Skip to main content
Clinical and Translational Science logoLink to Clinical and Translational Science
. 2025 Nov 12;18(11):e70409. doi: 10.1111/cts.70409

Single Center, Open‐Label, Randomized Crossover Trial on Drug–Drug Interactions of Levothyroxine/Magnesium‐Citrate and Levothyroxine/Magnesium‐Aspartate in Healthy Subjects—The ThyroMag Trial

Martin Christian Attinger 1,2, Stefanie von Felten 3, Claudia Lourenço Rodrigues 3, Jan Krützfeldt 2,4, Lorenz Risch 5, Jérôme Bonzon 1,2,✉
PMCID: PMC12605969  PMID: 41221788

ABSTRACT

Divalent cations such as calcium and ferrous sulfate interfere with the absorption of levothyroxine due to complexing. To our knowledge, the effects of magnesium on levothyroxine absorption have never been studied. The open‐label cross‐over pharmacokinetic study was conducted in 15 healthy, euthyroid adults. 1 mg of levothyroxine was administered in tablet form alone or co‐administered with either magnesium aspartate or magnesium citrate. Participants received all three treatments, separated by a washout period, but were randomly allocated 1:1:1 to one of three treatment sequences. We measured thyroxine (T4) over a 6‐h period after ingestion. The primary endpoint was the area under the curve (AUC) of thyroxine; secondary endpoints were C max and T max. Coadministration of magnesium aspartate significantly reduced thyroxine AUC by 12% (geometric mean ratio, GMR = 0.88, 95% CI 0.81–0.95, p = 0.002) and coadministration of magnesium citrate reduced thyroxine AUC non‐significantly by 7% compared with levothyroxine alone (GMR = 0.93, 95% CI 0.86–1.01, p = 0.076). C max was significantly reduced by 7% and T max was significantly increased by 17% when magnesium aspartate was co‐administered. The changes in C max and T max were smaller when magnesium citrate was co‐administered. In conclusion, magnesium reduces the absorption of levothyroxine. However, we found smaller effects compared to those already described for other divalent cations, possibly due to the liquid formulation. Hypothyroid patients should nonetheless take levothyroxine separated from magnesium‐containing formulations, especially if TSH levels are desired to be within a narrow range. If taken together, magnesium citrate may be a better option than magnesium aspartate.

Keywords: clinical trial, drug–drug interaction, levothyroxine absorption, magnesium, pharmacokinetics


Study Highlights.

  • What is the current knowledge on the topic?
    • ○
      The intake of levothyroxine with aluminum‐, calcium‐, and iron‐containing medications is not recommended, since complexes can form in the gastrointestinal tract and inhibit the absorption of levothyroxine. This interaction with other multivalent cations was however never proved for magnesium.
  • What question did this study address?
    • ○
      This study addresses the question of whether the concomitant intake of various magnesium preparations alters the absorption of levothyroxine.
  • What does this study add to our knowledge?
    • ○
      There is a pharmacokinetic interaction between levothyroxine and magnesium preparation, most likely due to complexing of the divalent cations with levothyroxine. However, the effect size is small and may not be clinically relevant for most hypothyroid patients. This holds true for the tested magnesium preparations in powdered form, that are dissolved in water before intake. It cannot be extrapolated to tablets or other galenic formulations.
  • How might this change clinical pharmacology or translational science?
    • ○
      Further studies with magnesium tablets instead of powders could clarify whether the AUC reduction is dependent on the galenic formulation. Also, the effect of the galenic form of levothyroxine could be evaluated in another study, e.g., comparing levothyroxine tablets with levothyroxine capsules taken alone or co‐administered with magnesium preparations.

1. Introduction

Hypothyroidism has a prevalence of 0.2%–1.3% in iodine‐sufficient parts of the world, with women being affected more often than men [1]. The prevalence of subclinical hypothyroidism is even higher in some countries [2]. Overt hypothyroidism, and in some cases, subclinical hypothyroidism (e.g., if clinically apparent or in case of infertility), is treated with the substitution of deficient thyroid hormone, usually in the form of levothyroxine sodium.

