Mature Human Milk Macronutrient Composition and Milk TSH Concentration in Women with Diagnosed and Clinically Managed Hypothyroidism Attending Lactation Counseling and Human Milk Bank Setting: A Cross-Sectional Study

1. Introduction

Human milk is a dynamic biological fluid whose composition reflects both maternal physiology and the nutritional needs of the infant. Although macronutrient content tends to be more stable during established lactation than in colostrum or transitional milk, it may still vary with lactation stage, metabolic status, endocrine function, inflammation, and sampling methodology [1,2]. Thyroid hormones are key endocrine regulators of metabolism, energy expenditure, mammary gland development, and lactation physiology; adequate thyroid function is also essential for infant growth and neurodevelopment [3].

Hypothyroidism is among the most common endocrine disorders in women of reproductive age and may persist into the postpartum period. Untreated or insufficiently treated hypothyroidism has been associated with delayed lactogenesis, reduced milk volume, and impaired milk ejection [4,5]. Potential mechanisms include altered prolactin signaling, disrupted lipid, protein, and carbohydrate metabolism, and reduced mammary gland responsiveness [6]. However, these mechanisms do not necessarily imply altered macronutrient composition in mature milk when thyroid dysfunction has been diagnosed and treated.

Previous studies examining thyroid dysfunction and human milk have often focused on colostrum or early lactation, used single milk samples, or did not fully account for maternal treatment status, iodine exposure, body mass index, hyperlactation, or pregnancy-related comorbidities [7,8,9,10]. In routine clinical practice, many women with hypothyroidism receive levothyroxine therapy before or during pregnancy and continue medical supervision during lactation [11]. Therefore, the specific question of whether medically managed hypothyroidism is associated with altered macronutrient composition of mature human milk remains insufficiently addressed.

Thyroid-stimulating hormone (TSH), thyroxine, and other thyroid-related hormones have been detected in human milk [12,13,14,15,16]. A recent study showed that TSH and thyroxine are measurable in own mother’s milk and donor milk during the first months of lactation, but the oral bioavailability and physiological relevance of milk TSH for the infant remain uncertain [16]. Importantly, milk TSH should not be assumed to be a substitute for contemporaneous maternal serum TSH or FT4 assessment.

This study aimed to compare the macronutrient composition and TSH concentration of mature 24-hour composite human milk samples from women with diagnosed and clinically managed hypothyroidism, including a small subgroup with Hashimoto’s disease, with samples from women without thyroid disorders. A secondary objective was to explore whether pre-pregnancy BMI or clinically defined hyperlactation was associated with milk composition or milk TSH levels, given the recognized links between thyroid dysfunction, gestational diabetes, and autoimmune thyroid disease in pregnancy [17,18]. By focusing on established lactation and 24-hour composite sampling in a lactation-counseling setting, the study addresses a gap left by earlier work that relied mainly on early lactation samples or less standardized collection procedures, while also requiring cautious interpretation because the cohort was selected clinically rather than population-based.

2. Materials and Methods

2.1. Study Design and Participants

This observational, cross-sectional study included 66 lactating women who sought lactation counseling at the Human Milk Bank of the Ludwik Rydygier Provincial Polyclinical Hospital in Toruń, Poland, between March 2020 and November 2021. Participants were recruited during routine consultations and provided written informed consent before enrollment. All participants were in established lactation at the time of sampling (≥ 4 weeks postpartum). Therefore, no early-vs-established lactation subgroup analysis was performed.

Participants were recruited from a clinical lactation counseling and human milk bank setting rather than from a population-based cohort. Thus, the sample represents women seeking lactation support and/or milk-bank consultation. The detailed indication for consultation was not collected as a mutually exclusive analyzable variable for every participant; however, the observed clinical context included breastfeeding difficulties such as excessive milk production or forceful flow, infant choking or regurgitation during feeding, milk-transfer or breastfeeding-management concerns, and questions about milk quality or possible donation. This selected recruitment setting was considered when interpreting the high prevalence of hyperlactation and the external validity of the findings.

