Selenium supplementation for Hashimoto's thyroiditis
Abstract
Background
Hashimoto's thyroiditis is a common auto‐immune disorder. The most common presenting symptoms may include anxiety, negative mood, depression, dry skin, cold intolerance, puffy eyes, muscle cramps and fatigue, deep voice, constipation, slow thinking and poor memory. Clinical manifestations of the disease are defined primarily by low levels of thyroid hormones; therefore it is treated by hormone replacement therapy, which usually consists of levothyroxine (LT4). Selenium might reduce antibody levels and result in a decreased dosage of LT4 and may provide other beneficial effects (e.g. on mood and health‐related quality of life).
Objectives
To assess the effects of selenium supplementation on Hashimoto's thyroiditis.
Search methods
We searched the following databases up to 2 October 2012: CENTRAL in The Cochrane Library (2012, Issue 10), MEDLINE, EMBASE, and Web of Science; we also screened reference lists of included studies and searched several online trial registries for ongoing trials (5 November 2012).
Selection criteria
Randomised controlled clinical trials that assessed the effects of selenium supplementation for adults diagnosed with Hashimoto's thyroiditis.
Data collection and analysis
Study selection, data extraction, assessment of risk of bias, and analyses were carried out by two independent review authors. We assessed the quality of the evidence of included studies using GRADE. We were unable to conduct a meta‐analysis because clinical heterogeneity between interventions that were investigated is substantial.
Main results
Four studies at unclear to high risk of bias comprising 463 participants were included. The mean study duration was 7.5 months (range 3 to 18 months). One of our primary outcomes－'change from baseline in health related quality of life'－and two of our secondary outcomes－'change from baseline in LT4 replacement dosage at end of the study' and 'economic costs'－were not assessed in any of the studies. One study at high risk of bias showed statistically significant improvement in subjective well‐being with sodium selenite 200 μg plus titrated LT4 compared with placebo plus titrated LT4 (relative risk (RR) 4.67, 95% confidence interval (CI) 1.61 to 13.50; P = 0.004; 36 participants; number needed to treat (NNT) = 2 (95% CI 2 to 3)).
Selenomethionine 200 μg reduced the serum levels of anti‐thyroid peroxidase antibodies compared with placebo in two studies (mean difference (MD) ‐917 U/mL, 95% CI ‐1056 to ‐778; P < 0.001; 85 participants) and (MD ‐345 IU/mL, 95% CI ‐359 to ‐331; P < 0.001; 169 participants). Pooling of the studies was not feasible due to marked clinical heterogeneity (I2 = 99%). In a further comparison within the first study where selenomethionine was combined with LT4 the reduction in TPO antibodies was even more noticeable (MD ‐1508 U/mL, 95% CI ‐1671 to ‐1345; P < 0.001; 86 participants). In a third study, where LT4 was added to both intervention arms, a reduction in serum levels of anti‐thyroid peroxidase antibodies favoured the selenomethionine arm as well (MD ‐235 IU/mL, 95% CI ‐374 to ‐95; P = 0.001; 88 participants). Although the changes from baseline were statistically significant in these three studies, their clinical relevance is unclear. Serum antibodies were not statistically significantly affected in the study comparing sodium selenite 200 μg plus titrated LT4 with placebo plus titrated LT4 (MD ‐25, 95% CI ‐181 to 131; P = 0.75; 36 participants).
Adverse events were reported in two studies (1 of 85 and 1 of 88 participants, respectively). Selenium supplementation did not appear to have a statistically significant impact on the incidence of adverse events (RR 2.93, 95% CI 0.12 to 70.00; and RR 2.63, 95% CI 0.11 to 62.95).
Authors' conclusions
Results of these four studies show that evidence to support or refute the efficacy of selenium supplementation in people with Hashimoto's thyroiditis is incomplete. The current level of evidence for the efficacy of selenium supplementation in the management of people with Hashimoto's thyroiditis is based on four randomised controlled trials assessed at unclear to high risk of bias; this does not at present allow confident decision making about the use of selenium supplementation for Hashimoto's thyroiditis. This review highlights the need for randomised placebo‐controlled trials to evaluate the effects of selenium in people with Hashimoto's thyroiditis and can ultimately provide reliable evidence to help inform clinical decision making.
Plain language summary
Selenium supplementation for Hashimoto's thyroiditis
Hashimoto's thyroiditis is a common disease in which a form of chronic inflammation of the thyroid gland results in reduced function of the gland. It is an auto‐immune disorder, which means that a person's own immune system attacks the thyroid gland, so that it no longer makes adequate quantities of thyroid hormones (hypothyroidism). Common clinical manifestations include feeling cold, depressive mood, dry skin, puffy eyes, constipation, weight gain, slowed heart rate, joint and muscle pain and fatigue. Some but not all people with Hashimoto's thyroiditis have an enlarged gland, also called a goitre. Hashimoto's thyroiditis is more common in women than in men and tends to run in families. Other auto‐immune diseases often occur simultaneously, such as vitiligo, rheumatoid arthritis and diabetes type 1. The disease does not always require treatment, but when it does, it is treated with synthetic thyroid hormone replacement (sometimes desiccated thyroid hormone is used, which is not synthetic). Selenium is an essential trace element that is required in small amounts for correct functioning of the immune system and the thyroid gland.
Four studies at unclear to high risk of bias comprising 463 participants were included. The mean study duration was 7.5 months (range 3 to 18 months). None of the studies addressed our key primary outcome－'health‐related quality of life'. Two of our secondary outcomes－'change from baseline in levothyroxine (i.e. thyroid hormone) replacement dosage at end of the study' and 'economic costs'－were not assessed either. One study at high risk of bias showed a statistically significant improvement in subjective well‐being with sodium selenite 200 μg plus levothyroxine compared with placebo plus levothyroxine (14/18 compared with 3/18, respectively). Selenomethionine 200 μg reduced the serum levels of anti‐thyroid peroxidase antibodies in three studies, and although the changes from baseline were statistically significant, their clinical relevance is unclear. Adverse events were reported in two studies, and selenium supplementation did not lead to more adverse events than were seen with placebo. One adverse event was reported in both studies in the selenomethionine 200 μg plus LT4 arm versus none in the control arm.
In conclusion, the results of these four studies do not provide enough evidence to support the use of selenium in the treatment of Hashimoto's thyroiditis.
Authors' conclusions
Summary of findings
| Selenium (+LT4) compared with placebo (+LT4) for participants with Hashimoto's thyroiditis | ||||||
| Patient or population: participants with Hashimoto's thyroiditis. | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Placebo (+ levothyroxine) | Selenium | |||||
| Change from baseline in health‐related quality of life | See comment | See comment | Not estimable | See comment | See comment | Not reported in any study |
| Change from baseline in assessment of symptoms such as mood, fatigue and muscle weakness | 167 per 1000 | 778 per 1000 | RR 4.67 | 36 | ⊕⊕⊕⊝ | |
| Proportion of participants reporting an adverse event | RR 2.71 | 258 | ⊕⊕⊝⊝ | Participants in placebo group counted twice (same participants in both comparisons) | ||
| Change from baseline in serum levels of anti‐thyroid peroxidase antibodies | See comment | See comment | Not estimable | 252 | ⊕⊕⊝⊝ | Data could not be pooled because of substantial clinical heterogeneity of participants, interventions and controls |
| Change from baseline in LT4 replacement dosage at end of study | See comment | See comment | Not estimable | See comment | See comment | Not reported in any study |
| Economic costs | See comment | See comment | Not estimable | See comment | See comment | Not reported in any study |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aKaranikas 2008 and Turker 2006 included levothyroxine in both treatment arms. Krysiak 2011 included levothyroxine in one arm combined with selenium. | ||||||
Background
Unfamiliar terms are listed in the 'Glossary of terms' (Table 1).
