Preconception body mass index and maternal thyroid function: a longitudinal observational study Authors: Mariacarla Moleti, Angela Alibrandi, Lelio Crupi, Giuseppe Paola, Laura Giovanna Perdichizzi, Luca Balbo, Sara Marchese, Giuseppe Sgrò, Serena D’Anna, Marta Ragonese, Roberta Granese, Francesco Vermiglio, Salvatore Cannavò Published: 2026 ABSTRACT ObjectiveThe objective of this study was to examine the association between preconception body mass index (p-BMI) and maternal thyroid function longitudinally evaluated over pregnancy and to explore whether early-pregnancy thyroid parameters are associated with adverse obstetric outcomes.MethodsUsing preconception BMI as a stratification variable, 1,107 pregnant women were classified as normal weight (NW), overweight (OW), or obese (OB). Thyroid function was assessed at ≤13 gestational weeks and at two subsequent time points. Thyrotropin (TSH) and free thyroxine (FT4) longitudinal trajectories were analyzed using generalized estimating equations, and the FT4–TSH relationship was evaluated by correlation analyses. As an exploratory analysis, logistic regression models were estimated to assess associations between first-trimester thyroid parameters and a composite adverse obstetric outcome (preterm delivery, labor dystocia, premature rupture of membranes, placental abruption, or abnormal amniotic fluid volume).ResultsAt early pregnancy, OW and OB women exhibited lower FT4 and higher TSH concentrations than NW women. Across gestation, FT4 declined and TSH increased in all p-BMI categories, with persistently lower FT4 and higher TSH levels in women with higher p-BMI and no significant p-BMI × gestational period interaction. Low FT4 in the first trimester was independently associated with adverse obstetric outcomes (odds ratio (OR) = 2.04, 95% CI: 1.32–3.11, P = 0.001), whereas the association with elevated TSH was weaker (OR = 1.71, 95% CI: 1.04–2.77, P = 0.03).ConclusionHigher p-BMI is associated with a persistently less favorable maternal thyroid hormone profile throughout pregnancy. Low FT4 concentrations during early pregnancy showed the strongest association with obstetric risk, supporting further investigation into early thyroid assessment in higher-risk populations. Introduction The relationship between thyroid function and obesity reflects a complex bidirectional interplay involving central regulation, peripheral sensitivity, and hormonogenesis (1). Thyroid hormones play a physiological role in adipose tissue regulation by modulating both lipolysis (2) and adipocyte differentiation (3), and both hypothyroidism and hyperthyroidism can affect body weight through alterations in basal metabolic rate, appetite regulation, and thermogenesis (4, 5). On the other hand, metabolic disturbances related to excessive adiposity and caloric overload may influence the hypothalamic–pituitary–thyroid (HPT) axis (1). In non-pregnant individuals, excess body weight is associated with higher serum thyroid-stimulating hormone (TSH) concentrations and subtle alterations in circulating thyroid hormone levels, even in the absence of overt thyroid disease (1, 6, 7). Proposed mechanisms include leptin-mediated stimulation of TRH secretion, changes in deiodinase activity, low-grade inflammation, and potential tissue-level resistance to thyroid hormone (5, 8). Pregnancy is a period of increased thyroid hormone requirements and dynamic HPT axis adaptations (9), with early gestation representing a critical window in which adequate maternal free thyroxine (FT4) availability is essential for placental development and fetal neurocognitive outcomes (9, 10, 11, 12, 13). Several studies have reported an association between higher maternal BMI and subtle alterations in thyroid function during pregnancy, including higher TSH levels and an increased likelihood of isolated hypothyroxinemia (14, 15, 16, 17, 18, 19, 20, 21). Yet, only a small number of investigations have characterized thyroid function from the very beginning of gestation (17, 19, 20, 21) and even fewer have incorporated concurrent measurements of TSH and FT4 (19, 20, 21), which offer complementary insights into early HPT axis adaptations. Early-pregnancy assessments also tend to be cross-sectional, and the extent to which gestational trajectories of TSH and FT4 vary according to preconception BMI (p-BMI) remains unclear (19, 21). A weakened inverse relationship between FT4 and TSH has been described in individuals with obesity (22, 23), raising the hypothesis that feedback sensitivity may be altered before conception and potentially persist during gestation. In addition, the physiological relationship between FT4 and TSH is itself modified during pregnancy due to pregnancy-related changes in thyroid physiology