Two groups were defined: the hypothyroidism group (HG; n = 31), comprising women diagnosed with hypothyroidism, including Hashimoto’s disease (n = 6), and the control group (CG; n = 35), comprising women without thyroid disorders. Thyroid disorders were diagnosed before conception or during the first trimester on the basis of clinical evaluation, medical documentation, and laboratory testing performed as part of routine preconception or early-pregnancy/perinatal care (including TSH, FT4, and thyroid autoantibodies when clinically indicated), following current endocrinology guidelines [19,20,21]. No blood samples for thyroid hormone assessment were collected specifically for this study at the time of milk sampling. In the HG, thyroid dysfunction had been identified before pregnancy in most cases, most commonly as primary hypothyroidism. Levothyroxine therapy was used when clinically indicated; doses were individualized and adjusted by the treating endocrinologist or obstetrician-gynecologist according to the patient’s thyroid hormone results and clinical response. In this cohort, 25/31 women (81%) used levothyroxine during pregnancy and 19/31 (61%) continued it during lactation; all women with Hashimoto’s disease received levothyroxine therapy during lactation. Exact levothyroxine dose at milk sampling and timing of milk collection in relation to medication intake were not systematically recorded.

None of the participants received individual therapeutic doses of iodine. During pregnancy and lactation, all participants used preparations containing prophylactic iodine doses (typically 150–200 µg/day), as recommended in Poland. In addition, Poland has maintained a national iodine prophylaxis model based on obligatory iodization of household salt since 1997; the iodine content of salt has been described as 30 ± 10 mg potassium iodide/kg NaCl or, since 2002, 39 ± 13 mg potassium iodate/kg NaCl [22]. Inclusion criteria were age ≥ 18 years, singleton pregnancy, term or near-term birth (≥ 37 weeks), established lactation, and ability to provide a complete 24-hour milk sample. Exclusion criteria were postpartum thyroiditis, untreated thyroid dysfunction, other endocrine disorders (e.g., type 1 diabetes, PCOS), medications known to affect lactation (dopaminergic agents, corticosteroids), active mastitis, smoking or substance use, and chronic diseases affecting metabolism (renal, hepatic, or autoimmune disease other than Hashimoto’s disease).

Sociodemographic and clinical data were collected using a structured 43-item questionnaire covering maternal characteristics, pregnancy course, mode of birth, infant feeding, chronic diseases, medication use, and lifestyle factors. Pre-pregnancy BMI was calculated using pre-pregnancy weight and height; BMI ≥ 25 kg/m² was classified as overweight/obesity according to the WHO adult BMI classification [23].

Hyperlactation (HL) was defined clinically because the literature does not provide a single universally accepted diagnostic threshold. In this study, HL was assessed by a lactation consultant and recorded when persistent symptoms suggested excessive milk production or excessive milk flow. Criteria included intense milk flow during latch or feeding, milk leaking from the infant’s mouth during breastfeeding, audible rapid swallowing suggestive of air swallowing while attempting to slow milk flow, choking or coughing during feeding, the infant being “flooded” with milk during breastfeeding, excessive infant weight gain relative to age norms, and/or marked regurgitation or burping after feeds after exclusion of other causes, such as sonographic features of pyloric stenosis. This definition is consistent with the clinical framing of the Academy of Breastfeeding Medicine, which recognizes hyperlactation as a condition that may be self-induced, iatrogenic, or idiopathic [24].

2.2. Milk Sampling

Milk collection followed standardized written instructions. Each participant collected milk over a 24-hour period in four intervals: 06:00–12:00, 12:00–18:00, 18:00–24:00, 24:00–06:00.

During each interval, mothers expressed 10–15 mL of milk before and after breastfeeding. For exclusively pumping mothers, 15 mL aliquots were taken from each expressed portion. Samples were stored at 4–8 °C and transported to the laboratory in a cooled container. The maximum storage time before analysis was 12 hours.

For each participant, all samples were pooled into a single composite sample representing the 24-hour period.