| Term | Explanation |
| Auto‐antigen | Usually a normal protein or complex of proteins (sometimes deoxyribonucleic acid (DNA) or ribonucleic acid (RNA)) that is recognised by the immune system of patients suffering from a specific auto‐immune disease. |
| Antibody | Produced by immune cells, B cells, to identify and neutralise foreign objects such as bacteria and viruses. The antibody recognises a unique part of the foreign target, called an antigen; this might also be an auto‐antigen. |
| Atherosclerosis | A condition in which an artery wall thickens as a result of the accumulation of fatty materials such as cholesterol. |
| Biosynthesis | An enzyme‐catalysed process in cells of living organisms by which substrates are converted to more complex products. |
| Cytokines | Small protein molecules, secreted by several types of cells to stimulate other cells. |
| CD4+ T‐helper cells | A subgroup of T‐cell lymphocytes, a type of white blood cell that plays an important role in the immune system, particularly the adaptive immune system. |
| CD8+ cytotoxic T cells | A subgroup of T‐cell lymphocytes that induce the death of cells infected with viruses (and other pathogens) or otherwise damaged or dysfunctional. |
| Dyslipidaemia | High cholesterol or fat levels in the blood. |
| Goitre | A swelling in the thyroid gland. |
| GPx4 | Phospholipid hydroperoxide glutathione peroxidase, a selenoprotein enzyme. |
| GPx | Glutathione peroxidase, a selenoprotein enzyme that has an antioxidant function. |
| Homeostasis | A state of balanced levels of the molecule in the human body. |
| Hyperglycaemia | High glucose levels in the blood. |
| IFN‐γ | Interferon‐gamma; a cytokine or type II interferon that is critical for innate and adaptive immunity against viral and intracellular bacterial infections and for tumour control. Aberrant IFN‐γ expression is associated with a number of auto‐inflammatory and auto‐immune diseases. |
| LDL | Low‐density lipoproteins or 'bad' cholesterol. |
| Macrophage | A type of immune cell that differentiates from monocytes in tissue and phagocytises (engulfs) foreign materials. |
| Pancreatitis | Inflammation of the pancreas. |
| Rheumatoid arthritis | A chronic, systemic inflammatory disorder that may affect many tissues and organs but principally attacks flexible (synovial) joints. |
| Selenoprotein | Selenium incorporated into proteins. |
| Stroke | A condition of impaired blood supply to the brain resulting in rapid loss of brain function(s). |
| Thyrocyte | Thyroid gland epithelial cells. |
| Vitiligo | An auto‐immune disorder that affects the skin, causing loss of pigment. |
Description of the condition
Hashimoto's thyroiditis (HT) or chronic lymphocytic thyroiditis is a common auto‐immune disorder. HT tends to run in families and affects women and men of all ages, although it is most often seen in middle‐aged women (Fink 2010; Stathatos 2012). Its prevalence is influenced by ethnicity, environmental factors such as iodine and selenium status, age and gender (Fink 2010; Stathatos 2012). Although data on its prevalence are limited at a global level, HT is estimated to affect 1% to 2% of adult women in the US (Hutfless 2011; Staii 2010) and is the most common cause of hypothyroidism in iodine‐sufficient areas of the world. However, subclinical hypothyroidism is more prevalent and occurs in 3% of men and approximately 8% to 10% of women (Chistiakov 2005). Concomitant and other auto‐immune diseases such as rheumatoid arthritis, diabetes mellitus type 1, multiple sclerosis and celiac disease are frequently seen in people suffering from HT (Chistiakov 2005; Stathatos 2012).
The most common presenting symptoms may include anxiety, negative mood, depression, dry skin, cold intolerance, puffy eyes, muscle cramps and fatigue, deep voice, constipation, slow thinking and poor memory (Canaris 2000; Carta 2004). The thyroid gland may enlarge and may present in its classical form as goitre; although thyroid enlargement is usually asymptomatic, a few patients have described thyroid pain and tenderness, sometimes requiring surgical intervention (Li 2011). The other major form of HT, not presenting with goitre, is atrophic auto‐immune thyroiditis, in which fibrosis is more dominant (Bülow Pedersen 2005). Variant forms of the disorder include silent (painless) thyroiditis and postpartum thyroiditis, both of which are transient but may be followed years later by thyroid failure (Lazarus 1996; Pearce 2003). Hypothyroidism due to HT in pregnant women is frequently associated with increased perinatal morbidity, miscarriage, postpartum thyroiditis and impaired neuropsychological development of the infant (Dosiou 2012; Stathatos 2012).
Although hypothyroidism is the characteristic functional abnormality of HT, the inflammatory process early in the course sometimes involves enough apoptosis to cause thyroid follicular disruption and thyroid hormone release, inducing transient hyperthyroidism (Fatourechi 1971). In rare cases, patients may cycle between hypothyroidism and Graves' disease (Kraiem 1992; Takasu 1990). The usual course of HT involves gradual loss of thyroid function. Patients who have mild (subclinical) hypothyroidism show overt hypothyroidism at a rate of approximately 5% per year (Huber 2002). Overt hypothyroidism, once present, is permanent in nearly all cases, except in some children and postpartum women in whom it is often transient.
Specific serum auto‐antibodies such as anti‐thyroid peroxidase antibodies (anti‐TPOAb) and anti‐thyroglobulin antibodies (anti‐TgAb) are characteristic of HT; serum thyroxine (T4) may be normal or low, and thyroid‐stimulating hormone (TSH) concentrations may be normal or high (Li 2011). Histopathologic examination merely shows diffuse lymphocytic infiltration and formation of germinal centres, although fibrosis can also be detected (Li 2011; Stuart 2011).
Clinical manifestations of the disease are defined primarily by low levels of thyroid hormones; therefore patients are treated by hormone replacement therapy, which usually consists of levothyroxine (LT4) (Özen 2011).
Pathogenesis
In HT, thyrocytes are attacked by a variety of cell‐ and antibody‐mediated inflammatory reactions, resulting in low levels of thyroid hormone (Mitchell 2007). Auto‐immunity can develop from the interaction of genetic susceptibility and environmental and endogenous factors (Chistiakov 2005; Saranac 2011; Tomer 2002). Several susceptibility genes have been identified, such as HLA‐DR, CD‐40, CTLA‐4, PTPN‐22 and thyroid‐specific genes (i.e. thyroglobulin and TSH receptor genes) (Saranac 2011; Stathatos 2012). Auto‐antigens, including tissue‐specific membrane receptors, enzymes and hormones, are presented by major histocompatibility complex (MHC) class II antigen‐presenting cells (APCs) to naive T cells and infiltrate the thyroid gland. Environmental factors such as high iodine intake, selenium deficiency and viral infection can increase the likelihood of this infiltration followed by clonal expansion of both T and B lymphocytes in the draining lymph nodes (Chistiakov 2005; Saranac 2011). Activated CD4+ T‐helper cells promote the release of interferon‐gamma (INF‐γ) by CD8+ cytotoxic T cells; this activates macrophages that capture the damaged thyroid cells, resulting in cytokine‐mediated cell death. In addition, auto‐antibodies (anti‐TSH receptor antibodies, anti‐thyroglobulin and anti‐TPOAb) produced by B cells cause antibody‐mediated cell death (Mitchell 2007). The end result consists of a gradual depletion of thyrocytes and replacement by mononuclear cell infiltration and diffuse fibrosis (Mitchell 2007).
