Messerli noted that most of the patients in the study were being treated with one of the seven available drugs, so that even if the effect is real, the findings may not apply to all of the medications.
Angiotensin-receptor blockers are widely used - some 82 million prescriptions were written in 2009, according to IMS Health Inc., which tracks drug sales. And their value is "compelling," according to Dr. Clyde Yancy, of Baylor University Medical Center in Dallas.
They reduce the risk of stroke and the risk of death from left ventricular dysfunction after heart attack, they slow progression in chronic kidney disease, and they improve outcomes in heart failure, Yancy said in an e-mail.
Yancy said he applauds Sipahi and colleagues but cautioned that the results are far from solid evidence of risk. "We usually describe these kinds of findings as a 'signal' but given the modest result," he said, "perhaps this is more of a 'hint'."
Above all, he said, patients on the drugs -- Yancy said he's included in that number -- should not stop treatment with what he called a "cornerstone of cardiovascular care."
Sipahi and colleagues said their analysis was prompted by the result of a 2003 trial of one of the drugs, which showed an unexpected increase in the proportion of fatal cancers among those getting the medication.
Since several other large randomized controlled trials have since reported, they decided to perform a meta-analysis. Literature searches turned up five trials with data on the occurrence of new cancers in a total of 61,590 patients. As well, five trials had data on specific solid tumors and eight trials had data on cancer-related deaths.
All told, nine randomized controlled trials were included in the analysis, with a total of 94,210 patients, they reported. The studies included data on four of the seven available angiotensin-receptor blockers - losartan (Cozaar), candesartan (Atacand), telmisartan (Micardis), and valsartan (Diovan).
The absolute risk increase for caner in the study was 1.2 percent over four years, Sipahi and colleagues said, which should be put in the context of an estimated 41 percent lifetime risk of cancer.
Using the data from those three trials, Sipahi and colleagues calculated that one excess cancer diagnosis would be made for every 105 patients treated with angiotensin-receptor blockers - the so-called "number needed to harm."
Combined with the large number of patients taking the drugs, that figure "is a very scary way of looking at it," Sipahi said in an interview.
Eight trials reported cancer-related deaths, but there was no significant difference between treated and control groups, they found.
Sipahi and colleagues said a strength of the study is that all of the data comes from randomized controlled trials, which should reduce the effect of confounding variables.
On the other hand, they said, the findings come from pooled data from trials not specifically designed to assess cancer, and cancer data was not available from several large trials of the drugs, which could lead to publication bias.
As well, they said, "meta-analyses are generally considered less convincing than a large prospective trial."
Nonetheless, Sipahi and colleagues concluded, the modest but significant safety signal warrants more investigation.
In the meantime, many doctors said prescribing patterns will probably not change.