Abstract
Why RCTs Are Necessary in Thyroid Diseases, in Particular Subclinical Thyroid Diseases
With improvements in biochemical testing and increasing numbers of relatively asymptomatic individuals being subjected to blood testing, subclinical thyroid diseases, both subclinical hypothyroidism (SCH) and subclinical hyperthyroidism/thyrotoxicosis (ScTox), have become the most frequent thyroid diseases. They are two- to fivefold more frequent than overt thyroid diseases (4). The clinical complications and functional consequences and effects on quality of life have been extensively addressed in several recent reviews of subclinical thyroid disease and will not be discussed here. The therapeutic options are also regularly discussed in congresses and in numerous papers expressing expert opinion. However, because of the lack of large RCTs, the evidence for choosing one treatment option over another is minimal. Thus, patient preference often determines the therapeutic choice. Because of the high frequency of these conditions and their potentially severe clinical consequences, the impact on public health could be considerable, making the selection of the optimal treatment an important issue.
For instance, it has been demonstrated by several observational studies that ScTox is associated with a 2.5- to 3-fold risk of atrial fibrillation (AF), a cardiac condition associated with adverse outcomes such as stroke (4). The impact of ScTox on mortality remains controversial; excess mortality has been suggested by some, but not all observational studies. ScTox could potentially be responsible for thousands of strokes and deaths in the world each year; alternatively, these epidemiological associations could be explained by a factor separately associated with both AF and ScTox, with no etiological link between them.
Similarly, looking at SCH, an association with adverse cardiovascular events has been substantiated by several meta-analyses of observational studies (5,6). Several short-term studies using relatively small patient groups and intermediary criteria have been published, but no large RCT of L-thyroxine therapy has ever been conducted using hard clinical endpoints. Indeed, there are several arguments against treatment of SCH, particularly in the elderly. An observational Dutch study of 85-year-olds found that SCH was associated with lower risk of mortality (7), and there was no evidence that L-thyroxine treatment improved cognitive endpoints (8). Indeed, studies of the biochemical control of hypothyroidism show that 15%–25% of patients are overtreated with L-thyroxine (9,10), with an associated increase in fracture risk (11,12) and also with potential adverse effects on cardiovascular mortality. In several European countries, the sales of L-thyroxine increase by more than 10% each year (13). It is unlikely that this increase could be due to an epidemic of overt hypothyroidism. This suggests that an unknown proportion of patients with SCH are being unnecessarily treated with L-thyroxine, a situation that carries a true risk of complications in terms of public health.
Why Are There So Few RCTs in Thyroid Diseases, and Why Are They So Difficult to Conduct?
One reason for the few RCTs in thyroid diseases is the absence of any commercial interest from pharmaceutical companies. Antithyroid drugs are old drugs without financial incentive for further studies. Proprietary formulations of L-thyroxine are challenged in most countries by generics, and as already mentioned, sales are spontaneously increasing without great marketing efforts. Thus, these companies have no interest in promoting clinical trials that could serve to restrict the prescription of these drugs. There is little chance that this situation will change unless a new drug appears on the market (which seems unlikely in the coming years).
Another reason is probably that endocrinologists, who lead clinical opinion for patients with thyroid diseases, have been slow to take on the culture of pragmatic clinical research (in contrast, for instance, to oncologists or cardiologists). In endocrinology, we have almost uniquely the ability to observe and measure detailed physiology, and our “default” treatment target for patients with many endocrine diseases is simply to restore the physiological hormone secretion pattern. In general, we have not considered too deeply whether restoring hormone secretions to the physiological or quasi-physiological state is always the best treatment strategy. Indeed, this is compounded by the fact that many of the conditions that secondary care and academic endocrinologists regularly deal with are rather rare, and obtaining large enough patient cohorts for RCTs for some conditions requires cumbersome multinational, multicenter studies, in which each center recruits only a small number of participants. So, with the exception of diabetes care, for which we know that patients will be asymptomatic with very different levels of glycemic control (hence large trials have been performed), we have been largely content to accept that the patient feeling better and having improved physiology are suitable outcomes for treatment in the absence of other evidence. Our own experience and data from the literature illustrates some of the difficulties in performing RCTs of subclinical thyroid diseases.
Information from ClinicalTrial.gov shows that three RCTs of radioiodine therapy for ScTox have been started, all in Europe, in the last 6 years. The three trials addressed mainly cardiac endpoints in an elderly population randomized either to treatment or to follow-up. Two of them (in the Netherlands and the United Kingdom, respectively) have been stopped after 1 or 2 years because of low recruitment. The last trial, in France, is ongoing with 134 patients, less than 50% of the planned recruitment, currently included. Since it is the last ongoing trial for this condition in the world, it has recently obtained additional funding to continue to the end. After numerous discussions with the investigators involved in these three RCTs, we would like to summarize the practical problems involved in conducting these trials and extrapolate more generally to trials on subclinical thyroid disease. Some of these difficulties relate to the patients, while others are linked to the characteristics of the clinical situation or to the healthcare system.
Patient-related problems
The main therapeutic questions in subclinical thyroid dysfunction concern elderly patients. It is well known that conducting trials in elderly patients is more difficult than in younger subjects because they are often less willing to actively participate in therapeutic decisions concerning their health. They may also be reluctant to sign informed consent or have difficulties in understanding the uncertainties inherent to the scientific design of clinical trials such as randomization or blinding. There are also pragmatic difficulties; for instance, the very basic question of traveling to clinical centers for follow-up can be quite problematic for some older patients. Public funding may not always cover these expenses. In addition, older people can find large hospitals and research centers daunting places, where they can become lost. So, many prefer to avoid these venues.