The Swiss drug label of levothyroxine states that it should not be taken together with aluminum‐ and iron‐containing medications, since complexes can form in the gastrointestinal tract and inhibit the absorption of levothyroxine [3]. This interaction could be shown for other multivalent cations, such as calcium [4], but never for magnesium, which is a divalent cation as well and thus likely to interfere in a similar manner. Magnesium preparations are often prescribed or bought over the counter for treating muscle cramps or as a supplementation in athletes. Furthermore, many multivitamin preparations contain magnesium. This makes the concomitant intake of levothyroxine with magnesium preparation likely, possibly leading to relevant variations in the bioavailability of levothyroxine and fluctuations in TSH levels.

2. Materials and Methods

This study (SNCTP, Swiss National Clinical Trials Portal, No. 000005658, ISRCTN, International Standard Randomized Controlled Trial Number, https://www.isrctn.com/, no. 18075783) was conducted at the University Hospital of Zurich, Zurich, Switzerland, from November 16, 2023 (first patient/first visit), until August 19, 2024 (last patient/last visit). All study procedures were approved by the competent Ethics Committee (Ethics Committee Zurich, no. 2023‐01493, approval October 6, 2023), and by the competent authorities (Swissmedic, no. 701685, approval September 4, 2023). All participants provided oral and written informed consent before study entry.

2.1. Study Design

The randomized clinical trial had an open‐label cross‐over design and was conducted in healthy participants. Each subject received 1 mg levothyroxine (LO) in the form of Euthyrox on one visit, which served as a comparator, 1 mg of levothyroxine with 10 mmol magnesium aspartate (LMA) in the form of Magnesiocard on a second, and 1 mg of levothyroxine with 12.4 mmol magnesium citrate (LMC) in the form of Magnesium Diasporal on a third visit to study the effect on the absorption of levothyroxine. Due to the cross‐over design of the trial, subjects received all treatments. The subjects were block‐randomized to three different treatment sequences in a 1:1:1 fashion with a block size of 3. The randomization list was prepared in advance by the trial statistician, and allocation concealment will be handled via sequentially numbered, opaque and sealed envelopes by a person at the study center who was not involved in the trial. Treatment sequences were chosen to ensure that each treatment appeared once in each position (1st, 2nd, 3rd), ABC, BCA and CAB, A being LO, B being LMC and C being LMA. With the cross‐over design, individual differences in absorption of levothyroxine were accounted for. The wash‐out period of at least 4 weeks was chosen long enough for a complete clearance of levothyroxine between administrations.

The primary outcome is the area under the curve (AUC) of thyroxine. The C max and T max of thyroxine are secondary outcomes. As magnesium salts are expected to reduce the absorption of levothyroxine, it was hypothesized that the two treatments with a magnesium compound would reduce the AUC of thyroxine compared to LO (resulting in a geometric mean ratio, GMR, < 1), but no difference was expected between the combination of LMC vs. LMA (GMR of 1). The null hypothesis was that the three treatments are bioequivalent regarding the AUC of thyroxine.

The trial protocol and statistical analysis plan, de‐identified participant data, statistical code and any other materials can be accessed upon request to the corresponding author.

2.2. Sample Size

The sample size was calculated to show a significant difference in the AUC of thyroxine over 6 h between LMA or LMC compared to LO with a power of 80% at a significance level of 5%. The treatment effect of LMA or LMC vs. LO with regard to the GMR of AUCs of thyroxine over 6 h was assumed to be 0.78, as estimated using data reported by Zamfirescu and Carlson [4] for levothyroxine with calcium‐citrate vs. levothyroxine alone. Total variance (sum of between and within subjects) was calculated based on figure 2 in Zamfirescu and Carlson as 0.093. The intra‐class correlation coefficient was assumed to be 0.50, i.e., equal variance between and within subjects of 0.047 (SD 0.216). We used the sample size formula for a simpler AB/BA design for continuous outcomes, as described in Chapter 4.3.1 of Chow et al. [5] and implemented in the R package TrialSize [6, 7]. The same number of subjects as for an AB/BA design (e.g., to compare levothyroxine with magnesium‐citrate to levothyroxine alone) will be needed for a 3‐period 3‐treatment cross‐over trial, as it just adds a treatment sequence. We calculated a total number of 14 subjects, which we rounded up to 15 in total (to be divisible by 3), 5 per sequence.

