Safety and efficacy in medicine is always a balance, and the evidence is increasingly that pharmaceuticals are associated with far more side effects than we currently appreciate.
But there are two qualifications regarding safety that are insufficiently appreciated. One is patient group, the other is indication, or therapeutic use. We all know that severe conditions can tolerate more severe side effects than mild ones. So, cancer patients treated with chemotherapy suffer a whole range of severe side effects that would be considered intolerable in a milder condition.
As an example, thalidomide has recently been re-introduced for the treatment of multiple myeloma, despite its well-known and severe intolerability and toxicology that led to its withdrawl for the treatment of insomnia and morning sickness. Thalidomide comes with its own risk management plan, but it is unusual in this respect; for the vast majority of pharmaceuticals, once approved for one indication, there is no legal prevention to their prescription for almost any other indication.
Sometimes this occurs very often. For instance, bone morphogenetic protein 2 or BMP-2, a recombinant bone-growth promoting hormone, is prescribed off-label in about 85% of cases. Another example which recently hit the headlines, was that of niacin, and other several prescription cholesterol drugs, which it has been claimed are used 75% of the time outside their regulatory label; despite being able to lower cholesterol, these drugs are generally not associated with beneficial cardiovascular outcomes. In the USA, there is a balance to be struck because of the constitutional first amendment right to free speech. This has recently thwarted the attempts to restrict what many see as irresponsible promotion of prescription drugs, outside their regulatory label, with unsurprising consequences to health.
This is important for two reasons. Firstly, the evidence of efficacy has not been demonstrated to regulatory standards in the non-approved use. However this article is focussed on the second reason, that the safety profile in the non-approved use is not necessarily acceptable. Moreover, where off-label use is widespread, the patient’s experience of prescription use is likely to be accompanied by poor outcome both in terms of safety and in terms of efficacy.
So, in the case of BMP-2, there has been a significant increase in complications alongside the increase in off-label use. For example, a particular area of concern is the ectopic growth of bone (i.e. outside the fusion area) in spinal fusion surgery, which is an off-label use of BMP-2. In some cases of cervical spinal fusion, there have been alarming reports of head and neck swelling, resulting in compression of the airway and neurological structures of the neck. There is further concern about the cancerous potential, in situations involving long-term administration of BMP-2, a known growth factor. Other serious adverse effects include immune stimulation and leg pain. The FDA has considered these side effects in association with the approved use, and approval is granted in that context; however the much more widespread use in other, less serious conditions, has not been granted approval. In such uses, not only are the supportive efficacy data lacking, but the safety profile may well be inappropriate. It is incorrect to assume that regulatory approval is a blanket assurance of a drug’s safety.
However, it is not always bad. Sometimes, things can work in a positive direction, and the balance can even be more positive in a secondary use. Let’s take rosiglitazone, for instance, which has been associated with a 0.4% excess risk of heart attack in type II diabetic patients. This is concerning since such patients need to take care of their cardiovascular health. A typical type II diabetic is middle aged or older, overweight and male: all are risk factors for myocardial infarction.
Now, rosiglitazone works by reducing sub-sensitivity to insulin, or so-called insulin resistance. This is a phenomenon in other medical conditions too, such as polycystic ovarian syndrome, or PCOS. Given the link, it is unsurprising that rosiglitazone has been trialled in PCOS, and with good results. What is more interesting, however, is to consider the risk profile for such patients with respect to the previous finding of excess risk for heart attacks. PCOS patients are always female and usually pre-menopausal, and at a much much lower risk for myocardial infarction.
Now let’s deal with different patient groups without change of indication. Consider for instance paroxetine, which has been widely attacked for its side effects, particularly of suicide risk in adolescents. In younger depressed patients, the clinical trial evidence shows paroxetine to be ineffective, but also increases suicide risk; despite the adverse profile in young patients, the approval in adults was not rescinded, because the suicide risk was counterbalanced by a statistical improvement in depression.
In these two different groups, we have a different safety/efficacy equation to consider. Again, it is certainly not the case, as some argue, that off-label prescription to all/any patient is ethically justified because the safety is assured to regulatory standards by the approval in one group.
The balance goes the other way when we are dealing with middle-aged mothers, who are common amongst depressed patients. They represent a group with a low propensity for suicide but in whom we should be more concerned about breast cancer risk. Paroxetine has been associated with a 36% reduced risk of breast cancer, in a retrospective study of nearly 3000 incident breast cancer cases on various SSRIs.
So, in conclusion, safety is a relativistic phenomenon: it is stratified by patient group and considered acceptable or not by virtue of the use for which the drug is administered. We need more, and better information to unravel these complexities. Importantly, I suggest that therapeutic use is always written on a drug’s prescription, and we find better ways to monitor the use for which drugs are prescribed.
