A Step Forward in Understanding the Relevance of Genetic Variation in Type 2 Deiodinase

Up to 20% of treated hypothyroid patients have residual hypothyroid complaints despite normalized thyroid-stimulating hormone (TSH) and free thyroxine (FT4) levels. The pathophysiological mechanism behind these persistent complaints is one of the major unresolved questions in thyroidology. In euthyroid individuals, ~20% of serum triiodothyronine (T3) results from direct thyroidal secretion, with the remainder derived from extrathyroidal conversion of T4 to T3 by type 1 and 2 deiodinase (D1 and D2). The vast majority of hypothyroid patients are treated with levothyroxine (LT4) monotherapy, and their serum T3 levels are therefore fully dependent on extrathyroidal T3 production. Thus, patients receiving LT4 monotherapy generally have higher T4/T3 ratios than euthyroid individuals. Some patients with normalized TSH levels will have serum T3 levels at the lower end or even less than the reference range, with high FT4 levels. D2 accounts for ∼70% of circulating serum T3 levels in humans, and D2-dependent local T3 production is an essential determinant of the thyroid hormone status of the brain (1). Therefore, decreased D2 expression or function has been hypothesized as a potential mechanism underlying the residual complaints in treated hypothyroid patients.

In the past decade, various studies have investigated the effects of common genetic variation in the DIO2 gene on serum thyroid parameters, thyroid-related outcomes, or D2 expression and function (2–13). The most frequently studied variant is rs225014, leading to a Thr to Ala conversion of codon 92 in exon 2 of the DIO2 gene (D2-Thr92Ala). This variant has been analyzed in many candidate gene studies in population-based and patient cohorts and has shown no effects on serum TSH, FT4, or T3 levels. In contrast, associations with other thyroid-related outcomes such as diabetes, psychological well-being, hypertension, and the risk of osteoarthritis have been reported for this variant (2–13). These candidate gene studies are relatively easy to perform and, if sufficiently powered, have the potential to detect even small effects of common genetic variants. However, these studies are also prone to false-positive results, especially when the study sample size is relatively small. Replication in independent cohorts is therefore mandatory to avoid such false-positive results and possible publication bias. In addition, in vitro data confirming the functionality of a particular genetic variant can provide further support for a true effect. Most of the previously cited studies included a relatively small sample size, and most of the associations with thyroid-related endpoints have not been replicated to date (13). However, the lack of associations of D2-Thr92Ala with serum thyroid function tests has been consistent among several studies, including a recent large population-based study by Wouters et al. (12) of 4336 individuals (12). This suggests that this variant does not have any noteworthy effect on serum thyroid parameters in the general population.

Nevertheless, much uncertainty remains concerning the clinical relevance of genetic variations in the DIO2 gene. In the current issue, Castagna et al. (14) report on their association and functional analyses of genetic variations in DIO2. The study included 140 thyroidectomized patients, who had not had thyroid dysfunction or thyroid function-altering medication before surgery. Only patients in whom the pre- and postoperative serum TSH levels did not differ by >0.5 mIU/L were included. Thyroid function tests were performed before (≤10 months) and after (≤6 months) thyroidectomy when thyroid function tests were stabilized on LT4 monotherapy. As expected, the mean FT4 levels were higher and free T3 (FT3) levels were lower after thyroidectomy. The TSH levels remained similar, consistent with the inclusion criteria applied by the investigators. In a subset of 102 patients, who had agreed to genetic testing, Castagna et al. (14) showed that the decrease in FT3 levels after thyroidectomy was greater in D2-92Ala carriers than in those without this variant. No such differences were seen between genotype groups for the remaining thyroid parameters. Because previously published studies did not detect any effects of D2-92Ala on serum thyroid parameters, including FT3 levels, it is remarkable that in the study by Castagna et al. (14), the decrease in FT3 levels after thyroidectomy in D2-92Ala carriers was larger than that in noncarriers (2–12). This discrepancy in findings can be explained by several factors. Most of the larger well-powered studies included population-based cohorts, with, therefore, predominantly euthyroid individuals (3, 5, 7, 9, 11, 12). Thus, the effects of the 92Ala variant might have been masked by thyroidal T3 production in these euthyroid individuals. Therefore, it could be that the effects of Thr92Ala become only apparent in athyreotic patients receiving LT4 monotherapy. However, several studies, including the previously mentioned study by Wouters et al. (12), found no differences in TSH, FT4, or FT3 levels or T3/T4 ratios between Thr92Ala genotype groups of hypothyroid patients treated with LT4 monotherapy (2, 12, 15). An alternative explanation for the differences in the findings can be found in the study design of the different studies and limitations thereof. First, similar to many of the previously cited candidate gene studies, the present study by Castagna et al. (14) is a relatively small study with only 52 D2-92Thr/Ala and 13 D2-92Ala/Ala patients, which made their study prone to false-positive results. As an example, the Ala/Ala group consisted of 13 patients, and their results for the differences in FT3 levels are shown in their Figure 1D, with a mean difference of −0.5 pg/mL. If, as a thought experiment, one removes the last patient with the largest difference and replaces this by duplicating the patient with the least difference, the change in FT3 levels would be only approximately −0.3 pg/mL, losing the dose–response relationship and the significance becoming questionable. Therefore, replication of the findings is crucial to render this result credible. Also, even if these results are true-positive findings, it is difficult to assess to which population these results would pertain, as several arbitrary cutoffs were used. For example, Castagna et al. (14) used 10 and 6 months as the cutoffs for inclusion of thyroid profiling before and after surgery, and only patients whose TSH levels before and after thyroidectomy did not differ by >0.5 mU/L were included. Furthermore, the DIO2 variants were only determined in 102 of the 140 patients, and it is not clear whether that group differed from the group in which the variants were not determined. All these factors make it difficult to find populations with similar characteristics and impossible to generalize the results to other populations. Finally, because the differences in serum FT3 levels between the genotype groups both before and after surgery were small (0.2 pg/mL), one could question whether these differences could yield clinically relevant effects. The same arguments regarding sample size and generalizability hold true for the association analyses of the D2-rs12885300 variant. Therefore, a large, well-powered study is needed to replicate the effects of D2-Thr92Ala on serum thyroid parameters in athyreotic patients receiving LT4 monotherapy before meaningful conclusions can be drawn.

