Ross Camidge, MD PhD

Few would argue that crizotinib, the first lung cancer drug to hit the market in six years when Pfizer introduced it under the trade name Xalkori in 2011, has proven to be a superb therapy for those with non-small cell lung cancer linked to a specific genetic mutation.

That it costs about $10,000 a month is established fact.

The drug’s cost is why crizotinib is a flashpoint for debate, despite its proven benefits to patients. As a collaboration between a University of Colorado School of Medicine oncologist and a CU health economist has shown, what looks like a concrete cost-benefit analysis can crumble under the pressure of a single changed assumption. Their findings could apply broadly at a time when the products of health-economics models can influence buying decisions affecting many thousands of patients a year.

The CU oncologist in question is Ross Camidge, MD, PhD, who has been treating ALK-positive lung-cancer patients with crizotinib since the CU Cancer Center was one of five U.S. sites for the drug’s phase 1 clinical trial. Running the numbers was Adam Atherly, PhD, a CU School of Public Health economist. Camidge and Atherly had collaborated on a 2012 studyof patient-screening costs related to crizotinib, so when the Journal of Thoracic Disease was looking for an editorial examining an astronomical crizotinib cost-benefit number published in February 2014, the editors knew where to turn.

Big bucks. The 2014 study calculated a cost of $255,975 per quality-adjusted life-year (QALY; used in health economics to measure the incremental cost per added unit of health) gained for patients taking their two $330 crizotinib pills each day. The Canadian authors deemed the drug to be “not cost effective.”

“What was striking was that the development of crizotinib in ALK-positive lung cancer has been one of the game changers,” Camidge said. “So it’s a bit of a blow when the Canadians come out and say, ‘Don’t use it because it’s not cost-effective.’”

Camidge was of a mind that “there’s no point in having a miracle drug if it’s stuck on a shelf and no one can reach it.” So he and Atherly looked closely at how the Canadian team made its calculations.

The CU team had no beef with the Canadians’ methodology or the modeling they used to analyze the expense of crizotinib. But by changing just one assumption, they arrived at a cost of $143,421 per QALY. That’s right: One changed assumption cut the estimated cost per unit of benefit by 44 percent. Such a drop could well influence Heath Canada and other single-payer systems that have long used such models to decide on whether or not to pay for a given drug.

That’s still more than the roughly $100,000 per QALY a single-payer state health care system typically pays, Camidge said. Their point, however, is that the concrete numbers flowing from even the best numerical models depend on underlying assumptions that can be squishy.

In this case, the assumption Camidge and Atherly changed was the percentage of normal quality of life a patient on crizotinib typically enjoys. The Canadian study estimated it to be 50 percent of normal. This might be a fair figure for someone fighting through rounds of chemotherapy, but crizotinib’s side effects (the official list includes vision disorders, nausea, diarrhea, vomiting, edema, and constipation) tend to be mild, more like those of antidepressants or diabetes medications.

Given that physicians in Canada have little direct experience with the drug, the people estimating the quality of life derived from this drug may not have had first-person experience with side effects. Camidge and Atherly simply upped the percentage of normal quality of life for crizotinib patients from 50 percent of normal to 90 percent, a better fit with Camidge’s long experience with patients on the drug.

Broader implications. Why be so concerned about this one drug’s fate in Canada, particularly when it’s still really expensive, even with the changed assumption? For one, Camidge said, financial cost-benefit analyses, long a staple of national health care systems in places such as Canada, the United Kingdom, and Australia, are poised to become more commonly applied as the
U.S. health care system increases its scrutiny of the cost of care.

Second, the audience for health-economic analyses is evolving, Camidge said. What used to be the domain of health economists is broadening to include oncology and other fields, as the publication of the original Canadian study in the high-profile Journal of Thoracic Disease shows. It’s important, Camidge said, that these new audiences understand that the products of these models, confidently reduced to single-digit-dollar exactitude, in fact depend on myriad factors of varying reliability.

In addition, Atherly said, cost-benefit calculations can change starkly with time – costs plunge when drugs go generic, and benefits can pile up as clinicians gain familiarity with therapies, for example.

“We’re trying to educate an audience that’s now being asked to consume this type of data for perhaps the first time,” Camidge said.

Added Atherly, “Clinicians need to be aware of how sensitive the results of these things are to each assumption.”
This is not to dismiss the value of cost-benefit modeling as a discipline, Atherly said.

“These are questions we need to ask when we’ve got the great new therapy,” he said. “Does the benefit justify the cost in a world where we can’t afford everything?”