Complexities in Drug Safety

Within pharmacovigilance, adverse reactions that occur within a short time of stopping or starting a particular drug are relatively easy to detect. They are often more readily suspected because of the close temporal relationship and are associated with a higher degree of biological and pharmacological plausibility.

Pharmacovigilance work to detect problems which occur with a longer latency period involves cases where the cause of the reaction is far more difficult to ascertain.

There are a variety of reasons for this phenomenon. It could be that medical records are not available, or that the patient cannot accurately remember how and when they were taking the product – and how and when they stopped. The relationship between the adverse event and the responsible medicine is less likely to be suspected by either the patient or their health professional, as there may be no clear logical, pharmacological or biological relationship apparent between taking the drug and the reaction arising. Suspicions between an Adverse Reaction occurring with a long latency period for a particular drug might only be detected by epidemiological means.

Well established drugs may have the potential to cause serious problems after many years on the market

There have been cases where a drug has been very well established on the market before the potential to provoke the Adverse Reaction has been detected. A robust pharmacovigilance system is essential to ensure that manufacturers and drug regulators are correctly monitoring, evaluating and processing any reports associated with even the longest established products, even after the patent has expired. An example here could be when various pharmaceutical companies that manufacture generic products have placed differing versions of a long-established drug onto the market and there could be reactions reported with just one version of the product, or if there were problems with quality in a batch from a manufacturer – whether a generic or an innovator company – that resulted in adverse reactions.

Examples of known adverse reactions associated with longer latency periods

An example could be the drug isoniazid, which has been estimated to have a latency period of around a year when considering unpredictable immune-mediated hypersensitivity reactions or idiosyncratic reactions. There are other examples of even longer latency periods. Practolol was a beta adrenergic receptor blocking agent which was found to provoke an oculomucocutaneous syndrome. It was withdrawn from sale in 1975 following a series of reports; the average latency period for this problem has since been estimated at over four years.

Particular therapeutic approaches may also present a risk of long latency adverse reactions or delay in detection. Gene therapy may represent one example, whereby the permanent changes to recipient cells may be associated with toxicities which are detected many years later. There may also be cases where viral vectors are used, raising the possibility of whether the potential exists for a latent infection to clinically manifest itself some years later. Some antiretrovirals have been found to be associated with multiple late onset problems related to toxicity. Long term follow up clinical trials are of paramount importance; controlled comparisons and epidemiological methods such as prospective cohort studies may be useful here.

Problems affecting the next generation not the patient

There have also been Adverse Reactions which occurred not for the patient who took the drug, but for their adult children. This was the scenario presented by Diethylstilbestrol (DES) which was prescribed as an oestrogen intended to prevent miscarriage. Used in the USA from 1941, it was not until 1970 that suspicions were raised over a series of cases of a rare vaginal cancer (clear cell adenocarcinoma, abbreviated to CCA). This disease is associated with patients in their seventies but was nonetheless being diagnosed in patients aged between 14 and 22 years old. A case control study finally confirmed the suspicion in 1971, some thirty years after the drug was placed onto the market: the cancer was associated with exposure of the embryo to stilbestrol taken by the mother. It was also later found that there were other serious problems associated with taking the drug, including malformations of the cervix and uterus; decreased fertility; ectopic pregnancies; and higher incidences of spontaneous abortions and preterm births.

Adverse reactions that mimic natural disease

It can be particularly difficult to detect adverse reactions that just represent an increase in frequency over the “normal” incidence of a disease. For example, if a medicine causes heart attacks but is used mostly in an elderly population (for example for treating arthritis), it could be difficult to distinguish what is caused by the drug from what would be expected to occur anyway in these patients without the drug. This is the situation with rofecoxib (Vioxx®) – a drug that was introduced for the treatment of arthritis but was found to have adverse effects on the heart.

False alarms – examples where suspicions around adverse reactions were not found to be associated with medicinal products

One contrary example of why a robust drug safety system must be operational during such long periods of time is the case of Bendectin™. It was prescribed for morning sickness in pregnancy over a 27 year period, by some 20 to 40% of all pregnant women in the USA. When the product was subsequently suspected to cause congenital abnormalities (as had been the case with thalidomide) it was withdrawn from sale. However, a series of case control studies did not find any relationship between the drug and the abnormalities. Given the widespread concern over the potential adverse reaction however, this product was not returned to the USA market even after pharmacovigilance scientists concluded that the suspicions were unfounded, although it was made available as a constituent of some medicines in Europe.

Pharmacovigilance work does not stop once a drug has been on the market for any number of years, nor when a patent simply expires. Every medicine on sale needs to be constantly monitored by drug safety systems supported by services that are responsive, comprehensive, legally compliant and effective – at every stage of the drug’s life cycle.

Reference and suggested further lay reading

1. Barton L. Cobert, MD. (2007). Manual of Drug Safety and Pharmacovigilance.
Massachusetts: Jones and Bartlett.