Description of the intervention
Selenium is an essential trace element that is required in small amounts for correct functioning of the immune system. The recommended daily intake for adults is 55 μg/day (Hu 2012). It is obtained from natural selenium rich sources such as brazil nuts, organ meat, muscle meat, cereals, shellfish and fish (Rayman 2008). The selenium content of food depends on local soil conditions, which can vary depending on geographical and geological factors (Rayman 2008). The serum selenium concentration is believed to be in the 70 to 130 ng/mL range (Bleys 2008). Selenomethionine and sodium selenite are the two most common oral forms of selenium supplementation that are available in variable dosages (100 and 200 μg/day) and are usually taken for HT (Toulis 2010; Turker 2006).
Adverse effects of the intervention
The upper tolerable intake level of selenium is 400 μg/day (Rayman 2008). Therefore, oral doses of selenium of less than 400 μg/day will not result in serious adverse effects over the short term (Monsen 2000). However, several adverse effects have been recorded with higher doses, resulting in chronic toxicity or selenosis (e.g. gastrointestinal upset, hair loss, white blotchy nails, garlic breath odour, fatigue, irritability, mild nerve damage) (Goldhaber 2003; Rayman 2008). It has also been reported that selenium may increase the likelihood of type 2 diabetes (Stranges 2007). The suggested mechanism is that selenium may suppress the production of insulin‐like growth factor‐1 (i.e. influencing glucose homeostasis). Moreover, selenium in high levels may promote the release of glucagon, resulting in hyperglycaemia (Stranges 2007). Likewise, high selenium blood levels may contribute to dyslipidaemia. The potential mechanism is not fully understood, but it has been proposed that elevated levels of selenium might result in high levels of selenoproteins that regulate cholesterol biosynthesis (Stranges 2010).
How the intervention might work
Recent advances in thyroid cell physiology have illustrated the key role that selenium plays in thyroid gland function (Köhrle 2005). Several enzymes in the thyroid gland are selenoproteins, meaning that selenium is incorporated in their molecular structure (Brown 2001; Köhrle 2005). One of the most vital of these enzymes, glutathione peroxidase (GPx), is involved in protecting the gland against oxidative damage. Hydrogen peroxide (H2O2), a free radical capable of inflicting oxidative damage, is required as substrate by thyroid peroxidase (TPO) for the iodination and coupling of tyrosyl residues in thyroglobulin to produce thyroid hormone. The active form of thyroid hormone, triiodothyronine (T3), is produced by de‐iodination of the prohormone T4 by type I and type II iodothyronine de‐iodinases (IDIs) in a two‐substrate 'ping‐pong' mechanism of reaction, along with degradation of H2O2 to water by GPx. IDIs, such as GPx, are also selenoproteins. If a selenium deficiency exists, these two enzymes cannot function properly, and the end result is ineffective production of T3 and inefficient protection against free radicals, inducing cell damage and auto‐immune destruction of the gland (Brown 2001; Köhrle 2005; Toulis 2010). In these conditions, selenium supplementation may be of benefit to patients with HT (Toulis 2010).
Why it is important to do this review
Several studies have suggested that selenium supplementation in patients with HT reduces antibodies levels (Gärtner 2002), results in a decreased dosage of LT4 and may provide other beneficial effects (e.g. on mood and health‐related quality of life (HRQoL)) (Ott 2011). On the basis of the last date of searches in 2007, one systematic review (Toulis 2010) concluded that high‐level evidence of the benefits of selenium supplementation for periods longer than three months is limited. The review authors also highlighted a lack of "meaningful clinical outcomes" selected and reported in the included trials; therefore at that time, routine selenium supplementation could not be recommended for HT. This previous review is discussed further in the section, 'Agreements and disagreements with other studies or reviews'.
Objectives
To assess the effects of selenium supplementation for Hashimoto's thyroiditis.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled clinical trials.
Types of participants
Adults (18 years of age and older) diagnosed with Hashimoto's thyroiditis.
Diagnostic criteria
As diagnosed by a physician and supported by serum levels of anti‐TPOAb and anti‐TgAb above the normal level of the laboratory's normal range.
Types of interventions
Intervention
-
Selenium 100 µg or 200 µg supplementation (sodium selenite or selenomethionine) alone or combined with titrated LT4 to maintain basal TSH within normal range.
Control
-
No control or no control plus titrated LT4 to maintain basal TSH within the normal range.
-
Placebo tablets or placebo tablets plus titrated LT4 to maintain basal TSH within the normal range.
Types of outcome measures
Primary outcomes
-
Change from baseline in HRQoL assessed using any validated quality‐of‐life instrument at end of study.
-
Change from baseline in symptoms such as mood, fatigue and muscle weakness assessed using any validated instrument at end of study.
-
Proportion of participants reporting an adverse event throughout the study period.
Secondary outcomes
-
Change from baseline in serum levels of anti‐thyroid peroxidase antibodies at end of study.
-
Change from baseline in LT4 replacement dosage at end of study.
-
Economic costs.
Timing of outcome measurement
We considered outcomes measured up to three months (short term), from three to six months (medium term) and after six months (long term).
Summary of findings table
We established a 'summary of findings Table for the main comparison' table using the following outcomes listed according to priority:
-
Change from baseline in HRQoL.
-
Change from baseline in assessment of symptoms such as mood, fatigue and muscle weakness.
-
Proportion of participants reporting an adverse event.
-
Change from baseline in serum levels of anti‐thyroid peroxidase antibodies.
-
Change from baseline in LT4 replacement dosage.
-
Economic costs.
Search methods for identification of studies
Electronic searches
We used the following sources from inception to 2 October 2012 for identification of trials:
-
The Cochrane Library.
-
MEDLINE.
-
EMBASE.
-
Web of Science.
We (EvZ) also searched databases of ongoing trials (ClinicalTrials.gov (www.clinicaltrials.gov/)), the Current Controlled Trials metaRegister (www.controlled‐trials.com/) and the EU Clinical Trials register (www.clinicaltrialsregister.eu/) on 5 November 2012. We have provided information including trial identifiers for recognised studies in the 'Characteristics of ongoing studies' table and the appendix 'Matrix of study endpoints (protocol/trial documents)'. For every included study, we tried to find its protocol in databases of ongoing trials, in publications of study designs or in both.
For detailed search strategies, please see Appendix 1 (searches were not more than six months old at the time the final review draft was checked into the Cochrane Information and Management System for editorial approval). We used PubMed's 'My NCBI' (National Centre for Biotechnology Information) email alert service to identify newly published studies using a basic search strategy (see Appendix 1).
For future updates, if additional key words of relevance are detected during any of the electronic or other searches, we will modify the electronic search strategies to incorporate these terms. We have included studies published in any language.
Searching other resources
We (EvZ) tried to identify other potentially eligible trials or ancillary publications by searching the reference lists of retrieved included trials, (systematic) reviews, meta‐analyses and health technology assessment reports.
Data collection and analysis
Selection of studies
To determine the studies to be assessed further, two review authors (AYA, EvZ) independently scanned the abstract, title or both sections of every record retrieved. We investigated all potentially relevant articles as full text. Where differences in opinion existed, they were resolved by a third party. We present an adapted PRISMA (Preferred Reporting Items for Systematic reviews and Meta‐Analyses) flow chart of study selection (Figure 1) (Liberati 2009).
Study flow diagram.
