to embrace the model of developing narrowly aimed drugs.
The system has been able to sustain the high costs of drugs like Kalydeco because the patient numbers are so small. “But if drugs for big diseases arrive with those high price tags, the whole reimbursement system will become unsustainable,” says Gagnon.
“There isn’t enough money in the health care bank to give drugs at those prices to everyone who will need them,” says the Milken Institute’s Munos.
Even worse, there’s a risk pharmaceutical companies will bring out superexpensive, narrowly focused drugs that work only a bit better than conventional drugs. A recent study in the BMJ medical journal concluded about half of all new cancer drugs arrived with enormous price tags but delivered only marginal benefits over much less costly drugs. That means the health care system ends up paying tens or even hundreds of millions of dollars more for very small patient benefits. Most European countries have mechanisms to prevent paying for that sort of minuscule health return on investment. France, for example, ties the payment for an individual patient’s treatment to whether the treatment significantly helped. In the United Kingdom, which takes an even firmer approach to the problem, the government-run health care system refuses to approve reimbursement for drugs that can’t demonstrate good clinical returns for the price.
But the U.S. doesn’t have a national mechanism for controlling drug prices. Payers like insurance companies, employers and Medicare can refuse to reimburse for overly expensive drugs, of course. The problem is that American patients don’t take well to being told they can’t have the latest and greatest medication, seeing any cost other than whatever monthly insurance premiums they pay as someone else’s problem. Austin Frakt, a health care economist with Harvard’s and Boston University’s schools of public health and adviser to the Department of Veterans Affairs on drug prices, notes that when public panels are held to examine VA drug-payment policy, patient groups often show up to defend high drug prices. “They’re concerned that any downward influence on prices, like restrictions on reimbursement, will mean that drugs won’t be developed for their diseases,” he explains.
So far, it’s been easier for insurers and the government to keep saying “yes” to high-priced drugs that deliver limited new benefits than to brave the political fire of saying “no” to dying patients and their families. That dynamic may not survive. If an array of new Alzheimer’s and diabetes drugs that are collectively effective on millions of people turns out to come with sky-high prices, society might face a difficult choice: rationing treatment, going broke making it available to all, or controlling prices in a way that would make drug development unprofitable and thus likely shut it off.
There’s a way to avoid this bind: help lower development costs so the new drugs could be priced more reasonably. It may be our only way out of a brewing health care disaster, which is why policymakers, the NIH, academia and pharma companies are all eagerly embracing strategies to make it happen. The effort largely revolves around finding better biomarkers for the two diseases—that is, biological indicators in patients that can be easily measured, and that provide early indicators of whether an experimental drug is actually doing something that’s likely to help the patient. A good biomarker can drastically cut the need for massive, long-running, superexpensive trials, by giving researchers a fast, easily measured and reliable answer to the basic question, “Does this drug work?” It can also give basic researchers clues as to which research avenues are most likely to pay off, avoiding dead ends and speeding the early stages of development.
In Alzheimer’s, researchers have tried to judge the effectiveness of experimental drugs by assessing how the drug affects cognitive decline in patients—by giving patients memory tests, for example. The problem is that most experts now believe that by the time cognitive decline has advanced enough to be clearly detected through memory tests, any window for hitting the brain with a drug capable of staving off the disease has long since closed, possibly for two or more decades. Researchers can give people the drug earlier on in life, before symptoms show—but they might have to wait 20 or 30 years to find out if the drug worked, or even if a test subject had Alzheimer’s in the first place.
What researchers need is a biological signal that can indicate a patient is developing Alzheimer’s years before cognitive decline shows up, and that can quickly measure whether a drug is slowing the progress of the disease. Though drugs that attack the brain plaques and tangled proteins that characterize Alzheimer’s haven’t worked so far in patients with cognitive decline, notes Lilly’s Skovronsky, plaques and tangles may in fact be the right biomarkers if they can be assessed in pre-cognitive-decline patients. That’s been hard to do in real time because the full extent of damage until recently could be clearly seen only in brain samples taken after the patient’s death. Lilly has been among those developing chemical agents that would allow clinicians to better spot plaques and tangles in living patients. “Now we can start to see who has how much plaque, and who has it in specific parts of the brain, and who has plaques but doesn’t have tangles yet,” says Skovronsky. “That would give us tools to enroll the right patients earlier on in the course of the disease, and then see if a drug is working.”
Brain inflammation and signs of small strokes may prove useful biomarkers as well, at least for some sets of patients. McGill University’s Gauthier says future drug trials for Alzheimer’s could screen patients via brain imaging and spinal-fluid testing, as well as plaque and tangle imaging, so as to include only those whose brains exhibit signs of some particular combination of all these markers.