2.3. Participants

Healthy female and male volunteers aged 18–65 were eligible to be enrolled in the study. The exclusion criteria were the presence of contraindications to the drugs under study, the need for any kind of drug therapy for the duration of the study, women who were pregnant, breastfeeding or intended to get pregnant during the course of the study, other clinically significant concomitant diseases (especially thyroid dysfunction, cardiovascular disease, arterial hypertension, renal failure and albumin deficiency), inability to follow the procedures of the study, participation in another study with investigational drugs within 30 days preceding and during the study, previous enrolment in the study and enrolment of the investigator, his/her family members, employees and other dependent people. Participants of childbearing potential were asked to practice an acceptable method of birth control. Physical and laboratory examinations as well as an electrocardiogram had to be considered as normal.

2.4. Study Drugs and Doses

Levothyroxine‐sodium is approved in Switzerland for the treatment of hypothyroidism. The component levothyroxine is identical to the endogenously synthesized thyroid hormone levothyroxine (T4). The oral bioavailability is approximately 80% and is reduced to 60% if the hormone is taken together with food. After absorption, levothyroxine is largely protein‐bound (99%), mainly to thyroxin‐binding globulin and thyroxin‐binding prealbumin. It is deiodized to the active metabolite L‐Triiodthyronine (T3). Both T4 and T3 are conjugated to glucuronic acid and sulfate to form inactive compounds. These are excreted via the biliary system. The half‐life of T4 is about 6–7 days [3]. Participants received 1 mg of levothyroxine in the form of 5 tablets of Euthyrox 200 μg with 200 mL of water (in case of LO) after fasting for at least 8 h.

Magnesium citrate is approved in Switzerland for the prophylaxis and treatment of dysrhythmias, neuromuscular hyperexcitability, tetany, muscle cramps (especially of the calves), myositis ossificans, (pre)eclampsia and the state of magnesium deficiency or increased requirement. It is contraindicated in acute or chronic renal failure. The salt dissociates completely in the stomach. The oral bioavailability is about 30%–60%. It is excreted by the kidney. Soft stool is a possible adverse event [8]. Participants received a sachet of Magnesium Diasporal 300 mg corresponding to 301 mg magnesium (or 12.4 mmol magnesium) in powder form, dissolved in 200 mL of water right before intake, to take the 5 tablets of levothyroxine (in case of LMC).

Magnesium aspartate preparations contain magnesium aspartate hydrochloride‐trihydrate. They are approved in Switzerland for the treatment of tachycardic dysrhythmias, tetany, muscle cramps (especially of the calves), (pre)eclampsia and states of magnesium deficiency or increased requirement. A contraindication is renal failure with anuria. Magnesium aspartate hydrochloride‐trihydrate is cleaved to divalent magnesium ions, chloride ions and aspartic acid; the latter is metabolized in the citrate cycle. Magnesium ions are excreted by the kidney. Possible adverse events include soft stool and diarrhea [9]. One sachet of Magnesiocard Gran 10 mmol corresponds to 10 mmol magnesium (or 243.0 mg magnesium) in powder form, dissolved in 200 mL of water right before intake, to take the 5 tablets of levothyroxine (in case of LMA).

2.5. Sampling Schedule

During a first visit, participants received all information necessary for a decision about participation and were granted at least a week to make a decision. During the second visit, participants signed the informed consent form. Eligibility criteria were evaluated, specifically a screening for contraindications of the study drug by taking a history, vital signs, electrocardiogram (ECG), a physical exam and measurement of TSH, fT4, fT3, serum magnesium, potassium, creatinine (to screen for renal failure), total protein, albumin and β‐HCG in women < 60 years of age (to screen for pregnancy). Each of the following visits started with a history, especially regarding adverse events, and taking of vital signs. An indwelling peripheral venous catheter facilitated subsequent blood drawings. Total serum thyroxine was measured at baseline (before drug intake), and then again after drug intake at 30, 60, 120, 240 and 360 min after the baseline measurement (total time 6 h). Total serum thyroxine was measured by an electrochemiluminescence immunoassay (ECLIA) on a Cobas 8000 Instrument (Roche Diagnostics, Rotkreuz, Switzerland). The assays had an imprecision of 4.2% (coefficient of variation, CV) at a level of 106 nmol/L, 4.1% at 147 nmol/L, and 4.2% at 189 nmol/L as obtained by commercially available internal quality control materials.