2.3. Determination of Macronutrient Composition

Macronutrient content—fat, crude protein, true protein, carbohydrates, total solids, and energy—was analyzed using the MIRIS Human Milk Analyzer (Miris AB, Uppsala, Sweden), based on mid-infrared transmission spectroscopy [25]. This method is widely used in clinical and milk-bank settings for rapid human milk macronutrient assessment, although validation studies indicate that accuracy varies by analyte and that protein measurements require cautious interpretation [25,26].

Before analysis, samples were heated to 40 °C and homogenized using a MIRIS Sonicator (1.5 s/mL). Each sample was analyzed in triplicate, and the mean value was used for statistical analysis. Daily zero-calibration was performed using manufacturer-provided standards.

2.4. Measurement of TSH Concentration in Human Milk

TSH concentrations were determined using a third-generation enzyme immunofluorescence assay (ELFA) on the VIDAS analyzer (bioMérieux, France), following the manufacturer’s analytical protocol [27]. Milk samples were centrifuged at 2000 rpm for 5 minutes, and 200 µL of supernatant was used for analysis. Each measurement was performed in duplicate. The assay is intended and manufacturer-validated for human serum or plasma rather than human milk; therefore, milk TSH values in this study should be interpreted as exploratory matrix measurements. No formal recovery, linearity, or dilution validation in the milk matrix was performed. This methodological limitation was considered when interpreting the findings.

2.5. Statistical Analysis

All statistical analyses were performed using IBM SPSS Statistics 26.0 (IBM Corp., Armonk, NY, USA). The distribution of continuous variables was assessed using the Shapiro–Wilk test and visual inspection of histograms and Q–Q plots. Approximately normally distributed variables were analyzed using parametric tests (t-test or ANOVA), whereas non-normally distributed variables were analyzed using non-parametric methods (Mann–Whitney U or Kruskal–Wallis tests).

The six milk-composition outcomes constituted the primary outcome family. The between-group comparison of apparent milk TSH concentration was treated as a single secondary outcome. BMI, hyperlactation, Hashimoto’s disease, and correlation analyses were considered exploratory.

Baseline continuous characteristics were compared using Welch’s two-sample t tests and are reported together with the HG−CG mean difference and its 95% confidence interval. Categorical characteristics were compared using Pearson’s chi-square test, whereas Fisher’s exact test was used for GDM and PIH because of small expected cell counts. Effect estimates for categorical variables were expressed as odds ratios comparing the HG with the CG, with corresponding 95% confidence intervals. Baseline comparisons were descriptive and were not included in the false discovery rate correction. Lactation stage was not modeled as an early-vs-established variable because all participants were already in established lactation. Between-group comparisons of the six milk-composition outcomes were performed using Welch’s two-sample t tests, which do not assume equal variances. For each outcome, the mean difference between groups (HG − CG) was reported together with its two-sided 95% confidence interval calculated using the Welch–Satterthwaite degrees of freedom, the exact p value, and Hedges’ g. Hedges’ g was calculated using the pooled within-group standard deviation with a small-sample correction; negative values indicate lower mean values in the HG than in the CG. To control the false discovery rate across the family of six milk-composition outcomes, p values were adjusted using the Benjamini–Hochberg procedure and are reported as q values. Differences in the prevalence of hyperlactation, GDM, PIH, and BMI category were assessed using Chi-square tests or Fisher exact tests when expected cell counts were small. Subgroup analyses of BMI, hyperlactation, and Hashimoto’s disease were exploratory because of the modest sample size and sparse subgroup counts.

Separate two-way ANOVA models were fitted for each milk-composition outcome, with study group (HG vs. CG), pre-pregnancy BMI category (<25 vs. ≥25 kg/m²), and the group × BMI interaction included as model terms. Homogeneity of variance was assessed using Levene’s test. Effect sizes were expressed as partial eta squared (η²p). To control the false discovery rate across the six BMI main-effect tests, the corresponding p values were adjusted using the Benjamini–Hochberg procedure. Interaction and subgroup findings were interpreted cautiously because of the modest sample size.