Data extraction and management
For studies that fulfilled inclusion criteria, three review authors (AYA, EvZ, ZF) independently abstracted relevant population and intervention characteristics using standard data extraction templates (for details see Characteristics of included studies, Table 2; Appendix 2; Appendix 3; Appendix 4; Appendix 5; Appendix 6; Appendix 7; Appendix 8; Appendix 9) with any disagreements resolved by discussion.
| Characteristic | Intervention(s) and comparator(s) | [N] Screened/eligible | [N] Randomised | [N] Safety | [N] ITT | [N] Finishing study | [%] Randomised finishing study | Follow‐upa |
| Karanikas 2008 | I: LT4 + 200 μg sodium selenite | 36 | 18 | ‐ | ‐ | ‐ | ‐ | 3 months |
| C: LT4 + placebo | 18 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| total: | 36 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| Krysiak 2011 | I1: Levothyroxine sodium | ‐ | 42 | ‐ | N/A | 41 | 98 | 6 months |
| I2: Selenomethionine 200 μg | 43 | ‐ | N/A | 42 | 98 | 6 months | ||
| I3: Levothyroxine sodium | 43 | ‐ | N/A | 42 | 98 | 6 months | ||
| C1: Placebo | 42 | ‐ | N/A | 40 | 95 | 6 months | ||
| total: | 170 | 165 | N/A | 165 | 97 | 6 months | ||
| Negro 2007 | I: 200 μg selenomethionine | 2227 | 85 | ‐ | ‐ | 77 | 91 | from 12 weeks' gestation to12 months' post partum |
| C: Placebo | 84 | ‐ | ‐ | 74 | 88 | |||
| total: | 169 | ‐ | ‐ | 151 | 89 | |||
| Turker 2006 | I: LT4 + 200 μg + L‐selenomethionine | ‐ | 48 | ‐ | ‐ | ‐ | ‐ | 3 months |
| C: LT4 + placebo | 40 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| total: | 88 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| Total | All interventions | 279 | ‐ | |||||
| All controls | 184 | ‐ | ||||||
| All interventions and controls | 463 | ‐ | ||||||
aDuration of intervention and/or follow‐up under randomised conditions until end of study.
‐ = not reported.
ITT: intention‐to‐treat; N/A: not applicable.
We sent an email request to the contact persons of included studies for further questions regarding the trials. The results of this survey are published in Appendix 10. Thereafter, we sought relevant missing information on the trial from the primary author(s) of the article.
Dealing with duplicate publications and companion papers
In the case of duplicate publications and companion papers of a primary study, we sought to maximise yield of information by simultaneous evaluation of all available data.
Assessment of risk of bias in included studies
Three review authors (AYA, EvZ, ZF) assessed each trial independently. We resolved possible disagreements by consensus.
We assessed risk of bias using The Cochrane Collaboration's tool (Higgins 2011; Higgins 2011a). We used the following bias criteria.
-
Random sequence generation (selection bias).
-
Allocation concealment (selection bias).
-
Blinding (performance bias and detection bias), separated for blinding of participants and personnel and blinding of outcome assessment.
-
Incomplete outcome data (attrition bias).
-
Selective reporting (reporting bias) ‐ see Appendix 5.
-
Other bias.
We judged risk of bias criteria as 'low risk', 'high risk' or 'unclear risk' and evaluated individual bias items as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We included a 'Risk of bias' graph figure (Figure 2) and a 'Risk of bias' summary figure (Figure 3).
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
We assessed the impact of individual bias domains on study results at endpoint and study levels.
For blinding of participants and personnel (performance bias), detection bias (blinding of outcome assessors) and attrition bias (incomplete outcome data), we evaluated risk of bias separately for subjective and objective outcomes.
We defined the following endpoints as subjective outcomes:
-
Change from baseline in HRQoL.
-
Change from baseline in assessment of mood and fatigue.
-
Proportion of participants reporting an adverse event.
We defined the following outcomes as objective outcomes:
-
Change from baseline in serum levels of anti‐thyroid peroxidase antibodies.
-
Change from baseline in LT4 replacement dosage.
-
Change from baseline in muscle weakness.
-
Economic costs.
Measures of treatment effect
We presented continuous outcomes on the original scale as reported in each individual study. Dichotomous outcomes were presented as risk ratios (RRs) and if significant were converted to the number needed to treat for an additional beneficial outcome (NNTB).
All outcomes data were reported with their associated 95% confidence intervals (CIs) and were analysed using a random‐effects model in RevMan (RevMan 2011) and the Mantel Haenzel test for dichotomous outcome data and invariance analysis for continuous outcome data, unless stated otherwise.
Unit of analysis issues
Cluster‐randomised trials
For future updates, if cluster randomised trials (i.e. groups of individuals randomly assigned to intervention or control) are identified from searches, these will be checked for unit of analysis errors based on the advice provided in Section 16.3.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). If studies are analysed that do not account for clustering, standard errors will be inflated for the effect of clustering and CIs and P values re‐calculated. If this is not possible, study results will be presented only as point estimates without P values or CIs.
Cross‐over trials
Unit of analysis issues can arise in studies in which participants have been randomly assigned to multiple treatments in multiple periods, or where an inadequate wash‐out period has been reported. We assessed the carry‐over and period effects in one study descriptively and analysed these data based on the advice provided in Section 16.4.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Studies with multiple treatment groups
Studies that are reported with multiple treatment groups have the potential for participant data to contribute to multiple comparisons. We assessed the comparisons for clinical importance and included only those that address the primary outcomes. In cases where all comparisons are of equal clinical value, we split the 'shared' group equally into the number of comparisons made, as discussed in Section 16.5.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Dealing with missing data
If data were missing from trials that were less than 10 years old, we tried wherever possible to contact the investigators or sponsors of these studies. We tried to re‐analyse data according to the intention‐to‐treat (ITT) principle whenever possible. For dichotomous outcomes, if authors had conducted a per‐protocol analysis, we carried out an ITT analysis by imputation setting the missing data to reflect treatment failure, checking the degree of imbalance of the drop‐out between arms to determine the potential impact of bias (Section 16.2.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011)). For continuous outcomes, a per‐protocol analysis was carried out in place of an ITT analysis.
In circumstances where partial data were presented in the primary research, we have calculated the change from baseline and associated standard deviation with an assumed correlation coefficient between baseline and follow‐up of 0.75, consistent with the nature of biomarker outcomes. In each case the calculation was repeated with an assumed weaker correlation of 0.5.
Assessment of heterogeneity
We assessed clinical heterogeneity by examining the characteristics of studies, the similarity between types of participants and the interventions. We planned to report heterogeneity as important if it was substantial (I2 between 50% and 90%, Higgins 2011); if the I2 statistic was greater than 90%, the meta‐analysis would not have been carried out. However, if heterogeneity could be explained by clinical reasoning and a coherent argument could be made for combining the studies, we planned to enter these into a meta‐analysis. In cases where the heterogeneity could not be adequately explained, we planned not to pool the data.
Assessment of reporting biases
In future updates, assessments of reporting bias will follow the recommendations on testing for funnel plot asymmetry (Egger 1997), as described in Section 10.4.3.1 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). These assessments will be performed for primary and secondary outcomes for meta‐analysis when a minimum number of studies are included to allow a reasonable estimate of the effect of intervention (nominally nine studies). Funnel plots will be presented only when some evidence of asymmetry is seen in the plots. Possible sources of asymmetry will be explored through an additional sensitivity analysis.
Data synthesis
In future updates, if adequate studies are identified from the searches, these data will be analysed in RevMan (RevMan 2011) and reported in accordance with the advice in Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). A random‐effects meta‐analysis will be carried out in studies that investigate similar interventions and report data that exhibited not more than moderate heterogeneity.
Subgroup analysis and investigation of heterogeneity
If an adequate number of studies had been reported, we planned to carry out subgroup analyses of the following primary outcomes:
-
Age.
-
Selenium status at baseline.
-
Type of selenium (selenomethionine or sodium selenite).
-
Selenium dose (100 or 200 µg/day).
-
Different baseline anti‐TPOAb.
-
Gender.