2.6. Statistical Analysis

The primary outcome, the AUC of thyroxine over 6 h, was calculated using two different types of interpolation: the “spline” method and the “trapezoid” method. For the primary analysis, the AUC calculated using the “spline” method was analyzed as log(AUC) by a linear mixed‐effects model with a random intercept per study participant and treatment (three categories, with levothyroxine alone as the reference category) and the period (1st, 2nd or 3rd treatment applied, 1st period as the reference category) as fixed effects. We estimated the GMR between LMC and LO as well as between LMA and LO together with a 95% confidence interval and corresponding p‐value via Satterthwaite's degrees of freedom method. Likewise, we reported the mean GMR between period 2 and period 1 and between period 3 and period 1. A sensitivity analysis was performed using the AUC values calculated by the “trapezoid” method to evaluate the robustness of the findings from the main analysis. Since bioequivalence studies are often performed in the field, we compared the between‐treatment GMRs to the bioequivalence range of 0.8–1.25. Distributional assumptions underlying the linear mixed‐effects models were assessed by diagnostic plots of the residuals.

The secondary outcomes C max (peak thyroxine) and T max (timing of peak thyroxine) were analyzed as log(C max) and log(T max) in the same way as the primary outcome. p‐values in the main text are reported using the categorization proposed by Bland [10] to describe the strength of evidence against the null hypothesis.

All participants were included in the analysis. One participant had a missing 5th measurement of thyroxine under one of the treatments. Because the AUC could be calculated despite this missing measurement, we did not have a missing primary outcome measurement. It is, however, uncertain whether the missing measurement would have been the maximum measurement of thyroxine and thus a relevant one for C max and T max. Nevertheless, we did not set these secondary outcomes to missing for this participant under the treatment where one measurement was missing, but rather used C max and T max based on the available five measurements. For details on this participant see Supporting Information and Figure 1.

FIGURE 1.

FIGURE 1

Individual thyroxine (T4) levels over 6 h period for each study participant stratified by the different treatments. The blue triangles depict the means at each time and the blue curve is a smoothed curve shown for illustrative purposes only. Levothyroxine taken alone is absorbed rapidly, leading to a T max at 120 min. In contrast, T max is higher when levothyroxine is taken together with either of the magnesium preparations.

The analysis was conducted with R version 4.4.1 [7] the mixed‐effects models were fitted using the R package lme4 [11].

3. Results

3.1. Descriptive Statistics

Table 1 shows the baseline characteristics, including age and sex, of the trial participants, overall (left column) and stratified by the treatment sequence they received. A total of 15 study participants were included, each randomly assigned to one of three specific treatment sequences. Each treatment sequence included 5 study participants. In the first sequence, LO‐LMC‐LMA, study participants first received LO, followed by LMC, and finally LMA. In the second sequence, LMA‐LO‐LMC, study participants received LMA first, followed by LO, and then LMC. In the third sequence, LMC‐LMA‐LO, study participants first received LMC, followed by LMA, and then LO. There are no missing data for the baseline characteristics (see Table 1).

TABLE 1.

Baseline characteristics age, sex and baseline serum thyroxine (T4) of the first treatment period (P1) for all participants included in the crossover trial overall and stratified by treatment sequence.

Variable Level Overall LO‐LMC‐LMA LMA‐LO‐LMC LMC‐LMA‐LO
n 15 5 5 5
Age (mean, SD) 24.40 (2.56) 24.60 (1.52) 23.80 (3.27) 24.80 (3.03)
Sex (%) Male 5 (33.3) 3 (60.0) 1 (20.0) 1 (20.0)
Female 10 (66.7) 2 (40.0) 4 (80.0) 4 (80.0)
Baseline T4 P1 (median [IQR]) 80.20 [73.65, 86.90] 86.80 [74.20, 88.20] 80.60 [80.20, 85.50] 73.50 [70.20, 73.80]

On average, the study participants included in the analysis were 24.4 years old. 33.3% were male and 66.7% female. The proportion of male and female study participants varied across the different treatment sequences. In the LO‐LMC‐LMA sequence, the distribution of male and female study participants was fairly balanced. However, in the LMA‐LO‐LMC and LMC‐LMA‐LO sequences, there was a predominance of female study participants, with 80% female and only 20% male study participants. Levels of baseline serum thyroxine of the first period (baseline_T4_P1) also differed slightly between sequences, with the highest median baseline T4 in the LO‐LMC‐LMA sequence (86.8 nmol/L) and the lowest in LMC‐LMA‐LO (73.5 nmol/L).