Because apparent milk TSH concentrations were not normally distributed, they were summarized using the median, interquartile range (IQR), and minimum–maximum range. Between-group comparisons were performed using the Mann–Whitney U test, and differences across four subgroups defined by study group and BMI category or hyperlactation status were assessed using the Kruskal–Wallis H test. Effect sizes for Mann–Whitney comparisons based on the asymptotic test were expressed as r, calculated as |Z|/√N. For comparisons involving small or unbalanced subgroups, including the Hashimoto’s disease subgroup, the exact Mann–Whitney U test was used where appropriate, and the effect size was expressed as the rank-biserial correlation. Effect sizes for Kruskal–Wallis tests were expressed as eta squared based on the H statistic (η²H). Associations between apparent milk TSH concentrations and macronutrient content were evaluated using Spearman’s rank-order correlation. To control the false discovery rate across the 12 Spearman correlation tests (six milk-composition outcomes evaluated separately in the two study groups), p values were adjusted using the Benjamini–Hochberg procedure and are reported as q values.

No a priori sample-size calculation was performed. The study sample comprised all eligible consenting participants enrolled during the study period. A sensitivity analysis indicated that, with 31 participants in the hypothyroidism group and 35 participants in the control group, a two-sided independent two-sample comparison at α = 0.05 had 80% power to detect a standardized between-group difference of approximately Cohen’s d = 0.70. Therefore, the study may not have detected small or moderate between-group differences.

All statistical tests were two-sided, and statistical significance was set at p < 0.05. Effect sizes were reported as Hedges’ g for between-group comparisons of milk composition, partial eta squared (η²p) for ANOVA effects, eta squared based on the Kruskal–Wallis statistic (η²H), and r or rank-biserial correlation for Mann–Whitney comparisons, as appropriate. Adjusted p values obtained using the Benjamini–Hochberg procedure are reported as q values. No formal equivalence or non-inferiority analysis with pre-defined clinical margins was performed; therefore, non-significant findings should not be interpreted as evidence of equivalence or absence of an effect. Secondary and subgroup analyses were considered exploratory.

3. Results

3.1. Participant Characteristics

A total of 66 lactating women were included: 31 in the HG with diagnosed hypothyroidism (including Hashimoto’s disease) and 35 in the CG without thyroid disorders. The groups did not differ significantly in maternal age, gestational age at birth, mode of birth, parity, or stage of lactation at the time of sampling (Table 1). All 66 participants in the analytic dataset had milk macronutrient and milk TSH results available. Information on eligible women who were not enrolled during routine counseling was not recorded systematically, which limits participant-flow reporting.

Thyroid dysfunction was diagnosed before conception or during the first trimester in most women in the HG. Levothyroxine therapy was used by 25/31 women (81%) during pregnancy and by 19/31 (61%) during lactation; all six women with Hashimoto’s disease received treatment throughout the study period [19,20,21]. Exact dose ranges and medication timing relative to milk collection were not available in the research dataset.

GDM was more frequent in the HG (7/31; 22.6%) than in the CG (4/35; 11.4%), but this difference was not statistically significant (Fisher’s exact test, p = 0.324). PIH occurred in 3/31 women (9.7%) in the HG and 5/35 women (14.3%) in the CG (Fisher’s exact test, p = 0.713). Overweight/obesity (BMI ≥ 25 kg/m²) affected more than 30% of women in both groups before pregnancy and after birth.

Hyperlactation was significantly more common in the control group (27/35; 77.1%) than in the hypothyroidism group (16/31; 51.6%) (χ² = 4.719, p = 0.030). These are within-group prevalences; the previously reported 62.8% and 37.2% values referred to the distribution of all hyperlactation cases across groups and have been removed to avoid misunderstanding.

3.2. TSH Concentrations in Human Milk

Because apparent milk TSH concentrations were non-normally distributed, the results are presented as medians, interquartile ranges, and minimum–maximum ranges. The median apparent milk TSH concentration was 0.0180 [0.0130–0.0208] µIU/mL in the control group (CG; n = 35; range: 0.0025–0.0690 µIU/mL) and 0.0155 [0.0093–0.0210] µIU/mL in the hypothyroidism group (HG; n = 31; range: 0.0000–0.1080 µIU/mL). The Mann–Whitney U test showed no statistically significant difference between the groups (U = 459.0, Z = −1.07, p = 0.286; r = 0.131).