Sensitivity analysis
In future updates, if adequate numbers of studies are identified, we will perform sensitivity analyses to explore the influence of the following factors on effect sizes:
-
Restricting the analysis to published studies.
-
Restricting the analysis while taking into account risk of bias, as specified earlier.
-
Restricting the analysis to very long or large studies to establish how much they dominate the results.
-
Restricting the analysis to studies using the following filters: diagnostic criteria, language of publication, source of funding (industry vs other) and country.
Results
Description of studies
For a detailed description of studies, see Characteristics of included studies, Characteristics of excluded studies,Characteristics of studies awaiting classification and Characteristics of ongoing studies.
Results of the search
The initial search identified 110 records; from these, nine full text papers were identified for further examination. We excluded the other studies on the basis of their titles or abstracts because they did not meet the inclusion criteria, they were not relevant to the question under study or they presented a duplicate report (see Figure 1 for the amended PRISMA [preferred reporting items for systematic reviews and meta‐analyses] flow chart). After the full text of the selected publications was screened, four studies (four publications) were deemed to meet the inclusion criteria. All studies were published in English. We contacted all authors of included studies and received a reply from two (Karanikas 2008; Krysiak 2011). We sought additional information from the authors of seven studies (Duntas 2003; Gärtner 2002; Gärtner 2003; Karanikas 2008; Krysiak 2011; Nacamulli 2010; Turker 2006). Six authors responded to these requests and provided further data (see Appendix 10).
After the search had been completed, an additional study (Krysiak 2012) was found, which is located in Characteristics of studies awaiting classification. This study has not been added to Figure 1 but will be considered in the next update of this review.
Included studies
A detailed description of the characteristics of included studies is presented elsewhere (see Characteristics of included studies and appendices). A succinct overview follows.
Comparisons
The four studies described different comparisons (see Appendix 2):
-
In Karanikas 2008 the treatment arm received levothyroxine (LT4) combined with 200 μg sodium selenite, while the control arm received LT4 with a placebo.
-
The study of Krysiak 2011 included four arms; one treatment arm with LT4, one with selenomethionine 200 μg, one with LT4 and selenomethionine 200 μg and one placebo arm.
-
In Negro 2007 selenomethionine 200 μg was compared with placebo.
-
Participants in the treatment arm in the study of Turker 2006 received LT4 combined with selenomethionine 200 μg, while the control arm received LT4 plus placebo.
Overview of study populations
A total of 463 participants were included in the four trials; 279 participants were randomised to intervention and 184 to control groups.
An unclear number of participants finished the study in the intervention and control groups because of the fact that only means were reported in two studies, and it was unclear whether all participants were entered into the analysis.
Individual sample size ranged from 36 to 170. For further details, see Table 2.
Study design
Studies were randomised controlled trials. All four trials adopted a parallel‐group superiority design, and all studies used a placebo control (Karanikas 2008; Krysiak 2011; Negro 2007; Turker 2006).
Two trials were multi‐centred (Negro 2007;Turker 2006), both with two centres.
In terms of blinding, one study was double‐blinded for participants and personnel (Krysiak 2011), no studies were single‐blinded for participants and in one study, blinding was not defined (Negro 2007). Outcome assessors were blinded in one study (Krysiak 2011). Investigators in two studies stated that the study was blinded, but no further details were given about the specific measures used to blind personnel and participants from knowledge of which intervention a participant was receiving (Karanikas 2008;Turker 2006).
Studies were performed between the years 2006 and 2011.
The duration of interventions ranged from three to 18 months, with a mean study period of 7.5 months.
No study included a follow‐up period.
None of the studies had a run‐in period.
None of the studies was terminated before regular end.
Settings
All studies were conducted in an outpatient setting in a hospital.
Participants
The participating population consisted of the following: women with auto‐immune thyroiditis (Karanikas 2008;Turker 2006), euthyroid women who had recently been diagnosed with Hashimoto's thyroiditis (Krysiak 2011) and pregnant women with positive anti‐TPO antibodies (Negro 2007).
Four trials included participants from economically developed countries.
Ethnic groups were distributed as follows: Caucasian (Karanikas 2008; Negro 2007); the other two studies did not provide details on ethnicity.
The duration of auto‐immune thyroiditis was not reported in any trial.
Only women were included in all studies (Karanikas 2008; Krysiak 2011; Negro 2007; Turker 2006).
The mean age of participants in the trials ranged from 28 to 47 years.
One trial reported co‐morbidities of participants (Turker 2006), one trial co‐interventions in participants (Negro 2007) and no trials co‐medications used by participants.
Criteria for entry into the individual studies are outlined in the Characteristics of included studies.
Diagnosis
Participants were diagnosed with auto‐immune thyroiditis in all four studies (Karanikas 2008; Krysiak 2011; Negro 2007; Turker 2006).
None of the studies confirmed the diagnosis of auto‐immune thyroiditis against standard diagnostic criteria. All four studies did not refer to standard diagnostic criteria but instead relied on third party diagnosis of auto‐immune thyroiditis before study enrolment.
Interventions
One study reported treatment before the start of the trial (Karanikas 2008) consisting of LT4.
None of the studies had a titration period.
Intervention was applied by the oral route once a day.
The daily dosage of sodium selenite or selenomethionine was 200 μg.
All studies used a matching placebo as the control intervention (Karanikas 2008; Krysiak 2011; Negro 2007; Turker 2006).
The duration of treatment ranged from three to 18 months, with a mean treatment duration of 7.5 months.
Outcomes
All studies explicitly stated a primary endpoint in the publication (Karanikas 2008; Krysiak 2011; Negro 2007; Turker 2006); none of the studies provided secondary endpoints.
Reporting of endpoints
One study assessed subjective well‐being (Karanikas 2008).
Anti‐TPO antibodies were measured from baseline in all studies (Karanikas 2008; Krysiak 2011; Negro 2007; Turker 2006).
Two studies reported on adverse events (Krysiak 2011; Turker 2006).
No studies investigated HRQoL, change from baseline in LT4 replacement dosage at end of study or cost of treatment.
For a summary of all outcomes assessed in each study, see Appendix 7.
Three studies provided a definition of endpoint measurement (Karanikas 2008; Negro 2007; Turker 2006) for the following outcomes: subjective well‐being, anti‐TPO antibody measurement and LT4 replacement.
Excluded studies
Six studies were excluded after careful evaluation of the full text of the publication (Balázs 2008; Contempré 1992; Duntas 2003; Gärtner 2002; Gärtner 2003; Nacamulli 2010) (see Figure 1).
The main reason for exclusion was that these appeared to be controlled clinical trials. Four studies were reported to be randomised, but after e‐mail contact with the investigators, these were classified as quasi‐randomised. For further details, see Characteristics of excluded studies.
Risk of bias in included studies
For details on risk of bias of included studies, see Characteristics of included studies.
For an overview of review authors' judgments about each risk of bias item for individual studies and across all studies, see Figure 2 and Figure 3.
We investigated performance bias, detection bias and attrition bias separately for objective and subjective outcome measures.
We defined 'objective outcome' measures as follows: change from baseline in serum levels of anti‐thyroid peroxidase antibodies, change from baseline in LT4 replacement dosage, change from baseline in muscle weakness and economic costs.
We defined 'subjective outcome' measures as follows: change from baseline in HRQoL, change from baseline in assessment of mood and fatigue and proportions of participants reporting an adverse event.
Allocation
Sequence generation
In two studies (Krysiak 2011; Negro 2007), the method used to generate the allocation sequence was described in sufficient detail; therefore, this domain was judged as low risk of bias for these studies. However, in the two remaining studies (Karanikas 2008; Turker 2006), sequence generation was based on prognostic factors such as serum level of anti‐TPO antibodies and age, and there was no indication that stratified randomisation had been used; accordingly, the domain was judged as at high risk of bias.