Baseline T4 levels overall and stratified by treatment period (1st, 2nd or 3rd treatment applied) are presented in Table 2. Each of the 15 study participants was treated and had a baseline measurement in all three periods. The overall median baseline T4 concentration was 80.6 nmol/L. When stratified by period, the median T4 concentration remained similar, with only small differences observed across the three periods. Period 3 had the highest median baseline T4, while Period 2 had the lowest. Overall, there were no missing values for baseline T4 levels across the periods.

TABLE 2.

Baseline serum thyroxine (T4) levels stratified by period.

Variable Overall Period 1 Period 2 Period 3 Missing
n 45 15 15 15
Baseline T4 (median [IQR]) 80.60 [74.20, 90.00] 80.20 [73.65, 86.90] 78.50 [74.45, 90.80] 83.30 [76.20, 92.00] 0

3.2. Primary Outcome

The GMRs that were estimated using the primary outcome (AUC of serum thyroxine) calculated by the “spline” method are presented in Table 3. There was no evidence (p > 0.1) of a difference in the geometric mean AUC between period 1 and 2, and only weak evidence (0.05 < p < 0.1) of a difference between period 1 and 3. In terms of treatment effects, there was strong evidence (0.001 < p < 0.01) that treatment LMA is associated with a reduction in AUC. Specifically, treatment LMA resulted in a 12% (95% CI 5%–19%) decrease in AUC compared to treatment LO. Additionally, there was weak evidence (0.05 < p < 0.1) that treatment LMC is associated with a 7% decrease compared to treatment LO. When comparing the between treatment GMRs to the bioequivalence range of 0.8–1.25, none of these treatment GMRs exceeded these limits. Therefore, all treatments may be considered bioequivalent to treatment LO, despite the differences described above.

TABLE 3.

Geometric mean ratios (GMRs) estimated by the linear mixed‐effects model with log(AUC) as the outcome variable.

Coefficient Lower 95% CI Upper 95% CI p
(Intercept) 658.88 604.13 718.6 < 0.0001
Period 2 vs. Period 1 1.04 0.96 1.13 0.31
Period 3 vs. Period 1 1.07 0.99 1.16 0.082
Levothyroxin + Magnesium Aspartat (LMA) vs. Levothyroxine alone (LO) 0.88 0.81 0.95 0.002
Levothyroxin + Magnesium Citrat (LMC) vs. Levothyroxine alone (LO) 0.93 0.86 1.01 0.076

The AUCs of thyroxine are visualized in Figure 2. Compared to LO, both treatments, LMA and LMC, showed lower mean AUC values. The lowest mean AUC was found for LMA. Under LMA, the majority of AUCs were decreased compared to LO. In contrast, the AUC under LMC was more variable, with some study participants showing an increase and others showing a decrease relative to LO or LMA.

FIGURE 2.

FIGURE 2

AUC values (based on spline method) for each treatment across all study participants. Individual participant trends are visualized with connected points and colored by treatment sequence. Crossbars indicate the mean AUC for each treatment.

The results of the sensitivity analysis using the AUC values calculated by the “trapezoid” method are presented in Table 4. These results are nearly identical to those obtained with the “spline” method (Table 3), showing that both interpolation methods (spline and trapezoid) lead to the same conclusions and highlight the robustness of the findings from the primary analysis.

TABLE 4.

Results of linear mixed‐effects model with log(AUC) as outcome variable.