and hormone-binding proteins, which complicates the interpretation of thyroid function tests and may alter the FT4–TSH feedback relationship (24). Whether the FT4–TSH relationship differs across p-BMI strata during pregnancy has not been systematically evaluated. Finally, the clinical implications of BMI-associated thyroid alterations remain debated. Early-pregnancy hypothyroxinemia and subtle thyroid dysfunction have been associated with adverse obstetric and neonatal outcomes (20, 25, 26), although evidence varies across populations with different iodine status and ethnicity. Exploring how p-BMI is associated with maternal thyroid function and whether early thyroid alterations are independently associated with subsequent pregnancy risk may provide insight into underlying mechanisms and help inform maternal risk stratification. In this context, we conducted a retrospective study in which p-BMI was used as a stratification variable to characterize early-pregnancy thyroid function and its longitudinal evolution over gestation, with the primary objective of exploring whether longitudinal patterns differed across p-BMI categories. Preconception BMI, rather than gestational weight, was selected as the primary stratification variable because it temporally precedes pregnancy-related physiological changes and is, therefore, less influenced by gestational weight gain, fluid retention, and pregnancy-induced metabolic and endocrine adaptations that may complicate the interpretation of thyroid function during pregnancy. In addition, differences in the FT4–TSH relationship in the first trimester and over the course of gestation according to the p-BMI category were examined. Finally, as an exploratory analysis, we investigated whether first-trimester thyroid parameters were associated with a composite adverse obstetric outcome. Patients and methods Study design and population This retrospective longitudinal observational study was conducted at a dedicated outpatient clinic for gestational thyroid disorders within a tertiary academic hospital (G. Martino University Hospital of Messina, Italy), located in a mildly iodine-deficient area (27). Between January 2014 and December 2023, a total of 1,593 consecutive pregnant women aged ≥18 years were screened for thyroid dysfunction. Since this study aimed at characterizing maternal thyroid function from early pregnancy and its longitudinal evolution across gestation according to preconception BMI (p-BMI), women were eligible if they underwent thyroid function testing in early gestation (≤13 gestational weeks (GW)) and during subsequent stages of pregnancy. Inclusion criteria included (i) singleton pregnancies, (ii) thyroid function testing at early gestation (≤13 GW), and (iii) availability of pregnancy and delivery data for outcome analyses. Exclusion criteria included (i) twin pregnancies, (ii) medically assisted reproduction, (iii) use of thyroid-interfering medications, including levothyroxine (LT4) prior to the first thyroid assessment, (iv) first thyroid testing >13 GW, and (v) missing or unreliable p-BMI data. Data collection and study cohort Preconception weight was routinely recorded at the first prenatal visit as part of the standard obstetric history, either obtained from clinical records when available or self-reported using a standardized questionnaire. For data quality purposes, all preconception weight values were systematically compared with weight measured at the first antenatal visit. This comparison was intended to identify implausible discrepancies exceeding the magnitude of variation reasonably expected between the preconception period and early gestation. Discrepancies greater than 5% or 3 kg were considered unreliable and led to exclusion, in line with validation studies reporting systematic underreporting of self-estimated preconception weight and average discrepancies in the range of 1.5–5 kg across preconception BMI categories (28). Height was measured at the first antenatal visit, and body mass index (BMI) was calculated as weight (kg)/height2 (m2). Based on the above-mentioned data quality checks and the prespecified inclusion/exclusion criteria, the final study cohort comprised 1,119 pregnant women (70.2% of those screened). According to p-BMI, participants were stratified into the following p-BMI categories: underweight (UW, p-BMI < 18.5 kg/m2, n = 12/1,119, (1.1%)); normal weight (NW, p-BMI: 18.5–24.9 kg/m2, n = 789/1,119, (70.5%)); overweight (OW, p-BMI: 25.0–29.9 kg/m2, n = 205/1,119 (18.3%)); and obesity (OB, p-BMI: ≥ 30.0 kg/m2, n = 113/1,119 (10.1%)). As the number of women in the underweight category was too small to allow meaningful statistical comparisons, this category was excluded from all analyses, and the