Allocation concealment
Reports of two studies (Krysiak 2011; Negro 2007) provided sufficient detail and reassurance that participants and investigators enrolling those participants could not foresee the upcoming assignment. For the other two studies (Karanikas 2008; Turker 2006), the method used to conceal the allocation sequence was not reported; thus, they received a judgment of unclear risk of bias for this domain.
Blinding
Three studies explicitly stated that blinding of participants and personnel was undertaken but did not provide sufficient information about blinding procedures (Karanikas 2008; Krysiak 2011; Turker 2006); the remaining study did not report any blinding (Negro 2007).
Most of the objective outcomes were based on blood tests; however, this is unlikely to have introduced bias into the outcome assessment. We judged this for three studies as having low risk of bias (Karanikas 2008; Krysiak 2011; Turker 2006). One study included thyroid ultrasound as well as an outcome; this can be potentially confounded by prior knowledge of treatment intervention (Negro 2007). Therefore we judged the domain for detection bias here as high risk of bias. Only one study assessed a subjective outcome (Karanikas 2008), but the method used to blind the assessment of subjective outcomes by participants was not described; therefore we judged this as having unclear risk of bias.
Incomplete outcome data
Only one study described a subjective outcome (Karanikas 2008); the other studies included only objective outcomes (Krysiak 2011; Negro 2007; Turker 2006).
Numbers of study withdrawals were described in two studies that had losses to follow‐up (Krysiak 2011; Negro 2007).
Analysis was reported as ITT in one study for the subjective outcome but not for the objective outcomes (Karanikas 2008). No ITT analysis was undertaken in the trials by Krysiak 2011 and Negro 2007.
Two studies did not report losses to follow‐up and reported only means of the outcomes without numbers of participants (Karanikas 2008; Turker 2006).
Selective reporting
The protocol for three of the studies was not available, but the prespecified outcomes and those mentioned in the methods section appeared to have been reported; therefore we judged this domain in these studies as having low risk of bias (Karanikas 2008; Krysiak 2011; Turker 2006). However, the free thyroxine (FT4) values were incompletely reported in Negro 2007, and we judged this as having unclear risk of bias (see also Appendix 5).
Other potential sources of bias
All four studies appeared to be free of other forms of bias, and we judged this domain as having low risk of bias.
Effects of interventions
Baseline characteristics
For details of baseline characteristics, see Appendix 3 and Appendix 4.
(1) Sodium selenite 200 μg plus titrated LT4 versus placebo plus titrated LT4
One study judged as having high risk of bias provided data for this comparison (Karanikas 2008).
Primary outcomes
Change from baseline in HRQoL assessed using any validated quality‐of‐life instrument at end of study
This outcome was not assessed.
Change from baseline in symptoms such as mood, fatigue and muscle weakness assessed using any validated instrument at end of study
Subjective well‐being (assessed with short form health survey) was improved in 14/18 participants receiving sodium selenite compared with 3/18 in the placebo group (RR 4.67, 95% CI 1.61 to 13.50; P = 0.004; number needed to treat (NNT) = 2 (95% CI 2 to 3).
Proportions of participants reporting an adverse event throughout the study period
This outcome was not assessed.
Secondary outcomes
Change from baseline in serum levels of anti‐thyroid peroxidase (TPO) antibodies at end of study
The anti‐TPO antibodies changed from 524 ± 452 IU/mL at baseline to 505 ± 464 IU/mL for the sodium selenite group and from 521 ± 349 IU/mL to 527 ± 354 IU/mL for the placebo group. The mean difference (MD) was estimated to be ‐25 (95% CI ‐181 to 131; P = 0.75; 36 participants).
Change from baseline in LT4 replacement dosage at end of study
This outcome was not assessed.
Economic costs
This outcome was not assessed.
(2) Selenomethionine 200 μg versus placebo
Two studies compared the efficacy of selenomethionine versus placebo (Krysiak 2011; Negro 2007).
Primary outcomes
Change from baseline in HRQoL assessed using any validated quality‐of‐life instrument at end of study
This outcome was not assessed.
Change from baseline in symptoms such as mood, fatigue and muscle weakness assessed using any validated instrument at end of study
This outcome was not assessed.
Proportions of participants reporting an adverse event throughout the study period
No adverse events were reported in either group (Krysiak 2011). This outcome was no assessed in the other study (Negro 2007).
Secondary outcomes
Change from baseline in serum levels of anti‐TPO antibodies at end of study
There was a clearly discernible end of study reduction in anti‐TPO antibody values, as compared to baseline, for participants in the selenomethionine group in both studies. The MDs were estimated as ‐917 IU/mL (Krysiak 2011) and ‐345 IU/mL (Negro 2007), both P < 0.001. Pooling of the studies was not feasible due to marked clinical heterogeneity, which was attributable to variability in the characteristics of the women included in the studies i.e. recently diagnosed euthyroid women not undergoing treatment with high baseline TPO antibodies (Krysiak 2011), versus pregnant women diagnosed with Hashimoto’s and low baseline TPO antibodies (Negro 2007). These results are presented in a forest plot, partitioned into two subgroups (I2 = 99%; P < 0.0001, see Analysis 1.1).This analysis demonstrates a clear reduction in serum levels of anti‐TPO antibodies between selenomethionine (200 μg) and placebo (see Analysis 1.1).
Change from baseline in LT4 replacement dosage at end of study
This outcome was not assessed.
Economic costs
This outcome was not assessed.
(3) Selenomethionine 200 μg plus titrated LT4 versus placebo
The study evaluating comparison (2) also compared the efficacy of selenomethionine plus titrated LT4 versus placebo (Krysiak 2011).
Primary outcomes
Change from baseline in HRQoL assessed using any validated quality‐of‐life instrument at end of study
This outcome was not assessed.
Change from baseline in symptoms such as mood, fatigue and muscle weakness assessed using any validated instrument at end of study
This outcome was not assessed.
Proportions of participants reporting an adverse event throughout the study period
In the active treatment group, 1/43 reported an adverse event versus 0/42 in the placebo group (RR 2.93, 95% CI 0.12 to 70.00).
Secondary outcomes
Change from baseline in serum levels of anti‐TPO antibodies at end of study
Anti‐TPO antibodies changed from 1810 ± 452 U/mL at baseline to 463 ± 104 U/mL at end of study in the group treated with selenomethionine plus titrated LT4 and from 1723 ± 410 IU/L to 1884 ± 346 U/mL in the placebo group. The MD was estimated to be ‐1508 U/mL (95% CI ‐1672 to ‐1345); P < 0.001; 86 participants). This demonstrated a clear reduction in serum levels of anti‐TPO antibodies between selenomethionine (200 μg) plus titrated LT4 and placebo.
Change from baseline in LT4 replacement dosage at end of study
This outcome was not assessed.
Economic costs
This outcome was not assessed.
(4) L‐selenomethionine 200 μg plus titrated LT4 versus placebo plus titrated LT4
This comparison was examined by one study at high risk of bias (Turker 2006).
Primary outcomes
Change from baseline in HRQoL assessed using any validated quality‐of‐life instrument at end of study
This outcome was not assessed.
Change from baseline in symptoms such as mood, fatigue and muscle weakness assessed using any validated instrument at end of study
This outcome was not assessed.
Proportions of participants reporting an adverse event throughout the study period
In the selenomethionine group, 1/48 reported an adverse event (gastric discomfort) versus 0/40 in the placebo group (RR 2.63, 95% CI 0.11 to 62.95).