Within the HG, an exploratory comparison was conducted between women with Hashimoto’s disease (n = 6) and women with hypothyroidism without Hashimoto’s disease (n = 25). The median apparent milk TSH concentration was 0.0100 [0.0091–0.0124] µIU/mL in women with Hashimoto’s disease (range: 0.0000–0.0235 µIU/mL) and 0.0170 [0.0100–0.0210] µIU/mL in women with hypothyroidism without Hashimoto’s disease (range: 0.0000–0.1080 µIU/mL). An exploratory comparison using the exact Mann–Whitney U test showed no statistically significant difference in apparent milk TSH concentrations between women with Hashimoto’s disease and women with hypothyroidism without Hashimoto’s disease (U = 49.0, exact p = 0.208; rank-biserial correlation = −0.347). Given the small number of participants with Hashimoto’s disease, this subgroup comparison should be regarded as exploratory and hypothesis-generating.

3.3. Macronutrient Composition of Human Milk

Macronutrient content of mature milk did not differ significantly between groups in the analyses performed (Table 2).

Welch’s two-sample t tests showed no statistically significant between-group differences in any of the six milk-composition outcomes (all p ≥ 0.169). None of the comparisons remained statistically significant after Benjamini–Hochberg correction (all q ≥ 0.814). The largest absolute standardized effect was observed for energy content (Hedges’ g = −0.345), but its 95% confidence interval for the mean difference included zero. These results should not be interpreted as evidence of equivalence.

3.4. Associations Between TSH and Milk Composition

Spearman’s rank-order correlation identified a nominal positive association between apparent milk TSH concentration and carbohydrate content in the CG (ρ = 0.363, p = 0.032), but this association did not remain statistically significant after Benjamini–Hochberg correction across the 12 correlation tests (q = 0.387). No other correlation was nominally significant (all p ≥ 0.122), and none remained statistically significant after false discovery rate correction (all q ≥ 0.387). Given the absence of concurrent maternal serum hormone data and limited milk-matrix validation of the TSH assay, these findings should not be interpreted as evidence that milk TSH reflects maternal thyroid status.

3.5. Associations of Pre-Pregnancy BMI with Human Milk TSH and Milk Composition

Pre-pregnancy BMI was categorized as <25 or ≥25 kg/m². The resulting subgroups comprised CG/BMI <25 (n = 24), CG/BMI ≥25 (n = 11), HG/BMI <25 (n = 20), and HG/BMI ≥25 (n = 11). Human milk TSH concentrations did not differ significantly among these four subgroups (Kruskal–Wallis H(3) = 2.73, p = 0.436; η²_H = 0.00).

Separate two-way ANOVAs were conducted for fat, crude protein, carbohydrates, total solids, true protein, and energy content, with study group (HG vs. CG) and pre-pregnancy BMI category as factors. No group × BMI interaction was observed for fat (F(1,62) = 0.38, p = 0.539, η²_p = 0.006), crude protein (F(1,62) = 0.04, p = 0.838, η²_p = 0.001), carbohydrates (F(1,62) = 0.45, p = 0.503, η²_p = 0.007), total solids (F(1,62) = 0.46, p = 0.502, η²_p = 0.007), true protein (F(1,62) = 0.06, p = 0.803, η²_p = 0.001), or energy content (F(1,62) = 0.09, p = 0.761, η²_p = 0.002). No significant main effects of study group were identified (all p ≥ 0.125).

Before correction for multiple comparisons, a main effect of BMI category was observed for energy content (F(1,62) = 4.91, p = 0.030, η²_p = 0.073). However, this association did not remain significant after Benjamini–Hochberg correction across the six BMI main-effect tests (q = 0.181). The main effects of BMI were also non-significant for fat (F(1,62) = 3.66, p = 0.060), crude protein (F(1,62) = 0.10, p = 0.753), carbohydrates (F(1,62) = 0.34, p = 0.562), total solids (F(1,62) = 2.32, p = 0.133), and true protein (F(1,62) = 0.21, p = 0.650). Thus, no milk-composition outcome showed a statistically significant association with pre-pregnancy BMI after correction for multiple testing.