final study cohort consisted of 1,107 pregnant women (Fig. 1). Study flow chart. Flow diagram depicting participant selection and study sample composition. Obstetric data were extracted from clinical records and verified against predefined clinical definitions, including confirmation of gestational age thresholds and review of clinical details when nonspecific obstetric complications were reported; women with incomplete pregnancy/delivery data were excluded from outcome analyses. Biochemical measurements and clinical management Maternal blood samples were routinely collected via morning venipuncture and processed within 3 h. Thyroid function was assessed at three predefined gestational windows (early pregnancy (≤13 GW), second trimester (GW 14–26), and third trimester (GW ≥ 27)), thereby providing repeated within-subject measurements for longitudinal analysis. TSH and FT4 concentrations were measured using Elecsys electrochemiluminescence immunoassays (ECLIA) on the cobas e-analyzers (Roche Diagnostics, GmbH, Germany). For TSH, the limit of detection was 0.005 mIU/L, with intra- and inter-assay coefficients of variation (CVs) <3 and <5%, respectively. The manufacturer’s reference interval for non-pregnant adults was 0.27–4.20 mIU/L. For FT4, the functional sensitivity was 0.3 pmol/L, with intra- and inter-assay CVs <5 and <6%, respectively. The reference range provided by the manufacturer was 11.9–21.6 pmol/L. TPOAb were measured using the Elecsys Anti-TPO ECLIA assay (cobas e-analyzers). The assay has a quantitation limit of 15 IU/mL (CV ≤ 20%), with intra- and inter-assay CVs of 4.0–7.7% and 5.5–11.7%, respectively. The upper reference limit (95th percentile) was ≤34 IU/mL. Clinical management aligned with contemporary evidence-based recommendations from major endocrine societies available at the time of clinical assessment (27, 28, 29). Specifically, LT4 replacement therapy was initiated when TSH exceeded an upper normal limit (UNL) of 2.5 mIU/L in the presence of thyroid autoimmunity and 4.0 mIU/L otherwise (29, 30). Treatment of isolated hypothyroxinemia was not systematic, as it was considered only when detected in early gestation, in accordance with European Thyroid Association guidance (31). In the absence of population- and trimester-specific reference ranges for FT4, the lower normal limit (LNL) provided by the assay manufacturer (11.9 pmol/L) was applied. Statistical analysis Descriptive statistics were performed to summarize baseline characteristics in the whole population and across p-BMI categories. Continuous variables are presented as median and interquartile range (IQR), and categorical variables as counts and percentages. As most of the tested variables were not normally distributed, nonparametric tests were employed: the Mann–Whitney U test (comparisons between two independent groups); the Kruskal–Wallis test with Bonferroni-corrected pairwise comparisons (comparisons among three or more groups). Categorical variables were compared using the chi-square (χ2) test. The primary analyses focused on the characterization and comparison of longitudinal trajectories of maternal TSH and FT4 across pregnancy according to p-BMI categories (NW, OW, and OB). To this end, since each woman contributed multiple thyroid function measurements during gestation, associations between p-BMI categories and thyroid function parameters during pregnancy were assessed using generalized estimating equations (GEE). This approach explicitly accounts for the correlation between repeated measurements obtained from the same individual and estimates population-averaged mean differences. Models included p-BMI class and gestational period as main effects, and a p-BMI-by-gestational period interaction was tested to assess whether p-BMI-related differences varied across gestation. Adjusted models were fitted by accounting for maternal age, thyroid autoimmunity, preconception iodine nutritional status based on prior iodized salt use, and parity, selected a priori because of their biological relevance and potential confounding effects on TSH and FT4 levels. Since levothyroxine treatment substantially modifies thyroid function parameters, the primary longitudinal analysis was restricted to thyroid function measurements obtained before treatment initiation, if any, to avoid the pharmacological effect of levothyroxine on hormone measurements and to describe endogenous thyroid hormone trajectories. Accordingly, all women contributed hormonal data to the longitudinal analyses at baseline (first trimester), with progressively fewer observations available at later gestational intervals as LT4 treatment was initiated in women newly diagnosed with thyroid hypofunction or due to loss to follow-up. Since this