Secondary outcomes
Change from baseline in serum levels of anti‐TPO antibodies at end of study
Anti‐TPO antibody levels decreased from 804 ± 484 IU/L to 572 ± 517 IU/mL in the selenomethionine group and from 770 ± 406 IU/mL to 773 ± 373 IU/mL in the placebo group. The MD was estimated to be ‐235 IU/mL (95% CI ‐374 to ‐95; P = 0.001; 88 participants); this demonstrated a reduction in serum levels of anti‐TPO antibodies between L‐selenomethionine (200 μg) plus titrated LT4 and placebo plus titrated LT4.
Change from baseline in LT4 replacement dosage at end of study
This outcome was not assessed.
Economic costs
This outcome was not assessed.
Subgroup analyses
We did not perform subgroup analyses because the number of studies was insufficient to allow estimation of effects in various subgroups.
Sensitivity analyses
To assess the impact of estimating the change from baseline correlation as 0.75, we changed this to 0.5 and noted no changes in study findings.
Assessment of reporting biases
Only one study was identified for each comparison; therefore, we were not able to assess reporting bias.
Discussion
Summary of main results
Four studies at unclear to high risk of bias comprising 463 participants were included. None of the studies addressed our principal primary outcome of 'health‐related quality of life' (HRQoL). Two of our secondary outcomes ('change from baseline in LT4 replacement dosage at end of study' and 'economic costs') were not assessed either. One study at high risk of bias showed a statistically significant improvement in subjective well‐being with sodium selenite 200 μg plus titrated levothyroxine (LT4) compared with placebo plus titrated LT4 (Karanikas 2008). Selenomethionine 200 μg supplementation was associated with a reduction in the serum levels of anti‐TPO antibodies in three studies (Krysiak 2011; Negro 2007; Turker 2006), and although the changes from baseline were significant, they were not considered to be clinically important. One study (Karanikas 2008), which assessed sodium selenite 200 μg plus titrated LT4, did not confirm this reduction in serum anti‐thyroid antibodies. Adverse events were reported in two studies, and selenium supplementation did not lead to a statistically significant increase in the number of adverse events when compared with placebo.
For further details, see the 'summary of findings Table for the main comparison'.
Three ongoing studies were identified that may eventually help to fill in some of the gaps in evidence for the efficacy of selenium as a supplement in people with Hashimoto's thyroiditis.
Overall completeness and applicability of evidence
The four studies at unclear to high risk of bias provided very limited data. No clinically relevant conclusions can be drawn on the basis of these four included studies. Hashimoto's thyroiditis has many very debilitating symptoms; therefore, outcomes such as change in HRQoL and improvement in symptoms such as mood, fatigue and muscle weakness are crucial meaningful markers of clinical status. Results of these studies provide incomplete evidence to support or refute the efficacy of selenium in people with Hashimoto's thyroiditis.
Quality of the evidence
Limitations in study design and implementation
Although study design in two of the included studies appeared to have been at best adequate, we judged the sequence generation of the other two studies as having high risk of bias. We were unsuccessful in our attempts to contact the investigators of these last two studies to clarify the methods used to generate the sequence and to conceal the allocation and to obtain details of blinding and losses to follow‐up (see Risk of bias in included studies section and Appendix 10 of this review). Furthermore, our key outcomes such as HRQoL and effects on mood, well‐being and fatigue were not addressed in any of the studies, with the exception of well‐being in one study, which was assessed as having high risk of bias. One of our remaining outcomes reflected changes in anti‐TPO antibodies, which, as long as they remain positive, can be considered to a large extent to be not clinically meaningful.
Indirectness of the evidence
Participants in the included study in general constituted a clinically representative sample matching the inclusion criteria; therefore, we had no significant concerns about the appropriateness of participants identified in the review.
Placebo‐controlled trials are still required to evaluate whether selenium supplementation has any potential beneficial effect on Hashimoto's thyroiditis. The results of these studies provide insufficient evidence to allow any firm conclusions to be drawn to support or refute selenium as additional therapy.
Patient‐relevant outcomes are a pre‐requisite for informing evidence‐based clinical decision making, but the importance of patient‐reported outcomes (PROs), specifically those used in evaluating the impact of the intervention on quality of life, appears to have been underestimated by investigators in all of the included studies.
Inconsistency of the results
In view of the clinical heterogeneity noted between the studies, and, more specifically, the comparisons evaluated, it was not possible to pool study data; and thus no inferences could be drawn about any possible inconsistency in the results.
Imprecision of the results
The primary outcome for this review was assessment of HRQoL, which was not measured in any of the included studies. The results of our secondary outcomes provided varying estimates of anti‐TPO antibody level reduction in each comparison. These effect estimates were generated from single studies that reported large reductions bound by tight confidence intervals. All estimates showed clear reductions, but it should be noted that these were generated from single studies and were subject to increased risk of bias.
Publication bias
Although our attempts to identify additional studies yielded three ongoing studies, the possibility of further unpublished research on this topic cannot be excluded. In future updates, and if additional trials are identified for inclusion, we will assess publication bias as specified in the Assessment of reporting biases section of this review.
Potential biases in the review process
We made every attempt to limit bias in the review process by ensuring a comprehensive search for potentially eligible studies. The authors' independent assessments of eligibility of studies for inclusion in this review minimised the potential for additional bias.
Agreements and disagreements with other studies or reviews
We identified another systematic review that attempted "to summarize available data and provide an evidence‐based recommendation regarding selenium supplementation in the treatment of Hashimoto's thyroiditis" (Toulis 2010). This review included a meta‐analysis of data extracted solely from trials that were 'blinded, randomized, placebo‐controlled in design'. Although this review relied on the consensus process negotiated between investigators and was therefore deemed reasonably transparent, we are in disagreement over the robustness of its methodological approach. Lack of clarity in the process and ultimately its limited reproducibility were illustrated by incomplete reporting of some of the important steps taken in study assessment and handling of missing trial details and data. It appears that no attempts were made to contact any of the investigators in the included studies for clarification of methods used to generate the sequence, allocation concealment or blinding or to retrieve missing data. Furthermore, and quite significantly, no risk of bias assessments of the included studies were undertaken. Of the four studies (Duntas 2003; Gärtner 2002; Karanikas 2008; Turker 2006) included in the meta‐analysis of this review (Toulis 2010), two were excluded in our review because through email contact, the trial investigators confirmed that these were quasi‐randomised (Duntas 2003; Gärtner 2002). In the other two studies, it was unclear whether participants had been randomly assigned according to strata, or whether the studies were also quasi‐randomised (Karanikas 2008; Turker 2006), and we were unsuccessful in our attempts to contact study investigators. The other systematic review included two additional studies (Gärtner 2003; Mazokopakis 2007), which were excluded from our review. We excluded Gärtner 2003 on the basis that email communication revealed that this study appeared to be quasi‐randomised, and although the study design was not an exclusion criterion for the systematic review of Toulis 2010, this study was excluded from the meta‐analysis. Mazokopakis 2007 was not considered eligible for our review as it was clear from the abstract that it was not a randomised controlled trial. Negro 2007 was excluded on the basis of inclusion criteria, which stated that no pregnant women would be included; however, the second phase of this study included women after delivery and could have been included.
Although the authors in Toulis 2010 sought to provide evidence‐based recommendations for selenium supplementation, they failed to indicate how the quality of the evidence was rated, or how the strength of subsequent recommendations was graded.
A recently published non‐systematic review was a valuable resource for increasing our knowledge and giving us a better understanding of the relationship between selenium and thyroid metabolism, the functions of selenium and its role in the different thyroid diseases. It did not include a systematic search of the literature, nor did it provide a critical appraisal of the studies cited as references in support of selenium supplementation for the management of Hashimoto's thyroiditis (Drutel 2013).
