The effect of time-restricted eating accompanied by weight loss on thyroid function in adults with type 2 diabetes

Abstract

Context

Time-restricted eating (TRE) is a lifestyle intervention associated with beneficial cardiometabolic effects. The impact of TRE-induced weight reduction on thyroid hormonal responses is not well elucidated.

Objective

To evaluate the effects of TRE accompanied by weight loss on thyroid function in adults with obesity and type 2 diabetes (T2DM).

Methods

We performed an exploratory analysis of a randomized, crossover, open-label study in adults with obesity and T2DM who underwent 2 dietary interventions: TRE and standard lifestyle. Each intervention period lasted 6 weeks. Thyroid function was assessed by TSH, free T4 (fT4), free T3 (fT3), and reverse T3 (rT3).

Results

Thirty-nine participants without known thyroid disease (age 56.3 years, 35.9% male, body mass index 32.4 ± 5.7 kg/m²) were included. TRE induced significant weight loss (−2.93 kg) compared to the standard lifestyle period (0.59 kg) (mean differences between intervention periods of −3.50 kg ± 2.9; P < .001). Despite this robust weight-lowering effect, there were no significant differences in TSH (0.49 mUI/L ± 3.0; P = .32), fT3 (−0.57 pmol/L ± 2.7; P = .19), fT4 (fT4 −0.55 pmol/L ± 3.8; P = .38), and rT3 (−0.94 ng/dL ± 5.9; P = .350) between intervention periods. Reinforcing these findings, there was no treatment effect interaction on any thyroid function test evaluated. In addition, there was no significant change in the fT3: fT4 and fT3:rT3 ratios between treatment interventions.

Conclusion

Despite inducing significant weight loss, TRE did not exert a significant effect on thyroid function in individuals with obesity and early T2DM.

Lifestyle modification is an important pillar in the management of diabetes and obesity. Time-restricted eating (TRE), a dietary intervention that limits food consumption to a consistent daily eating window with no explicit prescription of caloric intake or quality of food, has become increasingly popular (1-4). Clinical studies have demonstrated that TRE reduces energy intake of up to 30%, even in the absence of intentional caloric restriction (5, 6), and emerging evidence suggests that it may improve cardiometabolic health in the overall population (1-4), and pancreatic β-cell function and insulin resistance in patient with type 2 diabetes (T2DM) (7). Although a positive effect on oxidative stress and inflammatory markers has been suggested as potential mechanisms for the metabolic benefits of TRE (1-4), there are many questions remaining on the physiologic impact of intermittent fasting on diverse metabolic pathways.

One important aspect of energy balance regulation focuses on the interplay between obesity, weight loss, and thyroid function. Thyroid hormones (TH), including T4 and T3, regulate essential metabolic pathways involved in energy balance regulation (8). Although the influence of overt thyroid dysfunction (ie, hyperthyroidism and hypothyroidism) on body weight has been studied more extensively, the impact of weight changes on TH homeostasis is less clear. In a recent meta-analysis of 19 studies evaluating the impact of weight loss interventions on thyroid function, it was demonstrated that weight reduction leads to a significant reduction in free T3 (fT3) and TSH levels and increase in free T4 (fT4) (9). Conversely, in another meta-analysis evaluating the impact of Ramadan fasting in 1107 healthy adults, no significant changes in TH were observed (10). Similarly, a randomized controlled trial (RCT) comparing alternate-day fasting with daily caloric restriction in adults with subclinical hypothyroidism found that both dietary interventions produced similar weight reduction while TSH and fT4 levels remained unchanged (11). Given the heterogeneity of findings in the existing literature on this topic and the importance of elucidating hormonal changes associated with energy balance regulation following weight loss, further studies are warranted to understand the impact of TRE on TH. In this current report, we evaluated the effect of TRE accompanied by weight loss on thyroid function in adults with T2DM as an exploratory analysis of a crossover RCT. We hypothesized that TRE-induced weight loss could increase fT4 and rT3 while reducing fT3 and TSH as consequence of an energy-sparing state following negative caloric balance.

Material and methods

This is an exploratory analysis of a randomized, crossover, open-label study conducted at Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, Toronto, and approved by the Institution's Ethics Board (clinicaltrials.gov NCT05717127). The study protocol and main outcomes have been described in detail previously (6, 7).