approach may introduce informative censoring, as treatment initiation is related to thyroid function levels, sensitivity analyses were additionally performed by i) including all women and all measurements, including those obtained after treatment initiation, and ii) restricting the analysis to women who did not initiate levothyroxine therapy at any time during pregnancy. The results from GEE models are reported as regression coefficients (β) with 95% confidence intervals (95% CI). To allow comparison of thyroid function trajectories across BMI categories, model-based adjusted mean values (estimated marginal means – EMMEANS) were computed for each p-BMI group and time point, accounting for the covariates included in the underlying GEE model. Pairwise comparisons between estimated marginal means were performed with Bonferroni adjustment for multiple testing. Finally, as an exploratory analysis, logistic regression models were used to evaluate the prognostic value of early-pregnancy thyroid function for a composite adverse obstetric outcome, defined as the occurrence of at least one of the following: preterm delivery, labor dystocia (including prolonged or obstructed labor), premature rupture of membranes, placental abruption, or abnormal amniotic fluid volume. Models were adjusted for baseline maternal characteristics, including maternal age, parity, thyroid autoimmunity, iodized salt use, smoking habit, and preconception BMI category. First-trimester TSH and FT4 were each evaluated as the primary predictors of interest using clinically defined thresholds, namely TSH above the upper limit of normal (TSH > ULN) and FT4 below the lower limit of normal (FT4 0.05), making differential LT4 initiation an unlikely explanation for the observed longitudinal patterns. Consistent with this observation, sensitivity analyses, including all available thyroid function measurements either regardless of LT4 initiation or restricted to women who never initiated LT4 during pregnancy, yielded results consistent in direction and overall magnitude with those observed in the primary longitudinal analyses for both TSH and FT4 (Supplementary Tables S2 and S3; Supplementary Figs S1 and S2). Finally, given the greater between-group separation observed for TSH trajectories compared with FT4 trajectories, we examined whether the physiological relationship between FT4 and TSH differed according to preconception BMI. A significant inverse correlation was observed between FT4 and TSH concentrations in the overall cohort during the first trimester (Spearman’s rank correlation coefficient, ρ = −0.383; P < 0.0001). When stratified by p-BMI, the negative correlation proved strongest in NW women (ρ = −0.430; P < 0.0001) and attenuated in OW (ρ = −0.226; P = 0.001) and OB (ρ = −0.193; P = 0.044) groups (Fig. 3). First-trimester correlations between FT4 and TSH levels by preconception BMI category. Scatterplots displaying the inverse correlation between serum FT4 and TSH concentrations measured at ≤13 gestational weeks in women with normal weight (NW), overweight (OW), and obesity (OB). Spearman’s rank correlation coefficients (ρ) and corresponding P-values are shown in each panel. Fisher’s Z-tests formally confirmed that, compared with NW, the FT4–TSH relationship in the first trimester was significantly weaker in both OW and OB groups (P = 0.005 and P = 0.010, respectively), with no difference between OW and OB. The above-mentioned findings remained largely unchanged when limited to TPOAb-negative women (whole cohort: ρ = −0.333; P < 0.0001; NW: ρ = −0.397; P < 0.0001), although the correlations in the OW and OB groups were no longer statistically significant (ρ = −0.090; P = 0.262 and ρ = −0.140; P = 0.180, respectively). Similar patterns were observed in the second and third trimesters, with a persistent inverse association between FT4 and TSH across p-BMI categories. The attenuation of the FT4–TSH relationship in OW and OB women remained evident at later gestational stages, with borderline statistical significance in the OB group at the third-trimester assessment only (P = 0.050). First-trimester thyroid parameters and obstetric outcomes Pregnancy outcome data were available for 963 of 1,107 women (87.0%). Overall, 249 of 963 women (25.9%) experienced at least one adverse event included in the composite obstetric outcome. The distribution of obstetric complications was similar across p-BMI categories (172/696 (24.7%) in the NW group, 53/175 (30.3%) in the OW group, and 24/92 (26.1%) in the OB group; P = 0.32). Univariate logistic regression analyses were performed to explore the association between first-trimester thyroid parameters and the composite obstetric outcome, together with selected baseline maternal characteristics (Table 4). Both TSH > ULN and FT4 < LNL in the first trimester were associated with an increased risk of adverse obstetric outcomes, with a larger effect size observed for low FT4 (OR = 2.07, 95% CI: 1.36–3.14; P < 0.0001) compared with high TSH (OR = 1.71, 95% CI: 1.05–2.72; P = 0.027). Table 4Univariate regression analyses for composite obstetric outcome – exploratory analysis. Bold values indicate statistically significant associations (P < 0.05).CovariatesOR95% CIP-valueAge1.000.97–1.030.857Thyroid autoimmunity1.300.88–1.890.177Iodized salt0.890.66–1.180.410Parity0.780.64–0.94 0.011 Smoking habit0.990.76–1.280.941OW1.320.91–1.900.133OB1.070.64–1.740.774FT4 < LNL2.071.36–3.14 <0.0001 TSH > ULN1.711.05–2.72 0.027 OW, overweight; OB, obesity; ULN, upper limit of normal; LNL, lower limit of normal. Univariate regression analyses for composite obstetric outcome – exploratory analysis. Bold values indicate statistically significant associations (P < 0.05). OW, overweight; OB, obesity; ULN, upper limit of normal; LNL, lower limit of normal. In multivariable logistic regression models separately evaluating first-trimester TSH or FT4 (Table 5), first-trimester FT4 < LNL remained independently associated with an increased risk of adverse obstetric outcomes (OR = 2.04, 95% CI: 1.32–3.11; P = 0.001). The association between first-trimester TSH > ULN and the composite obstetric outcome was weaker but remained statistically significant (OR = 1.71, 95% CI: 1.04–2.76; P = 0.029). Table 5Multivariable logistic regression models evaluating first-trimester thyroid parameters as predictors of a composite adverse obstetric outcome – exploratory analysis. Bold values indicate statistically significant associations (P < 0.05).Model A – TSH > ULNModel B – FT4 < LNLOR95% CIP-valueOR95% CIP-valueFirst-trimester thyroid parameters* TSH > ULN1.711.04–2.76 0.029 - - FT4 < LNL - - 2.041.32–3.11 0.001 Baseline maternal characteristics Age (years)1.020.99–1.060.1861.020.99–1.050.224 Thyroid autoimmunity1.220.82–1.800.3101.220.81–1.790.327 Iodized salt use0.920.69–1.240.5990.920.68–1.230.571 Parity0.710.57–0.88 0.002 0.720.58–0.89 0.003 Smoking habit1.010.77–1.310.9401.040.79–1.360.760 Preconception BMI: overweight1.390.95–2.000.0851.430.97–2.070.064 Preconception BMI: obesity1.080.64–1.760.7691.120.66–1.840.661*Primary predictors.ULN, upper limit of normal; LNL, lower limit of normal. Each primary predictor was not included in the alternate model. Multivariable logistic regression models evaluating first-trimester thyroid parameters as predictors of a composite adverse obstetric outcome – exploratory analysis. Bold values indicate statistically significant associations (P < 0.05). Primary predictors. ULN, upper limit of normal; LNL, lower limit of normal. Each primary predictor was not included in the alternate model. Discussion In this large retrospective longitudinal cohort, we examined the association between preconception BMI and maternal thyroid function over the course of gestation and whether early thyroid alterations were linked to subsequent obstetric risk. Across more than 1,000 pregnancies with repeated thyroid assessments, higher p-BMI was consistently associated with subtle but persistent shifts in TSH and FT4, as well as with a weaker inverse FT4–TSH relationship across p-BMI categories. Notably, early-pregnancy FT4, rather than TSH, emerged as the thyroid parameter most strongly associated with obstetric complications, supporting the notion that even mild impairments in thyroxine availability at conception/early gestation may carry clinical relevance even after accounting for preconception BMI (11, 15, 25). Consistent with the existing literature, we observed significantly lower FT4 and higher TSH concentrations at early pregnancy in women with overweight and obesity compared to their normal-weight counterparts (15, 16, 19, 21, 32). Most prior studies, however, evaluated thyroid function at a single time point, limiting the ability to characterize temporal dynamics. Among the few longitudinal investigations, studies from both Pop et al. (19) and Collares et al. (16) reported BMI-related differences in early gestation and described subsequent gestational patterns. However, their primary focus was on gestational weight gain or metabolic outcomes rather than on thyroidal trajectories per se. In contrast, our study provides a longitudinal characterization of thyroid function across all trimesters, showing that p-BMI-related differences in FT4 and TSH are already evident by the first trimester and persist throughout pregnancy. The absence of significant BMI × trimester interaction in our GEE models indicates that women with elevated BMI maintain a persistently less favorable hormonal profile. In