In diabetes, doctors and researchers have relied on blood glucose levels as a biomarker, but that’s proven unreliable as a way of telling if a drug will head off later organ damage, such as heart disease and blindness, or early death. “It’s clear we need drugs that don’t just lower glucose, but that reduce the complications from diabetes,” says Smith. “Now we’re also looking at biomarkers for heart disease, renal disease, and other complications, so we can see if a drug is really engaging the right targets.” Once the most useful biomarkers are identified, adds Smith, they can be enlisted to help researchers find specific genes that may be associated with a certain pattern of biomarker indications. Trials can then be focused on patients with those genes, a strategy that should work for both diseases.
With biomarkers providing faster, clearer indications of drug effectiveness on patients earlier along the disease pathway, drug companies should in theory be able to get faster, clearer results with fewer trials on smaller numbers of patients—perhaps hundreds instead of the usual thousands. That could mean there’s a good chance better drugs could be developed more quickly at lower cost.
THAT LEAVES THE DRUG development world with a big question: Who’s going to do the work needed to find better biomarkers? It’s a challenging question because biomarkers can tumble into the gap between the product-driven investments of pharma companies, and the basic biology questions that academic researchers tend to focus on. Closing that gap requires finding a way to better connect academia and pharma. “It requires a structural change in how we do drug development,” says Skovronsky.
The solution is multipronged. On the academic side, researchers need more interest and incentive in coming up with insights that are drug-development friendly. And on the pharma side, companies need to get past their fixation on only those ideas and approaches that they can lock away from competitors early on with patents, freeing them up for more collaboration with academia, government and each other.
At the University of Louisville, Bates runs the Research Evaluation and Commercialization Hub, one of several centers funded by the NIH designed to nudge academic researchers into closer collaborations with the pharma industry and other health care-related industries. “We provide a lot of guidance about how the industry, the market and the FDA work,” she explains. “And we provide funding for proof-of-concept studies, to encourage researchers to find out as quickly as possible if their work might lead to a viable product, and if so, how to get it more quickly into the development pathway.”
To promote collaboration even more broadly, NIH has established the “Accelerating Medicines Partnerships” for both Alzheimer’s and diabetes. The AMP programs fund consortia of industry, government and academic researchers tasked with coming up with lists of prime new candidates for biomarkers and drug targets. Richard Hodes, director of the National Institute on Aging, whose division houses the Alzheimer’s AMP program, says investigators in the consortium are looking at brain tissue, genomes, protein expression and metabolic pathways, and are applying high-powered bioinformatics tools to the data in order to identify the best candidates. Though it’s too soon to assess results, there’s excitement among researchers and industry that the effort will generate critically useful information. “It’s fairly unheard of to have that sort of collaboration from the most basic level right up into preclinical trials,” says Hodes. Also unusual, he adds, is the level of information-sharing between pharma companies, who have proven willing to put aside competitive interests at least through the preclinical stage. Much the same is happening with the diabetes AMP program, says Smith, whose NIH division runs the consortium.
Novo Nordisk’s von Herrath reports another industry shift: Bringing patients in to help shape the hunt for drugs. “We want to find out up front what they think would constitute a perfect outcome for them,” says von Herrath. “If we don’t engage them early on, we risk coming up with a solution that doesn’t give them what they want, which means we’ll end up with low patient compliance in taking the medication”—an enormous problem in health care right now, estimated to add as much as $300 billion annually to health care costs in the U.S. In the case of diabetes, says von Herrath, patients with the disease told his group that they wanted a drug that would give them freedom from having to worry about glucose levels for at least three years. “Knowing that is impacting our choice of targets,” he says.
Even if all these efforts pay off in lowering the costs of developing new generations of drugs effective against Alzheimer’s and diabetes, chances are the cost to the health care system of providing them to tens of millions of patients will still be staggering. But making those high additional costs a major policy target would be a mistake, says Kenneth Thorpe, who heads the health policy and management department at Emory University’s School of Public Health, as well as the Partnership to Fight Chronic Disease. Those drug costs may be daunting, he says, but if the drugs are effective, they’re likely to pay for themselves and more by reducing the costs of other types of care to Alzheimer’s and diabetes patients, and especially of having to hospitalize them.
“Ninety percent of U.S. spending in medicine is related to the prevalence of chronic disease,” he notes. “Instead of fixating on high drug prices, we need to think in terms of total per-patient spending.” He cites one study of Medicare patients with multiple chronic diseases, which found that each dollar spent on a prescription drug that helped manage one of the diseases returned $1.80 in other reduced care costs.
If that’s the case, it would make more sense to be worried about underspending than overspending on Alzheimer’s and diabetes drugs. But while accepting enormous price tags for a new generation of more effective drugs might well be a reasonable tradeoff, it increases the need for policy aimed at making sure we get the maximum possible benefit with the diseases that hit the most of us the hardest. We want industry to be rewarded for health impact, not for coming up with smarter ways to boost profits. For now, at least, having to worry about the costs of a new generation of drugs that are effective against Alzheimer’s and diabetes is a problem we ought to all be hoping we face—and the sooner, the better, considering the alternative.
David H. Freedman is co-founder and Executive Editor of Global HealthCare Insights Magazine, and a contributing editor at The Atlantic. His most recent book is “Wrong: Why Experts Keep Failing Us.”