As previously described (7), participants with previously diagnosed T2DM, body mass index (BMI) ≥ 25 kg/m², and hemoglobin A1c between 5.5% and 9.0% were randomized to the initial intervention to follow either a TRE intervention (4 hours of eating window between 4 to 8 Pm or 5 to 9 Pm with 20 hours of daily fasting) or a standard eating schedule (no specific recommendation regarding hours of fasting) for a period of 6 weeks (Table 1). At baseline visit, all participants received instruction on healthy lifestyle recommendations and were encouraged to follow the standard healthy eating pattern for the duration of the trial (7). Exclusion criteria included treatment with insulin, glucagon-like peptide-1 receptor agonists, sodium-glucose cotransporter 2 inhibitor, and/or sulfonylureas. Although thyroid disease was not an exclusion criterion, participants included in this analysis did not have known thyroid disease. Each intervention period was 6 weeks followed by an approximately 4-week washout period before crossing over to the other arm. During the fasting intervention, participants were allowed to drink water, tea, and coffee without any sugar added. As previously described in detail (6, 7), we have ensured compliance to the study protocol using a self-administered dietary tool. The entire study protocol consisted of 20 weeks in duration.

Table 1.

Baseline characteristics of study participants at randomization

	Total (n = 39)
Demographic
Male sex—n (%)	14 (35.9)
Female sex—n (%)	25 (64.1)
Premenopausal	2 (5.1)
Postmenopausal	23 (58.9)
Age (years)	56.3 ± 9.4
Ethnicity—n (%)
White	17 (43.6)
Asian	13 (33.3)
Black	2 (5.1)
Mixed and other	7 (17.9)
Hispanic—n (%)	2 (5.1)
Duration of diabetes (years)	2.89 ± 2.41
Hemoglobin A1c (%)	6.6 ± 0.7
Fasting insulin (pmol/L)	125 (83-165)
Treatment of diabetes—n (%)
Diet alone	12 (30.8)
Metformin alone	24 (61.5)
Metformin + dipeptidyl peptidase-4 inhibitor	3 (7.7)
Current smoking—n (%)	1 (2.6)
Alcohol intake (g/day)	3.1 ± 11.6
Hypertension—n (%)	17 (43.6)
Chronic kidney disease—n (%)	0 (0)
Sleep apnea—n (%)	7 (17.9)
Habitual physical activity scorea	9 (8.4-9.34)
Anthropometrics
Body weight (kg)	89.0 ± 18.3
Body mass index (kg/m2)	32.4 ± 5.7
Resting basal metabolic rate (kcal)	1787 ± 442
Thyroid function
TSH (mIU/L) (reference range 0.4-4.00)	1.7 (1.4-2.4)
Free T4 (pmol/L) (reference range 12-21)	15.3 ± 2.4
Free T3 (pmol/L) (reference range 3.1-6.8)	4.7 ± 0.5
Reverse T3 (ng/dL) (reference range 8-25)	18.0 ± 4.9

Data are mean ± SD, if normally distributed; or median (25th-75th), if skewed.

^a Assessed using the Baecke Physical Activity Questionnaire (7).

At each study visit at baseline, 6 weeks, 10 weeks, 16 weeks, and 20 weeks, participants underwent metabolic characterization, anthropometrics, and had their thyroid function assessed (Fig. 1). Blood samples were collected in the morning after an overnight fast. Free T4, fT3, and TSH were quantified by immunoassay (fT4: Roche Cobas Pro e801 AB_3095310, analytical range 0.5-100 pmol/L; fT3 Roche Cobas Pro e801 AB_3741142, analytical range 0.6-50 pmol/L; TSH: Roche Cobas Pro e801 AB_3095311, analytical range 0.005-100 mIU/L). Reverse T3 (rT3) was assessed by liquid chromatography-tandem mass spectrometry (Quest Diagnostics, reference range 8-25 ng/dL). The basal energy expenditure of participants was assessed by using MedGem indirect calorie meter device that is a clinically validated and Food and Drug Administration 510K-cleared, class II, medical device for measurement of resting metabolic rate. The MedGem device measures oxygen consumption using a single-use disposable mouthpiece and a nose clip for collection of expired air to determine resting metabolic rate (12, 13).

Figure 1.

Flow diagram showing study design.

Statistical analyses

The effect of TRE on thyroid function was evaluated by the comparison of changes in thyroid hormones (TSH, fT4, fT3, and rT3) between treatment intervention periods by paired Student t-test (Table 2). In addition, to assess whether TRE impacts thyroid function as compared to standard lifestyle intervention (treatment interaction effect), the trajectories of the thyroid function tests were evaluated by general linear model (within-subjects factor) (Fig. 2). Finally, as indirect measures of the activity of thyroid deiodinases and peripheral conversion of fT4 to fT3, we analyzed differences in fT3:fT4 ratio and fT3:rT3 ratio between the study intervention periods (Fig. 3) (14, 15). Statistical analyses were conducted with The Statistical Package for the Social Sciences software version 23.0 (SPSS Inc., Chicago, IL, USA). Two-tailed P values < .05 were considered statistically significant.

