Carcinogenicity of lipid-lowering drugs

Authors: Newman T, Hulley S.
Source: JAMA. 275:55-60. January 3, 1996.
Institutions: School of Medicine, UCSF.
Financial support: not specified.


This article has created considerable concern among the public about the safety of cholesterol-lowering medication. The authors report on carcinogenicity data culled from the Physician's Desk Reference and then argue against drug therapy of hyperlipidemia in the primary prevention of coronary disease. The authors' points are rebutted in an accompanying editorial.

Carcinogenicity data

The authors summarize animal carcinogenicity data for lipid-lowering drugs, obtained from the 1994 PDR. These data are then compared to data for commonly used antihypertensive medications. The method for comparing animal doses to human doses was changed for lovastatin and gemfibrozil in 1993: prior to that year, the doses were compared on an administered, mg/kg basis; thereafter blood-levels achieved (area under the curve) were compared. Since the small animals used in the studies metabolize these drugs more quickly than humans, much higher doses are required in the animals to achieve comparable blood-levels.

The results are given in a table, which reveals that the fibric acid derivatives and the statins cause a number of different malignancies and benign tumors (in particular liver, thyroid and Leydig cell) in rats and mice, at blood levels that are often comparable to those seen with human dosing regimens. No studies were cited for niacin; cholestyramine was found to enhance intestinal carcinogenesis only in the presence of other carcinogenic factors.

Another table presents similar data for a number of commonly used antihypertensive agents, such as ACE inhibitors, beta-blockers, calcium channel blockers and diuretics. For 41 drugs for which results were reported, only 32% were found to be carcinogenic (15% malignant, 17% benign).

Authors' Discussion

The authors note that carcinogenicity in laboratory animals cannot be directly extrapolated to humans, particularly if an established mutagenic mechanism has not been found (as in the drugs examined here). They argue, however, that the evidence for cancer-causation in rats and mice is worrisome, and that the exposures that cause tumors in animals, when expressed as blood-level areas under the curve are comparable to blood-levels achieved in humans on therapeutic doses.

In terms of the individual drugs, the authors are most critical of the fibrates (gemfibrozil), because of a higher total mortality in two large studies. For the statins, they note that there was no increase in cancer deaths after 5.4 years of follow-up in the Scandinavian Simvastatin Survival Study, but express concern that the latency period for the development of cancer could be 10 years or longer. They express much less concern about niacin and the resins.

Beyond the data presented here, the authors make a number of points that they have made in the past, in other articles. They reaffirm the role of lipid-lowering therapy in the secondary prevention of coronary disease (hyperlipidemic patients with documented atherosclerotic disease, at high risk of coronary events). They suggest that, for primary prevention, where the short-term risk of coronary events is low, meta-analyses of trials suggest that the overall risk of cholesterol lowering may exceed the benefits. They note that meta-analyses of trials looking specifically at cancer deaths found no significant increase when follow-up was extended, but caution that medications may not have been continued for extended periods in these trials.


Authors: Dalen J, Dalton W
Source: JAMA. 275:67-9.

In an accompanying editorial, Dalen and Dalton rebut the points made by Newman and Hulley.

They argue that the rodent carcinogenicity data is flawed: the rats used have a high spontaneous cancer rate, and 50% of chemicals tested in these animals will be identified as carcinogens. Although cancer was produced in the animals at blood levels similar to those obtained with therapy in humans, in order to achieve these levels, much higher mg/kg doses had to be administered to the animals (because of rapid metabolism). Such high oral doses can lead to very high tissue exposure in the liver and GI tract, causing non-specific damage and possibly neoplasia on that basis. In fact, many of the tumors noted in the animals were in the GI tract and liver.

Turning to the analysis of clinical trial data, the authors point out that one of three meta-analyses found no increase in cancer deaths, the other two analyses found only a very small excess cancer death rate in the treatment groups (0.002-0.003), and that these were accounted for by two trials, a clofibrate and a dietary trial. Looking at the late follow-up data from several large trials, there is also no evidence of increased cancer incidence here.

Dalen and Dalton conclude that primary prevention of coronary disease by pharmacological treatment of hyperlipidemia should not be questioned based on the data presented here, although the appropriate age to begin screening is not yet clear and has important cost-benefit implications.


The article by Newman and Hulley was picked up by the media, and has caused quite a bit of anxiety so far. In my opinion, a few points need to be emphasized: 1/9/96

Reader comments

Date: Sat, 10 Feb 96
From: Howard Homler < >
Subject: carcinogenicity of lipid-lowering agents

 I would have far preferred to see their article listed as an opinion piece or editorial-type format, rather than as an original article. It did cause a lot of distress to my patients and the news reports were very misleading. Although I feel strongly that journals should print other than "mainstream" views, it's important to realize that publication is assumed by the public to mean "truth" and "endorsement."

 Howard Homler MD, FACP
Sacramento, California

June 21, 1996

Dr. Petros Hiou from Greece points out that my statement (last sentence under carcinogenicity data, above):

"For 41 drugs for which results were reported, only 32% were found to be carcinogenic (15% malignant, 17% benign)."

makes no sense. Benign carcinogenicity? Should read: "32% were found to induce tumors", or something similar. Thanks Dr. H.

Journal Club on the Web home page
Submit a comment about this article

Site Meter