DISCUSSION
In the present study, we have not been able to demonstrate that the use of established risk factors for thyroid cancer is useful to discriminate obese patients with a higher risk of harbouring thyroid nodules and, therefore candidates to rule out thyroid cancer.
No professional medical society recommends population-based screening for thyroid cancer [20], and in the cases where this strategy has been implemented it has rendered no benefit in clinical terms [21]. However, there is a continued increase in thyroid cancer prevalence in most countries that is generally considered as overdiagnosis and overtreatment [22] and probably indicates, among other reasons, a continued non-judicious use of thyroid ultrasound in some clinical settings [23, 24]. Our aim in this study was to analyse the effectiveness of a screening strategy advocated as cost-effective and therefore recommendable from a theoretical framework [15], adding knowledge to the sensible use of thyroid ultrasound. Our data do not support the advocated strategy and therefore supports the current recommendations in asymptomatic obese adults [13].
A significant difference between the theoretical framework used in the seminal study [15] and our actual results is the prevalence of risk factors. The main discrepancy lies in our higher prevalence of hypothyroidism. Although the prevalence of high TSH may be different in different populations [25, 26], the reported prevalences, also for treated hypothyroidism [27], are generally lower than in our study. A possible explanation for our higher prevalence could be the presence of obesity itself, since it has been reported that high TSH levels are more prevalent in obese patients, at least in children and adolescents [28] and morbid obese patients [29]. The prevalence of TPOAb was similar to some other studies [26] but again higher than usually expected.
Different studies have shown an association between thyroid cancer and the presence of autoimmune thyroiditis [30, 31] or higher TSH levels [32]. Although these associations are not a constant finding [33, 34], some authors have suggested a role for periodic US to detect thyroid malignancy in these situations [30]. In our population the presence of these potential risk factors was not associated to an increased risk of thyroid nodularity or cancer. We can hypothesize that this finding may be due to a different relevance of these factors in the obese population, specially the higher incidence of high TSH. In this sense it has been argued that higher TSH levels in obese patients are not an evidence of thyroid dysfunction but, rather, an adaptation process [35]. Interestingly, in a recent report, thyroid nodules were less common in children and adolescents with autoimmune thyroiditis or high TSH than in their counterparts [36].
The percentage of thyroid nodules with indication for FNAB was higher than expected [15]. This result is in line with other authors’ results [37] and suggests that the prevalence of thyroid nodularity in the obese population is probably higher than in the non-obese population.
Several meta-analyses have found an association of obesity with an increased thyroid cancer risk [3]. Although the present study was not designed nor powered to evaluate the incidence of thyroid cancer or the utility of generalized US screening for its detection in obese patients, we found an incidence above the reported incidence rates for the general population [8] although in line with other populations where a systematic screening has been performed [21]. Most detected thyroid cancers in our study were of low risk and good prognosis and since any screening strategy should counterbalance potential health benefits with the risk of generating an excess in diagnosis, treatment and costs [8, 21], it does not seem that any screening strategy of thyroid cancer in obese individuals will be cost-effective and, in any case, the use of the assessed risk factors is not useful.
Furthermore, this screening strategy results in the detection of a considerable number of nodules without indication for FNAB that could be subjected to follow-up increasing resource use and costs and generating anxiety for patients without any noticeable benefit.
Our study has some limitations. We designed the study to detect differences in the incidence of thyroid nodules and not thyroid cancer, which is the final purpose of the screening strategy. However, detection of thyroid nodules is the first step in the diagnosis of most thyroid cancers [16] and therefore the absence of an increased incidence of thyroid nodules in the obese population with additional risk factors makes the screening strategy ineffective.
Some variability may rise from the fact that ultrasound was not performed centrally. However, the use of a specific classification system might have reduced the effects of this limitation [38].
Additionally, we used as a surrogate for the presence of thyroiditis the positivity of TPOAb. Although other serological markers can be used, TPOAbs are considered the best serological marker to establish a diagnosis of Hashimoto’s thyroiditis [39]. However, an association between thyroid cancer and anti-thyroglobulin antibodies (TgAb) has been reported [31], and therefore we cannot rule out that the use of TgAbs would have changed the results.
Also, the population studied is predominantly morbid obese patients, which may not be representative of other obese populations. It seems reasonable, however, to consider that our study focuses on the obese population at highest risk of cancer and in which the effect of additional risk factors should be more evident. Moreover, BMI was not different between patients without or with thyroid nodules.
Furthermore, there is some risk of selection bias. For unknown reasons, not all obese patients attended were offered participation and, this fact may have selected the population finally evaluated. Additionally, we decided to exclude patients with other autoimmune diseases. The reason for doing so, was to not overestimate de prevalence of high TPOAbs as it is known that the prevalence of thyroid autoimmunity is increased in the presence of other autoimmune disorders [40].
Despite these limitations, the number of patients evaluated give us confidence that the results are reliable and therefore we can conclude that the use of the studied risk factors (specifically TSH and TPO) is not useful to discriminate obese patients with a higher risk of harbouring thyroid nodules.