In 2003, the Food and Drug Administration approved a lung cancer drug called Iressa, but not because the medicine helped patients live longer. Instead, the agency endorsed the drug based on the fact that it reduced the size of tumors in about 10 percent of patients.
This “surrogate” measure was expected to predict future increases in survival time. But it didn’t. A subsequent trial found that Iressa did not prolong patient lives. And in 2012 — after nearly a decade on the market at a cost to the US health care system of around $285 million — the FDA, at the manufacturer’s request, revoked the drug’s approval .
This is a cautionary tale of what can happen when regulatory decisions are based on proxy measures without proof that drugs are extending lives.
In the case of Iressa, the FDA eventually corrected course and pulled the ineffective drug. But a new study suggests that many more cancer agents of questionable benefit remain on the market — even as Congress is debating a bill that would increase use of surrogate measures to approve drugs.
Dr. Vinay Prasad  from the Oregon Health and Sciences University and Dr. Chul Kim from the National Cancer Institute looked at all cancer medicines approved between 2008 and 2012 on the basis of surrogate markers. In clinical studies, some drugs lengthened the time after treatment before doctors detected a cancer worsening. Others led to tumor shrinkage on scans.
The researchers searched for subsequent reports of these 36 drugs boosting survival rates. They found evidence of such meaningful gains for only 14 percent, or just five of them. That’s a small number.
The increased use of surrogate measures reflects a growing clamor for speedier approvals of drugs, especially cancer medicines and other treatments that can combat life-threatening ailments. And drug makers benefit because the trials needed to win FDA approval are shorter and smaller, which means gathering patient data costs less.
But as the authors note, the results suggest the FDA may be approving drugs that, it so happens, are often expensive and sometimes toxic, yet do not improve survival rates. And when pricey medicines do not perform as expected, the overall cost of health care can rise.
“We should think twice about using surrogates in situations where it does not correlate with prolonging life,” Prasad told us.
The report, which appeared in JAMA Internal Medicine,  arrives just as federal lawmakers consider legislation that critics say could lower standards for drug approvals by, in part, widening reliance on surrogate endpoints. Known as the 21st Century Cures Act, the bill  passed the House in July, and the Senate is expected to take up a version next year.
The new analysis suggests the FDA should slow the pace of early approvals. But many experts counter that there’s more to cancer treatment than simply pushing back the date of a patient’s last breath.
“If you have an incurable disease and the goal [of treatment] is to reduce symptoms, there are lots of ways to improve the quality of life that do not necessarily improve the overall survival,” said Dr. Clifford Hudis, an oncologist at Memorial Sloan Kettering Cancer Center in New York.
Few clinical studies gather data on symptom improvements or other kinds of subjective measures that might matter most to patients. But according to Hudis, “It is fundamental. If the tumor is smaller, you should feel better.”
Dr. Mikkael Sekeres, an oncologist at the Cleveland Clinic, also points out that dying patients don’t have time to wait for the kinds of lengthy trials needed to prove survival benefits. “If you look at it from the standpoint of a patient who has a terrible cancer,” he said, “I would want to try a drug that has promise as soon as possible.”
Prasad said that he doesn’t oppose approvals based on surrogate endpoints, especially for drugs used to treat patients in dire circumstances. But he does believe the FDA should be doing more to ensure drug makers run required follow-up studies to determine survival benefits. And he worries that the agency has a spotty record.
For instance, in fiscal 2012, the last year for which data were available , one-third of the all follow-up studies mandated by the FDA — for drugs to treat cancer and other diseases — did not have specific milestones or completion dates to submit data.
The FDA can punish recalcitrant companies for not conductin g post-approval studies by declaring their drugs are misbranded and imposing fines. However, the agency rarely if ever does so. Nor does it often revoke approvals, as it did for Iressa, if or when drugs fail to show a survival boost.
This is an unhealthy precedent. It may be OK to take a leap of faith when you approve a drug, but you need to follow it up. At a time when the FDA is under ever more pressure to approve drugs, more oversight is required to ensure these treatments fulfill their promise. And if not, we need to be willing to cut our losses when some medicines simply don’t pan out.
Editor’s note: This first appeared today as the weekly Pharmalot column  in The Boston Globe.