Coefficient Lower 95% CI Upper 95% CI p
(Intercept) 658.99 604.7 718.17 < 0.0001
Period 2 vs. Period 1 1.04 0.96 1.12 0.32
Period 3 vs. Period 1 1.07 0.99 1.16 0.086
Levothyroxin + Magnesium Aspartat (LMA) vs. Levothyroxine alone (LO) 0.88 0.81 0.95 0.001
Levothyroxin + Magnesium Citrat (LMC) vs. Levothyroxine alone (LO) 0.93 0.86 1.01 0.077

3.3. Secondary Outcomes

The results of the analysis using log(C max) as the outcome variable are presented in Table 5. The analysis provided moderate evidence (0.01 < p < 0.05) that treatment LMA is associated with a reduction in C max. Specifically, treatment LMA resulted in a 7% decrease in C max compared to treatment LO. In contrast, there was no evidence (p > 0.1) that treatment LMC is associated with a reduced C max compared to LO. Similarly, the period effects for log(C max) showed no evidence (p > 0.1) of differences between periods 1 and 2 or periods 1 and 3.

TABLE 5.

Secondary outcomes: Results of linear mixed‐effects model with log(C max) and log(T max) as outcome variables.

log(C max) as outcome variable log(T max) as outcome variable
Coefficient Lower 95% CI Upper 95% CI p Coefficient Lower 95% CI Upper 95% CI p
(Intercept) 165.78 154.01 178.45 < 0.0001 4.27 3.93 4.64 < 0.0001
Period 2 vs. Period 1 1.04 0.97 1.1 0.25 0.99 0.91 1.07 0.82
Period 3 vs. Period 1 1.05 0.99 1.12 0.13 0.93 0.85 1.01 0.066
Levothyroxin + Magnesium Aspartat (LMA) vs. Levothyroxine alone (LO) 0.93 0.88 0.99 0.03 1.17 1.07 1.26 0.0005
Levothyroxin + Magnesium Citrat (LMC) + Levothyroxine alone (LO) 0.99 0.93 1.05 0.65 1.08 1 1.17 0.057

The results of the analysis using log(T max) as the outcome variable are also presented in Table 5. The analysis provided very strong evidence (p < 0.001) that treatment LMA is associated with an increase in T max. Specifically, treatment LMA resulted in a 17% increase in T max compared to treatment LO. Similarly, there was strong evidence (0.001 < p < 0.01) that treatment LMC is associated with an 8% increase in T max compared to LO. There is no evidence (p > 0.1) of a difference in T max between period 1 and 2, and only weak evidence (0.05 < p < 0.1) of a difference between period 1 and 3.

3.4. Tolerability

There were no adverse events detected during the study. Vital signs at baseline taken at every visit showed no clinically relevant changes.

4. Discussion

Levothyroxine is usually taken in the morning at least 30 min before the first meal [3] due to the negative effects of food on its absorption. The fasting state reduces variability in serum TSH concentrations. This is especially important when a narrow target range of TSH is desired to avoid iatrogenic subclinical thyroid disease [12]. Besides food, various medications have been shown to influence levothyroxine absorption, for example phosphate binders (e.g., sevelamer and lanthanum carbonate), basic antacids (e.g., aluminum hydroxide), bile acid sequestrants (e.g., cholestyramine), orlistat, simethicone and sucralfate [13, 14, 15, 16]. The proposed mechanism for these interactions is complexing [16].

Also, for the proton pump inhibitors, an effect on TSH levels has been described [17]. However, the mechanism is not complex, but either an alkalization of the gastric environment that impairs complete dissolution of the tablet, or an induction of UGT enzymes that increases the biliary clearance of thyroxine [16]. Alkalization of the gastric environment is a relevant factor when tablets need to be dissolved in the gastric contents. The raised pH by magnesium salts could be a contributing factor to the impaired absorption.

Further, a relevant pharmacokinetic interaction between levothyroxine and divalent cations through complexing had been previously shown for ferrous sulfate [18] and calcium [19]. Zamfirescu et al. described a reduction of levothyroxine absorption by about 20%–25% when taken together with calcium carbonate, calcium citrate or calcium acetate [4].