Hyperlactation occurred in 14 of 22 women with pre-pregnancy BMI ≥25 kg/m² (63.6%) and in 29 of 44 women with BMI <25 kg/m² (65.9%). This difference was not statistically significant (Fisher’s exact test, p = 1.000).

3.6. Association of Hyperlactation with Human Milk TSH Concentrations

Human milk TSH concentrations were compared across four subgroups defined by study group and hyperlactation status: CG without hyperlactation (n = 8), CG with hyperlactation (n = 27), HG without hyperlactation (n = 15), and HG with hyperlactation (n = 16). The corresponding median TSH concentrations [IQR] were 0.0133 [0.0096–0.0261], 0.0185 [0.0143–0.0203], 0.0105 [0.0045–0.0180], and 0.0183 [0.0124–0.0215] µIU/mL, respectively. The Kruskal–Wallis test showed no statistically significant differences among the four subgroups (H(3) = 4.69, p = 0.196; η²_H = 0.027).

In the overall sample, the median milk TSH concentration was 0.0185 [0.0135–0.0210] µIU/mL among women with hyperlactation and 0.0125 [0.0068–0.0220] µIU/mL among women without hyperlactation. This difference did not reach statistical significance (Mann–Whitney U = 637.0, p = 0.056; rank-biserial correlation = 0.288). Exploratory analyses conducted separately within the study groups also showed no significant association between hyperlactation and milk TSH concentration (CG: U = 123.5, p = 0.555; HG: U = 162.0, p = 0.101). Therefore, the available data did not provide evidence of an association between hyperlactation and human milk TSH concentration.

4. Discussion

This study evaluated macronutrient composition and TSH concentration in mature 24-hour composite human milk samples from women with diagnosed and clinically managed hypothyroidism, including a small subgroup with Hashimoto’s disease, compared with controls without thyroid disorders. In this cohort, no statistically significant between-group differences were observed in milk macronutrient composition or milk TSH concentration. These findings are clinically reassuring, but they should be interpreted within the limits of a small cross-sectional study and should not be read as proof of equivalence, confirmed euthyroidism at milk sampling, or absence of biological effect.

The lack of statistically significant differences in milk composition is compatible with the clinical expectation that timely diagnosis, routine monitoring, and levothyroxine therapy may reduce endocrine disturbances relevant to lactation [11,19,20,21]. Thyroid hormones influence lipid, protein, and carbohydrate metabolism, and untreated hypothyroidism may impair lactogenesis and milk ejection [4,5,6]. However, this study did not include untreated women and did not include contemporaneous serum TSH/FT4 measurements at the time of milk sampling; therefore, it cannot determine whether women were biochemically euthyroid during sampling, assess dose-response relationships with levothyroxine, or determine whether different degrees of thyroid control would have affected milk composition.

Milk TSH concentrations were low and not significantly different between groups. Earlier studies detected TSH, T3, and/or T4 in human milk, and more recent work confirmed measurable TSH and thyroxine in own mother’s milk and donor milk during the first months of lactation [12,13,14,15,16]. Nevertheless, the physiological relevance and oral bioavailability of milk TSH remain unresolved. In the present study, no association between apparent milk TSH concentration and macronutrient content remained statistically significant after correction for multiple testing. Apparent milk TSH concentration was also not significantly associated with maternal pre-pregnancy BMI or hyperlactation. Because the ELFA assay used here is validated for serum/plasma and not formally validated for human milk, and because serum thyroid function was not measured concurrently, milk TSH should be interpreted only as an exploratory milk biomarker, not as a marker of maternal thyroid control or infant thyroid exposure.