addition, the observed differences remained largely unchanged after excluding women with thyroid autoimmunity, suggesting non-autoimmune, BMI-related differences in thyroid function not specifically examined in prior longitudinal studies. Interestingly, we observed only limited differences between overweight and obese women, with largely similar pattern of thyroid function disruption in both groups. The above-mentioned observation may suggest a continuum across increasing adiposity and thyroid regulation. Although few studies have directly compared thyroid function trajectories between overweight and obese pregnant women (19), similar patterns of TSH elevation and/or lower FT4 values have been reported across the spectrum of increased BMI in pregnancy, often without a clear distinction between overweight and obesity categories (15, 16, 17, 18, 21, 32, 33, 34). The progressive attenuation of the physiological inverse correlation between FT4 and TSH with increasing p-BMI categories may reflect a potential disruption in HPT axis feedback sensitivity in the setting of excessive adiposity. Several mechanisms may underlie this pattern (8). Increasing adiposity has been linked to altered HPT axis regulation through leptin-mediated TRH stimulation, effects on anorexigenic pathways, and modulation of deiodinase activity (8, 35, 36, 37, 38). Reduced TSH receptor and TRα-1 expression in adipose tissue in severe obesity further supports peripheral hormone resistance (39, 40, 41). Low-grade inflammation and cytokine-mediated inhibition of thyroidal iodide uptake may additionally affect hormonogenesis (41, 42, 43, 44), potentially contributing to the lower FT4 levels observed in our cohort and others (19, 20, 22, 26, 45, 46). Beyond the pathophysiological mechanisms, the observed pattern of BMI-related disruption in early gestational thyroid function is clinically meaningful when viewed in the broader context of pregnancy adaptation. Women with overweight and obesity appear to enter pregnancy with a less favorable thyroidal set-point and their subsequent hormonal trajectories across gestation largely reflect the baseline differences observed at the first trimester. Importantly, however, our longitudinal design does not allow us to determine whether these gestational trajectories themselves exert an independent influence on obstetric outcomes, and our outcome analyses were intentionally limited to first-trimester thyroid parameters. Future appropriately designed studies with repeated hormone-outcome linkages would be needed to clarify whether deviations in thyroid trajectories add prognostic value beyond early hormone levels. It is worth noting that the lower FT4 levels observed in women with higher BMI were not only statistically significant but also clinically relevant, as they resulted in a higher prevalence of isolated hypothyroxinemia, particularly among women with obesity in the early second trimester. This condition has been associated with adverse fetal neurodevelopmental outcomes and abnormal placental function (11), particularly when occurring early and in the context of increased maternal adiposity (47). In this view, our study adds to the growing body of evidence linking maternal thyroid status at early gestation to pregnancy complications. Women with reduced FT4 levels during the first trimester exhibited an approximately twofold increased risk of experiencing obstetric complications, even after adjusting for p-BMI and a range of other predictors known from the literature to be associated with adverse pregnancy outcomes (e.g., maternal age, parity, smoking, autoimmunity) (46). Notably, thyroid autoimmunity was not independently associated with obstetric outcomes in our cohort whereas low FT4 levels were, reinforcing the notion that thyroid hormone availability in early pregnancy may represent a more relevant determinant of obstetric risk than thyroid autoimmunity per se (24, 48, 49). In this context, it is worth noting that all women in our cohort underwent thyroid function testing in early pregnancy and were managed upon diagnosis of thyroid dysfunction in accordance with available guidelines. Levothyroxine treatment initiated after the first-trimester assessment may have attenuated the observed associations, and the reported estimates should, therefore, be interpreted as conservative estimates of the association between early pregnancy thyroid status and obstetric outcomes. Although causality cannot be established, our findings suggest that early pregnancy may already represent a relatively late stage for mitigating the downstream effects of impaired thyroid function. If confirmed, this relationship would support the value of preconception screening and early thyroid