In this study, the pharmacokinetic interaction between levothyroxine and divalent cations could be extended to magnesium. However, the effect size we showed was smaller than previously described for other cations. We demonstrated an AUC reduction of 12% for levothyroxine co‐administered with magnesium aspartate and 7% for levothyroxine co‐administered with magnesium citrate, the latter not being statistically significant. One possible reason could be the smaller diameter of the magnesium ion in comparison with the calcium and iron ions, leading to less complexing. The other, and most likely more relevant factor is the galenic formulation of the magnesium preparations. A powder dissolved in water is more likely to pass the stomach along the small curvature quickly, thus minimizing the interaction potential with levothyroxine tablets.

This study has some limitations. Although every effort was made to minimize bias, it cannot be ruled out that individual differences in intestinal function affecting the absorption of levothyroxine may exist within the same subject between application appointments. Also, the trial was conducted in healthy volunteers without hypothyroidism, which could also affect the efficiency of absorption. It has been shown that the bioavailability of levothyroxine could be slightly higher in hypothyroid patients compared to euthyroid healthy volunteers [20, 21, 22]. Therefore, the AUC reduction caused by divalent cations might also differ. Although the AUC reduction was still formally within the bioequivalence interval, the clinical relevance of our results cannot be ruled out. Particularly in patients with a low dose requirement of levothyroxine, a reduction in absorption in the size of 12% could possibly lead to relevant TSH fluctuations and even clinical symptoms. Further studies with clinical endpoints would be necessary to investigate this further.

Hypothyroid patients should nonetheless be advised to take levothyroxine separated from magnesium‐containing formulations, especially if TSH is desired to be within a narrow range. The effect size of AUC reduction by magnesium might be higher in patients with overt hypothyroidism and physiologic T4 supplementation. If taken concomitantly, magnesium citrate seems to be a better option than magnesium aspartate, since it demonstrated less interaction potential.

Author Contributions

M.C.A., S.F., C.L.R., J.K., L.R., and J.B. wrote the manuscript. M.C.A., S.F., and J.B. designed the research. M.C.A, and J.B. performed the research; S.F. and C.L.R. analyzed the data.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: cts70409‐sup‐0001‐supinfo.docx.

CTS-18-e70409-s001.docx (330.8KB, docx)

Acknowledgments

We thank all the participants who took part in this study and thus helped realize this project. Open access publishing facilitated by Universitat Zurich, as part of the Wiley ‐ Universitat Zurich agreement via the Consortium Of Swiss Academic Libraries.

Attinger M. C., von Felten S., Rodrigues C. L., Krützfeldt J., Risch L., and Bonzon J., “Single Center, Open‐Label, Randomized Crossover Trial on Drug–Drug Interactions of Levothyroxine/Magnesium‐Citrate and Levothyroxine/Magnesium‐Aspartate in Healthy Subjects—The ThyroMag Trial,” Clinical and Translational Science 18, no. 11 (2025): e70409, 10.1111/cts.70409.

Funding: The authors received no specific funding for this work.

Previous Presentation: The results were presented as a poster with an oral presentation by Martin Christian Attinger at the spring meeting of the Swiss Society of General Internal Medicine in Basel, 2025, May 22.