Hyperlactation was frequent in this cohort and more common in the control group. This finding requires cautious interpretation because all women were recruited from a human milk bank/lactation-consultation setting, where oversupply, forceful milk flow, concerns about infant feeding, or questions about donation may themselves prompt referral. Hyperlactation may be self-induced, iatrogenic, or idiopathic and is not necessarily endocrine in origin [24]. Differences between groups may therefore reflect referral patterns, breastfeeding management, pumping practices, infant feeding behavior, or selection bias rather than a direct thyroid-related mechanism. Given that untreated hypothyroidism has been associated with impaired milk production [4,5], thyroid-related differences in lactation physiology might hypothetically influence the propensity to develop milk oversupply. However, the present cross-sectional data cannot distinguish a biological mechanism from differences in referral patterns, breastfeeding management, or milk-expression practices. In this cohort, hyperlactation was not significantly associated with apparent milk TSH concentration. A key strength of the study is the use of 24-hour composite milk samples, which reduce variability associated with diurnal changes and foremilk–hindmilk differences. The study also focused on mature milk, a stage less frequently examined in the context of maternal thyroid disorders, and included clinically relevant variables such as pre-pregnancy BMI, and hyperlactation. At the same time, the article should be viewed as a focused clinical milk-composition study rather than a metabolomic profiling study. Its contribution to Metabolites lies in the endocrine-metabolic framing of lactation and milk macronutrient composition, but it does not provide broad metabolomic or mechanistic pathway data.

Several limitations must be emphasized. First, the sample size was modest, especially for subgroup analyses; the Hashimoto’s disease subgroup included only six women and all disease-specific comparisons should be regarded as exploratory. The sensitivity analysis showed that the available sample provided 80% power only for standardized between-group differences of approximately d = 0.70 or larger; therefore, small and moderate differences may have remained undetected. Second, maternal serum TSH, FT4, thyroid antibody titers, exact levothyroxine dose, and timing of milk sampling in relation to medication intake were not systematically collected, preventing confirmation of euthyroidism, analysis of dose variability, and direct correlation between serum and milk thyroid markers. Third, the milk TSH assay was not formally validated for the human milk matrix, which limits interpretation of absolute milk TSH concentrations. Fourth, maternal diet was not analyzed in detail, and iodine status was not measured; dietary variability and iodine exposure could therefore not be evaluated as contributors to milk composition. Fifth, the cross-sectional design precludes causal inference and cannot assess changes across lactation. Sixth, the study population consisted of women attending a human milk bank or lactation consultation, which may limit generalizability to the broader population of breastfeeding women with hypothyroidism or to all potential milk donors.

Overall, the results suggest that in this selected cohort, diagnosed and clinically managed hypothyroidism was not associated with statistically significant differences in mature human milk macronutrient composition or milk TSH concentration. These findings may support cautious clinical reassurance for breastfeeding women with treated hypothyroidism, while reinforcing the need for individualized thyroid monitoring, appropriate levothyroxine dose adjustment by treating physicians, and lactation support. They should not be used to infer donor milk eligibility on their own, because the study did not evaluate milk-bank screening criteria, microbiological safety, micronutrients, iodine concentration, thyroid hormone transfer to infants, or infant outcomes.

5. Conclusions

In this small cross-sectional cohort of women attending lactation counseling, diagnosed and clinically managed hypothyroidism, including an exploratory subgroup with Hashimoto’s disease, was not associated with statistically significant differences in mature human milk macronutrient composition or milk TSH concentration compared with controls without thyroid disorders. Exploratory analyses did not identify statistically significant associations between milk TSH levels and maternal pre-pregnancy BMI, hyperlactation or macronutrient content, but these subgroup analyses were limited by sample size and should be interpreted cautiously. The findings should not be taken as evidence of equivalence, confirmed euthyroidism during milk sampling, or absence of any possible biological effect.

Future research should include larger and more diverse cohorts, untreated and suboptimally treated thyroid dysfunction where ethically and clinically appropriate, contemporaneous maternal serum TSH/FT4 and antibody measurements, levothyroxine dose and medication-timing data, detailed dietary assessment, iodine-status assessment, formal validation of thyroid hormone assays in human milk, longitudinal sampling across lactation, and infant outcomes to clarify whether thyroid-related hormones in milk have physiological relevance for the infant.