optimization, particularly in higher-risk populations, such as women with overweight or obesity or those with thyroid autoimmunity (50, 51). The main strengths of this study include its large, well-characterized cohort, the standardized and consistent laboratory methods applied across the entire study period, and the availability of repeated thyroid function measurements. Unlike large registry-based studies that focus exclusively on first-trimester thyroid function, our dataset included repeated hormone assessments across pregnancy, which allowed us to describe early BMI-related thyroid alterations along with real-world trajectories of TSH and FT4 in a sizeable population of unselected pregnant women, an aspect not captured in previous work. Several limitations should also be acknowledged. First, as an observational retrospective study, causal inference or directionality could not be established and residual confounding may persist despite adjustment. In addition, statistical power to detect differences in obstetric outcomes across BMI strata was limited by the relatively small number of women with obesity. Consequently, analyses of p-BMI-related differences in obstetric risk were considered exploratory and should be interpreted with appropriate caution, with the composite obstetric outcome serving as an overall indicator of obstetric burden. Furthermore, not all pregnancy outcomes could be consistently included due to limitations in outcome ascertainment and data availability inherent to the retrospective longitudinal design. Information on pregnancy losses was not systematically available with sufficient detail for reliable classification, and pregnancies ending in early loss could not contribute to the longitudinal analyses by design. In addition, gestational diabetes and hypertensive disorders were not included because diagnostic criteria and outcome ascertainment were not uniform across the study period, and raw oral glucose tolerance test data were not systematically available, precluding retrospective harmonization. FT3 concentrations and biomarkers of leptin or inflammation were not measured, limiting the pathophysiologic interpretation of the observed BMI-related alterations. Finally, as the source and exact timing of preconception weight assessment relative to conception were not systematically recorded, some degree of measurement error cannot be excluded. Nonetheless, the use of preconception BMI as a categorical stratification variable, rather than as a continuous measure, together with the application of internal data consistency checks, is likely to have reduced the potential impact of measurement error on group classification. Finally, pregnancy-specific thyroid reference intervals were not available for the study population and fixed assay-specific thresholds were applied. However, the primary analyses focused on continuous FT4 and TSH concentrations and their longitudinal trajectories across p-BMI strata rather than on categorical diagnoses, limiting the impact of fixed cut-offs on the main findings. Reference limits were used only to define baseline thyroid status in early pregnancy, assessed at a median gestational age of approximately 10 weeks, when gestational alterations in FT4 measurements are still relatively limited. Nevertheless, our results support the notion that maternal BMI and early gestational thyroid function jointly contribute to pregnancy risk, even in the absence of overt thyroid disease, and underscore the need for future studies aimed at clarifying whether targeted preconceptional or early-pregnancy interventions may improve maternal and obstetric outcomes. By delineating distinct longitudinal patterns of thyroid adaptation across BMI strata, this study offers pathophysiological insights that may help refine early-pregnancy risk stratification and inform the selection of target populations for future interventional studies. Supplementary materials Declaration of interest The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported. Funding This research did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector. Author contribution statement MM: Conceptualization, Data Curation, Formal Analysis, Investigation, Writing – Original Draft, Writing – Review & Editing (lead); AA: Formal Analysis (lead), Writing – Original Draft, Writing – Review & Editing; LC: Formal Analysis (supporting), Writing – Review & Editing; GP: Data Curation, Formal Analysis, Writing – Review & Editing (supporting); LGP, LB, SM, GS, SDA, MR RG: Data Curation, Formal Analysis, Writing – Review & Editing (supporting); FV, SC: Conceptualization, Formal Analysis, Writing – Review & Editing (supporting) – Shared senior authorship. All authors reviewed and approved the final manuscript.