Table 2.

Changes between baseline and intervention endpoint for TRE and standard lifestyle respectively, and comparison of these changes (TRE minus standard lifestyle)

	TRE (n = 39)	Standard lifestyle (n = 39)	TRE minus standard lifestyle (n = 39)	P
Body weight and metabolism
Weight (kg)	−2.93 ± 2.3	0.59 ± 1.7	−3.50 ± 2.9	<.001
BMI (kg/m2)	−1.08 ± 0.86	0.22 ± 0.64	−1.31 ± 1.09	<.001
Resting basal metabolic rate (kcal)	−25.1 ± 240	−26.0 ± 253	−6.9 ± 376	.91
Thyroid function
TSH (mUI/L)	0.04 ± 1.1	0.46 ± 2.6	−0.49 ± 3.0	.32
Free T4 (pmol/L)	−0.53 ± 1.7	0.09 ± 3.7	−0.55 ± 3.8	.38
Free T3 (pmol/L)	−0.18 ± 0.6	0.38 ± 2.5	−0.57 ± 2.7	.19
Reverse T3 (ng/dL)	−1.1 ± 4.2	−0.32 ± 5.0	−0.94 ± 5.9	.35

Data are mean ± SD, if normally distributed; or median (25th-75th), if skewed. Skewed variable were log transformed for analyses.

Figure 2.

Impact of time restricted eating (TRE) on (A) weight, (B) resting basal metabolic rate, (C) TSH, (D) free T4, (E) free T3, and (F) reverse T3. Data shown before and after 6-week period of standard lifestyle and TRE, followed by 4-week period of washout. Means and SE (error bars) are shown. P values for study intervention × time interaction. *For clarity, data are shown by study intervention and does not reflect randomization order.

Figure 3.

Thyroid function changes between baseline and intervention endpoint for each intervention period: standard lifestyle and time-restricted eating (TRE). (A) Change in free T3: free T4 ratio, and (B) change in free T3:reverse T3 ratio.

To assess the informativeness of potential nonsignificant results, we evaluated the precision and magnitude of the 95% CIs of the changes in fT4 and fT3. The study's capacity to rule out meaningful effects was determined by comparing the width of the 95% CI to a predefined minimum clinically important difference for fT4 of 2 pmol/L (which corresponds to a 10% to 20% change in circulating pool) and fT3 of 1.5 pmol/L (which represents average 25% reduction in active tissue hormone) (16, 17). A finding was considered a “true” null result if the 95% CI excluded the minimum clinically important difference, indicating high precision (18).

Results

Thirty-nine participants (mean age 56.3 years, 35.9% male) were included in this crossover trial and randomized to either an initial intervention consisting of 6 weeks of TRE or standard lifestyle followed by a subsequent 4-week washout period before crossing over to the other intervention. Table 1 shows the baseline characteristics of study participants. Participants had a baseline glycated hemoglobin 6.6 ± 0.7% (mean ± SD) and BMI of 32.4 ± 5.7 kg/m² with a resting basal metabolic rate of 1787 ± 442 Kcal. None of the participants reported known thyroid disease. The baseline TSH was 1.7 mIU/L (1.4-2.4) [median (25th-75th)], with mean ± SD fT4 level of 15.3 pmol/L ± 2.4, mean fT3 4.7 pmol/L ± 0.5, and mean rT3 levels of 18.0 ng/dL ± 4.9. As reported previously (7), TRE was well tolerated and none of the participants discontinued the intervention due to side effects.