Clinical situation–related problems
First, subclinical thyroid diseases are generally asymptomatic and most often diagnosed incidentally. Even if some patients may have complaints, they are often not related to the thyroid status and this means that patients will not have a symptomatic benefit from the treatment. In this way, subclinical thyroid dysfunction is more a state of clinical risk than an overt disease, and this has similarities with hypertension or hypercholesterolemia.
Secondly, the complications of subclinical thyroid diseases are not easy to explain to patients, in particular the risk of AF in patients with ScTox. AF is also more a condition than a disease; it is generally asymptomatic but carries potentially devastating consequences. Furthermore, pertinent clinical endpoints (AF in ScTox, coronary events in SCH) are not specific to thyroid diseases but depend on numerous other risk factors, some clearly identified, others not. These other risk factors will have a strong influence in determining the absolute risk. In designing RCTs, the dilemma is to choose between selecting high-risk patients or in having studies with wider inclusion criteria but lower event rates. The former will be associated with an increased event rate with a shorter duration of the study and a lower number of patients needed in the trial, but it may increase recruitment difficulties and potentially raise ethical problems. The latter entails a longer duration of the study and a marked increase in the number of patients needed because of the lower event rate, but recruitment could be easier and the results could be extrapolated to a much larger population.
Third, the treatment proposed in the trials of ScTox has been radioiodine, which carries the important risk of inducing overt hypothyroidism. The situation, contrary to that for diabetes or hypercholesterolemia, is that treatment will create a new disease in a patient who does not have any current complaint related to his/her thyroid status, condemning him/her to a lifelong medical treatment. In the French study, after patients have given their informed consent but before randomization, they are asked about their preference either for immediate radioiodine treatment or follow-up without treatment. A preliminary analysis has shown that two thirds of the patients would choose follow-up (presented at the International Thyroid Congress in Paris, 2010).
Finally, all endocrinologists know that radioiodine treatment requires some explanations to the patient because “radioactivity” can frighten some of them. Indeed, the same month that the UK trial started recruiting (November 2006), the murder of Alexander Litvinenko by radioactive polonium poisoning was front-page news. Thus, the idea of receiving a radioactive treatment for an asymptomatic condition does not appeal to many people or their relatives, and this contributes to patients' reluctance to participate in these clinical trials.
Regarding SCH, one study has recently been funded by the European Union (the TRUST trial, a randomized double-blinded study planning to recruit 3000 patients over 65 years of age;
In contrast to the difficulties of studying chronic subclinical disorders, two relatively recently published studies on hypothyroidism have been conducted in the setting of pregnancy (14,15). Despite the specific difficulties of performing RCT during pregnancy, this experience indicates that a relatively short duration study with a clear rationale of the endpoint are probably crucial for obtaining the participation of clinicians and of patients.
Health system–related problems
The three trials on SCH have been conducted in countries with different health system organizations. However, some characteristics are common. First, subclinical thyroid disorders are mainly diagnosed by general practitioners (GPs) who in most countries are not regularly involved in RCTs. Even if GPs refer patients to an endocrinologist, this is often in a context of private practice. These office-based endocrinologists may be reluctant to refer their patients to larger centers conducting these trials for financial reasons. Furthermore, RCTs are often conducted in tertiary care centers. In the absence of a trial, they would not be routinely involved in the care of many of these patients, or would usually only become involved in the care of more complicated patients; for instance, patients who already have AF due to ScTox. Since treatment would be required, these patients would be excluded from many of the most pertinent RCTs relating to ScTox.
How Can We Improve Evidence-Based Knowledge of Thyroid Diseases?
Whereas thyroid diseases are frequent and their treatment could have important effects on public health, the clinical research volume in this area is currently low. Even if the main reason is the absence of interest of pharmaceutical companies, public funding needs to compensate and invest in a few well-powered and definitive clinical trials that could save the public healthcare systems, and individuals who pay directly for their own care, large direct costs, as well as provide future savings related to better long-term health. In particular, there should be future trials of ScTox in elderly patients, studies of SCH in both young and old patients, and appropriate stratification for associated cardiovascular risk factors. These trials should use robust clinical endpoints, in particular for cardiovascular morbidity and mortality. This requires convincing public agency sponsors of the potential benefits of these trials and a broad ethos of cooperation to achieve (i) rational screening strategies to link to trial recruitment aims, (ii) educational efforts, in particular through patient associations, to convince patients to participate, and (iii) cooperation with primary care providers and private office–based and district hospital endocrinologists and academic endocrinologists for the recruitment and follow-up of patients. Moreover, as thyroid diseases are second only to diabetes as the most frequent endocrine pathology, conducting these trials could add structure to the field and prepare the ground for many further trials. Eventually, we can achieve the practice of more evidence-based thyroidology, not to mention evidence-based management of pituitary, adrenal, and gonadal diseases, as well as diabetes care.
Footnotes
Author Disclosure Statement
S.H.P. and B.G. have received speakers fees from Merck-Serono. S.H.P. has been funded to work on subclinical thyroid disease by the Medical Research Council, UK grant G0500783. B.G. has been funded by the French PHRC to conduct a multicenter trial on subclinical hyperthyroidism.