Our assessments of the overall quality of the evidence and conclusions on the efficacy of selenium supplementation for Hashimoto's thyroiditis were largely in agreement with the recently updated topic summary in DynaMed, a clinical reference derived from systematic literature surveillance with explicit critical appraisal criteria (DynaMed 2013).
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Selenomethionine versus placebo, Outcome 1 Anti‐TPO antibody levels.
| Core elements | Issues to consider | Status of research for this review |
| Evidence (E) | What is the current state of the evidence? | This systematic review identified one randomised controlled trial (RCT). Incomplete evidence of efficacy and safety of selenium for Hashimoto's thyroiditis. |
| Population (P) | Diagnosis, disease stage, co‐morbidity, risk factors, gender, age, ethnic group, specific inclusion or exclusion criteria, clinical setting | Inclusion criteria:
Exclusion criteria:
|
| Intervention (I) | Type, frequency, dose, duration, prognostic factor | Selenium 100 µg or 200 µg supplementation (sodium selenite or selenomethionine) plus titrated LT4 to maintain basal TSH within normal range for at least 3 months. |
| Comparison (C) | Type, frequency, dose, duration, prognostic factor |
|
| Outcome (O) | Which clinical or patient‐related outcomes will the researcher need to measure, improve, influence or accomplish? Which methods of measurement should be used? | Primary outcomes
Secondary outcomes
|
| Time Stamp (T) | Date of literature search or recommendation | 1 November 2012 |
| Study Type | What is the most appropriate study design to address the proposed question? |
|
| Selenium (+LT4) compared with placebo (+LT4) for participants with Hashimoto's thyroiditis | ||||||
| Patient or population: participants with Hashimoto's thyroiditis. | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Placebo (+ levothyroxine) | Selenium | |||||
| Change from baseline in health‐related quality of life | See comment | See comment | Not estimable | See comment | See comment | Not reported in any study |
| Change from baseline in assessment of symptoms such as mood, fatigue and muscle weakness | 167 per 1000 | 778 per 1000 | RR 4.67 | 36 | ⊕⊕⊕⊝ | |
| Proportion of participants reporting an adverse event | RR 2.71 | 258 | ⊕⊕⊝⊝ | Participants in placebo group counted twice (same participants in both comparisons) | ||
| Change from baseline in serum levels of anti‐thyroid peroxidase antibodies | See comment | See comment | Not estimable | 252 | ⊕⊕⊝⊝ | Data could not be pooled because of substantial clinical heterogeneity of participants, interventions and controls |
| Change from baseline in LT4 replacement dosage at end of study | See comment | See comment | Not estimable | See comment | See comment | Not reported in any study |
| Economic costs | See comment | See comment | Not estimable | See comment | See comment | Not reported in any study |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence: | ||||||
| aKaranikas 2008 and Turker 2006 included levothyroxine in both treatment arms. Krysiak 2011 included levothyroxine in one arm combined with selenium. | ||||||
| Term | Explanation |
| Auto‐antigen | Usually a normal protein or complex of proteins (sometimes deoxyribonucleic acid (DNA) or ribonucleic acid (RNA)) that is recognised by the immune system of patients suffering from a specific auto‐immune disease. |
| Antibody | Produced by immune cells, B cells, to identify and neutralise foreign objects such as bacteria and viruses. The antibody recognises a unique part of the foreign target, called an antigen; this might also be an auto‐antigen. |
| Atherosclerosis | A condition in which an artery wall thickens as a result of the accumulation of fatty materials such as cholesterol. |
| Biosynthesis | An enzyme‐catalysed process in cells of living organisms by which substrates are converted to more complex products. |
| Cytokines | Small protein molecules, secreted by several types of cells to stimulate other cells. |
| CD4+ T‐helper cells | A subgroup of T‐cell lymphocytes, a type of white blood cell that plays an important role in the immune system, particularly the adaptive immune system. |
| CD8+ cytotoxic T cells | A subgroup of T‐cell lymphocytes that induce the death of cells infected with viruses (and other pathogens) or otherwise damaged or dysfunctional. |
| Dyslipidaemia | High cholesterol or fat levels in the blood. |
| Goitre | A swelling in the thyroid gland. |
| GPx4 | Phospholipid hydroperoxide glutathione peroxidase, a selenoprotein enzyme. |
| GPx | Glutathione peroxidase, a selenoprotein enzyme that has an antioxidant function. |
| Homeostasis | A state of balanced levels of the molecule in the human body. |
| Hyperglycaemia | High glucose levels in the blood. |
| IFN‐γ | Interferon‐gamma; a cytokine or type II interferon that is critical for innate and adaptive immunity against viral and intracellular bacterial infections and for tumour control. Aberrant IFN‐γ expression is associated with a number of auto‐inflammatory and auto‐immune diseases. |
| LDL | Low‐density lipoproteins or 'bad' cholesterol. |
| Macrophage | A type of immune cell that differentiates from monocytes in tissue and phagocytises (engulfs) foreign materials. |
| Pancreatitis | Inflammation of the pancreas. |
| Rheumatoid arthritis | A chronic, systemic inflammatory disorder that may affect many tissues and organs but principally attacks flexible (synovial) joints. |
| Selenoprotein | Selenium incorporated into proteins. |
| Stroke | A condition of impaired blood supply to the brain resulting in rapid loss of brain function(s). |
| Thyrocyte | Thyroid gland epithelial cells. |
| Vitiligo | An auto‐immune disorder that affects the skin, causing loss of pigment. |
| Characteristic | Intervention(s) and comparator(s) | [N] Screened/eligible | [N] Randomised | [N] Safety | [N] ITT | [N] Finishing study | [%] Randomised finishing study | Follow‐upa |
| Karanikas 2008 | I: LT4 + 200 μg sodium selenite | 36 | 18 | ‐ | ‐ | ‐ | ‐ | 3 months |
| C: LT4 + placebo | 18 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| total: | 36 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| Krysiak 2011 | I1: Levothyroxine sodium | ‐ | 42 | ‐ | N/A | 41 | 98 | 6 months |
| I2: Selenomethionine 200 μg | 43 | ‐ | N/A | 42 | 98 | 6 months | ||
| I3: Levothyroxine sodium | 43 | ‐ | N/A | 42 | 98 | 6 months | ||
| C1: Placebo | 42 | ‐ | N/A | 40 | 95 | 6 months | ||
| total: | 170 | 165 | N/A | 165 | 97 | 6 months | ||
| Negro 2007 | I: 200 μg selenomethionine | 2227 | 85 | ‐ | ‐ | 77 | 91 | from 12 weeks' gestation to12 months' post partum |
| C: Placebo | 84 | ‐ | ‐ | 74 | 88 | |||
| total: | 169 | ‐ | ‐ | 151 | 89 | |||
| Turker 2006 | I: LT4 + 200 μg + L‐selenomethionine | ‐ | 48 | ‐ | ‐ | ‐ | ‐ | 3 months |
| C: LT4 + placebo | 40 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| total: | 88 | ‐ | ‐ | ‐ | ‐ | 3 months | ||
| Total | All interventions | 279 | ‐ | |||||
| All controls | 184 | ‐ | ||||||
| All interventions and controls | 463 | ‐ | ||||||
| aDuration of intervention and/or follow‐up under randomised conditions until end of study. ‐ = not reported. ITT: intention‐to‐treat; N/A: not applicable. | ||||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Anti‐TPO antibody levels Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 1.1 1700 IU/mL levels at baseline | 1 | 85 | Mean Difference (IV, Random, 95% CI) | ‐917.0 [‐1029.16, ‐804.84] |
| 1.2 600 IU/mL at baseline | 1 | 169 | Mean Difference (IV, Random, 95% CI) | ‐345.0 [‐358.79, ‐331.21] |