Table 2 shows the changes in body weight and thyroid function tests between baseline and intervention endpoint for each intervention period (TRE and standard lifestyle), and comparison of these changes. TRE resulted in significant weight loss during the TRE treatment period (−2.93 kg ± 2.3) compared to the standard lifestyle period (0.59 kg ± 1.7; mean differences between intervention periods of −3.50 kg ± 2.9; P < .001). Despite this robust weight-lowering effect, there were no significant differences in resting basal metabolic rate between TRE and standard lifestyle periods (−6.9 Kcal ± 376, P = .91). Importantly, TRE induced a small change in fT4 (−0.53 pmol/L ± 1.7) and fT3 (−0.18 pmol/L ± 0.6), which was also observed after the standard lifestyle period (fT4 0.09 pmol/L ± 3.7, fT3 0.38 pmol/L ± 2.5). However, these differences did not reach statistical significance (mean difference between intervention periods for fT4 −0.55 pmol/L ± 3.8; P = .38 and for fT3 −0.57 pmol/L ± 2.7; P = .19). Importantly, although this RCT was powered for the detection of changes in pancreatic β-cell function (primary outcome) (7), the 95% CI of fT4 (−0.55; 95% CI, −1.78 to 0.68) and fT3 (−0.57; 95% CI, −1.45 to 0.31) yielded a precise estimate of the effect. In other words, given that the entire CI falls outside the range of 2.0 pmol/L (for fT4) and 1.5 pmol/L (for fT3) (16, 17), these data suggest that any true effect is likely smaller than what is considered clinically meaningful, effectively ruling out a clinically significant treatment effect (18). In addition, there were no differences in changes on TSH and rT3 levels between intervention periods (mean difference between intervention periods for TSH −0.49 mUI/L ± 3.0; P = .32, and rT3 −0.94 ng/dL ± 5.9; P = .35) (Table 2).

Figure 2 illustrates the trajectories of bodyweight, resting metabolic rates, and thyroid function tests during each treatment intervention (P values showed for within-subjects factor overtime, indicating the treatment effect of TRE). As evidenced in panel A, TRE intervention induced significant weight loss (P < .001), but there was no significant treatment effect on resting metabolic rate (panel B, P = .181), TSH (panel C, P = .183), fT4 (panel D, P = .393), fT3 (panel E, P = .309), and rT3 (panel F, P = .282). In parallel to these results, Fig. 3 demonstrates the changes in the fT3: fT4 (panel A) and fT3:rT3 (panel B) ratios between baseline and intervention endpoint for each intervention period, showing no significant differences between treatment periods that further reinforced that TRE had no clinically significant impact in thyroid hormones homeostasis in our study population. In addition, sensitivity analyses evaluating potential sex differences in the impact of TRE on thyroid function revealed no significant sex × time interactions for TSH (P = .93), fT4 (P = .70), fT3 (P = .44), or rT3 (P = .16).

Discussion

This crossover RCT demonstrates that TRE did not exert a significant effect on thyroid function in individuals with obesity and early T2DM, despite inducing significant weight loss. Specifically, TRE did not promote clinically significant changes to TSH, fT4, fT3, and rT3 levels in relation to standard lifestyle. In addition, no significant changes in the indirect assessment of deiodinase activity, fT3:fT4 and fT3:rT3 ratios, were observed.

Achieving long-term weight loss maintenance poses a major challenge in the clinical management of obesity and T2DM. That is possibly because physiologic mechanisms that are triggered by weight reduction induce compensatory changes in energy balance homeostasis after weight loss (19-21). It has been reported that following lifestyle-induced weight loss, levels of circulating mediators of appetite that encourage weight regain are switched toward appetite stimulation (19-21). In this context, changes in TH such as increases in rT3 and fT4 coupled with reduction in fT3 that ultimately impact energy expenditure toward an energy-sparing state have been reported in some studies (9, 22). However, the evidence regarding the effects of weight loss induced by dietary interventions on thyroid function in adults with obesity remains heterogenous with some reports showing a reduction in both fT3 and fT4 (23) or no changes in thyroid function (24). It is important to note that increased BMI is associated with higher TSH in the general population, likely via leptin pathway, which triggers hypothalamic stimulation of pituitary TSH, which could be reversed by weight reduction (25).

Studies that specifically evaluated the impact of TRE-induced weight loss on thyroid function are scarce. A recent secondary analysis of a RCT evaluating the impact of TRE with or without low-carbohydrate diet on myocardial status and thyroid function in participants with metabolic syndrome as compared to low-calorie diet did not show any significant differences in thyroid function between groups, suggesting no pleiotropic effect of TRE on thyroid function (26). This was consistent with results of other trials that failed to show any significant changes in TSH after TRE intervention (27, 28). Consistent with these findings, our study observed no significant alterations in thyroid function through a comprehensive serial assessment of TSH, fT4, fT3, rT3, and indirect measurements of deiodinase activity (fT3:fT4 and fT3:rT3 ratios) over the course of a crossover trial on TRE (20 hours of fasting) intervention in participants with obesity and T2DM despite significant reductions in body weight and BMI. These findings contribute to the existing literature on dietary weight loss interventions, in particular TRE and their endocrine effects, providing further evidence regarding the impact of TRE, an increasingly used strategy for weight management. One possible explanation for the lack of impact of TRE-induced weight loss on TH in our study could be that the weight loss achieved over 6 weeks was not sufficient to trigger an adaptive response in thyroidal axis. However, previous data suggest that moderate weight loss already impact thyroid homeostasis (29). In addition, a previous meta-analysis evaluating both dietary and bariatric treatment for weight loss, indicate that there is a linear relationship between weight loss and changes in TH without a clear threshold (9).