References

  • 1. Taylor P. N., Albrecht D., Scholz A., et al., “Global Epidemiology of Hyperthyroidism and Hypothyroidism,” Nature Reviews. Endocrinology 14, no. 5 (2018): 301–316. [DOI] [PubMed] [Google Scholar]
  • 2. Valdes S., Maldonado‐Araque C., Lago‐Sampedro A., et al., “Population‐Based National Prevalence of Thyroid Dysfunction in Spain and Associated Factors: di@bet.Es Study,” Thyroid 27, no. 2 (2017): 156–166. [DOI] [PubMed] [Google Scholar]
  • 3. Merck‐Schweiz‐AG , “Euthyrox Drug Label, 2023”.
  • 4. Zamfirescu I. and Carlson H. E., “Absorption of Levothyroxine When Coadministered With Various Calcium Formulations,” Thyroid 21, no. 5 (2011): 483–486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Chow S. C., Shao J., Wang H., and Lokhnygina Y., Sample Size Calculations in Clinical Research, vol. 20, 3rd ed. CRC Biostatistics Series (Chapman and Hall/CRC, 2018). [Google Scholar]
  • 6. Zhang L. J., Wang Z. L., Wang L., Cui L., Sokolove J., and Chan I., “A Simple Approach to Incorporating Historical Control Data in Clinical Trial Design and Analysis,” Statistics in Biosciences 14, no. 2 (2022): 216–236. [Google Scholar]
  • 7. R Core Team , R: A Language and Environment for Statistical Computing [Internet] (R Foundation for Statistical Computing, 2025), https://www.R‐project.org/. [Google Scholar]
  • 8. Doetsch‐Grether‐AG , “Magnesium Diasporal Drug Label,” 2017.
  • 9. Biomed‐AG , “Magnesiocard Drug Label,” 2023.
  • 10. Bland M., An Introduction to Medical Statistics (Oxford Academic, 2015). [Google Scholar]
  • 11. Bates D., Mächler M., Bolker B. M., and Walker S. C., “Fitting Linear Mixed‐Effects Models Using lme4,” Journal of Statistical Software 67, no. 1 (2015): 1–48. [Google Scholar]
  • 12. Bach‐Huynh T. G., Nayak B., Loh J., Soldin S., and Jonklaas J., “Timing of Levothyroxine Administration Affects Serum Thyrotropin Concentration,” Journal of Clinical Endocrinology and Metabolism 94, no. 10 (2009): 3905–3912. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Siraj E. S., Gupta M. K., and Reddy S. S., “Raloxifene Causing Malabsorption of Levothyroxine,” Archives of Internal Medicine 163, no. 11 (2003): 1367–1370. [DOI] [PubMed] [Google Scholar]
  • 14. John‐Kalarickal J., Pearlman G., and Carlson H. E., “New Medications Which Decrease Levothyroxine Absorption,” Thyroid 17, no. 8 (2007): 763–765. [DOI] [PubMed] [Google Scholar]
  • 15. Sherman S. I., Tielens E. T., and Ladenson P. W., “Sucralfate Causes Malabsorption of L‐Thyroxine,” American Journal of Medicine 96, no. 6 (1994): 531–535. [DOI] [PubMed] [Google Scholar]
  • 16. Liu H., Lu M., Hu J., et al., “Medications and Food Interfering With the Bioavailability of Levothyroxine: A Systematic Review,” Therapeutics and Clinical Risk Management 19 (2023): 503–523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Sachmechi I., Reich D. M., Aninyei M., Wibowo F., Gupta G., and Kim P. J., “Effect of Proton Pump Inhibitors on Serum Thyroid‐Stimulating Hormone Level in Euthyroid Patients Treated With Levothyroxine for Hypothyroidism,” Endocrine Practice 13, no. 4 (2007): 345–349. [DOI] [PubMed] [Google Scholar]
  • 18. Campbell N. R., Hasinoff B. B., Stalts H., Rao B., and Wong N. C., “Ferrous Sulfate Reduces Thyroxine Efficacy in Patients With Hypothyroidism,” Annals of Internal Medicine 117, no. 12 (1992): 1010–1013. [DOI] [PubMed] [Google Scholar]
  • 19. Singh N., Singh P. N., and Hershman J. M., “Effect of Calcium Carbonate on the Absorption of Levothyroxine,” Journal of the American Medical Association 283, no. 21 (2000): 2822–2825. [DOI] [PubMed] [Google Scholar]
  • 20. Colucci P., Yue C. S., Ducharme M., and Benvenga S., “A Review of the Pharmacokinetics of Levothyroxine for the Treatment of Hypothyroidism,” European Endocrinology 9, no. 1 (2013): 40–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Read D. G., Hays M. T., and Hershman J. M., “Absorption of Oral Thyroxine in Hypothyroid and Normal Man,” Journal of Clinical Endocrinology and Metabolism 30, no. 6 (1970): 798–799. [DOI] [PubMed] [Google Scholar]
  • 22. Hasselström K., Siersbaek‐Nielsen K., Lumholtz I. B., Faber J., Kirkegaard C., and Friis T., “The Bioavailability of Thyroxine and 3,5,3′‐Triiodothyronine in Normal Subjects and in Hyper‐ and Hypothyroid Patients,” Acta Endocrinologica 110, no. 4 (1985): 483–486. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1: cts70409‐sup‐0001‐supinfo.docx.

CTS-18-e70409-s001.docx (330.8KB, docx)

Articles from Clinical and Translational Science are provided here courtesy of Wiley

RESOURCES