The major strength of the study by Castagna et al. (14) lies in its functional studies on the D2-Thr92Ala variant. Various functional studies have previously been performed on D2-Thr92Ala. Canani et al. (16) showed decreased D2 activity in muscle and thyroid samples from patients with type 2 diabetes mellitus who were homozygous for D2-92Ala compared with D2-92Thr carriers (16). However, this decreased D2 activity could not be shown in transiently transfected D2-92Ala cells (5, 16), suggesting that the observed effect of 92Ala in the former study might be driven by a variant elsewhere on the genome that is in high linkage with DIO2-Thr92Ala. In contrast, a more recent study by McAninch et al. (17) has shown that D2-92Ala has a longer half-life and accumulates in the Golgi in transfected human embryonic kidney (HEK) cells, suggesting that this variant itself does have functional consequences. The investigators also studied the gene expression profiles in the cerebral cortex of 19 postmortem D2-92Ala carriers. Although differences in the expression of several genes could be detected, the expression of T3-response genes remained unchanged, suggesting that the effects of the 92Ala variant on cognitive endpoints might be mediated via mechanisms other than changes in thyroid hormone levels.

Castagna et al. (14) performed a series of functional experiments that provide new insights into the functional consequences of the D2-92Ala variant. Because in vivo detection of endogenous D2 is not possible owing to the lack of functional antibodies, they elegantly used knock-in 3XFlag-D2 mice. D2 was expressed in the endoplasmic reticulum perinuclear space during cell differentiation in both D2-92Thr- and D2-92Ala-transfected mouse muscle stem cells. Neither were differences found in protein stability between the genotype groups. Next, they showed lower intracellular T4-to-T3 conversion in D2-92Ala- compared with D2-92Thr-transfected myoblasts from DIO2-null mice. A similar effect was shown in the pituitary cells from DIO2-null mice, as the response to T4-induced suppression of TSH was significantly lower in the D2-92Ala variant compared with D2-92Thr–transfected pituitary cells. Therefore, by systematically investigating various aspects of D2 function and expression that could be affected by the variant, the investigators have convincingly shown that in their system the D2-92Ala variant leads to less conversion of T4 to T3 in vivo, despite identical subcellular localizations and somewhat higher protein expression. Because the latter experiments were performed using D2-transfected cell lines, these findings do suggest that the observed effects result from the 92Ala variant itself and are not mediated via a variant that is in high linkage disequilibrium. In addition, the fact that in the previously cited study by Canani et al. (16) and Peeters et al. (5) D2-92Ala did not influence T4-to-T3 conversion in transfected HEK and COS cells could point toward cell-dependent effects of this variant. The observation that this variant has a longer half-life and accumulates in the Golgi in HEK cells, while especially the expression profiles of non-T3 response genes were affected in the cerebral cortex, further supports this idea of cell-specific effects of this variant (17).

In conclusion, although the clinical relevance of genetic variations in the DIO2 gene is still disputable, the study by Castagna et al. (14) has provided important novel evidence for functional effects of the D2-92Thr/Ala variant in multiple cell lines. Although previous data have convincingly shown that this variant has no effects on serum thyroid hormone parameters in the general population, the data in the study by Castagna et al. (14) suggest that athyreotic D2-92Ala carriers might be more prone to a decrease in T3 levels. However, no differences in three other cohorts of LT4-treated patients with similar or larger sample sizes were found in previous studies (2, 12, 15). Therefore, there is a strong need for replication in well-powered studies of LT4-treated athyreotic patients to clarify whether these patients are indeed more vulnerable to the effects of this variant. Importantly, these studies should also investigate whether these effects contribute to the residual complaints that impair the quality of life of many treated thyroid patients. Until then, no conclusions can be drawn regarding the potential clinical relevance of D2 variants in the management of patients with thyroid disease.

Abbreviations:

 
• D1

  type 1 deiodinase

 
• D2

  type 2 deiodinase

 
• FT3

  free triiodothyronine

 
• FT4

  free thyroxine

 
• HEK

  human embryonic kidney

 
• LT4

  levothyroxine

 
• T3

  triiodothyronine

 
• T4

  thyroxine

 
• TSH

  thyroid-stimulating hormone.

Acknowledgments

Disclosure Summary: The authors have nothing to disclose.

References

Author notes