Interestingly, previous data by Zheng et al showed a significant reduction in T3:T4 ratio with TRE between pre- and postintervention, which indicates reduced peripheral conversion of T4 to T3 likely representing a reduction in type 1 deiodinase activity (25). This reduction may reflect a diminished sympathetic nervous system activity following weight reduction (30). Because T3 activates thermogenesis and induces an increase in energy expenditure and oxygen consumption, a reduction in T3 may represent an adaptative response to caloric restriction and weight reduction. However, in this previous report (25), there were no differences between changes in T3:T4 ratio induced by TRE when compared to the low-calorie diet group. Discrepancies between our findings and those of Zheng et al (25) may be attributable to differences in the study populations: our cohort consisted of individuals with treated T2DM and overweight/obesity, whereas Zheng et al (25) studied individuals with metabolic syndrome. In addition, the TRE protocols differed because Zheng et al (25) employed a 16-hour fasting window with either early or late TRE, whereas our protocol used a 20-hour fasting window with early TRE exclusively. Expanding on the physiology of intermittent fasting and thyroid regulation, our study demonstrated that TRE did not promote significant changes on both fT3:fT4 and fT3:rT3 ratio compared to standard eating in the context of significant weight reduction.

Our study has some limitations. Our RCT was powered for the primary outcome of changes in β-cell function between intervention periods (7); therefore, the findings should be interpreted as exploratory. However, we believe that the observed associations are clinically informative, given that our sample size and crossover design allowed us to rule out clinical meaningful differences between the effects of TRE and standard eating on changes in fT4 and fT4. Furthermore, the relatively short follow-up period may introduce potential limitations. Specifically, the 4-week washout interval between interventions may have been insufficient to allow for full stabilization of TSH levels, potentially confounding the assessment of the independent effects of TRE and weight loss on thyroid function. On the other hand, the evaluation of fT4, fT3, rT3, and ratios counteract this potential limitation, as the half-life on these endogenous hormones is shorter (ie, fT3 has a very short half-life of less than 24 hours, and fT4 usually stabilizes within a 4-week period). There are unanswered questions that our study design is not positioned to address. For example, we studied late TRE (eating window from 4-8 or 5-9 Pm) such that the impact of early TRE on thyroid homeostasis was not assessed. In addition, the interplay between thyroid homeostasis and TRE was not evaluated in the context of participants on levothyroxine replacement and/or potential modification on sleep pattern, conditions that could potentially interfere with thyroid regulation. Another limitation is that our sample was restricted to individuals with treated T2DM who were also overweight or obese; consequently, the effects of TRE in individuals with obesity alone could not be assessed.

In conclusion, in this crossover RCT, we demonstrated that TRE did not exert a significant effect on thyroid function in individuals with overweight and early T2DM compared to standard eating pattern. These findings suggest that weight loss induced by TRE may be associated with less impact on adaptative responses in thyroid function in participants without thyroid disease adding relevant insights to the understanding of the impact of TRE on metabolism.

Abbreviations

BMI

body mass index

fT3

free T3

fT4

free T4

RCT

randomized controlled trial

rT3

reverse T3

T2DM

type 2 diabetes mellitus

TH

thyroid hormone

TRE

time-restricted eating

Author contributions

C.K.K. conceptualize the study. Y.J. and C.K.K. acquired the data. C.K.K. performed the statistical analyses. Y.J. and C.K.K. wrote the manuscript. All authors contributed to interpretation of the data and critical revision of the manuscript. All authors approved the manuscript.

Disclosures

C.K.K. reports research grants from Boehringer Ingelheim, outside of the submitted work. D.S.F. reports an investigator-initiated grant from Dexcom, in-kind donations from Dexcom and Tandem and honoraria from Novo-Nordisk, Sanofi and Ypsomed, outside the submitted work. R.R. reports grants from Boehringer Ingelheim, grants and personal fees from Novo Nordisk, personal fees from Sanofi, personal fees from Eli Lilly, outside the submitted work. All other authors declare no competing interests.

Data availability

Deidentified data can be made available under restricted access from the corresponding author, for academic purposes, subject to a material transfer agreement and approval of the Mount Sinai Hospital Research Ethics Board.

Clinical trial information

Trial registration:  clinicaltrials.gov NCT05717127.