Letter to the Editor: Re-evaluating fixed thyroid thresholds in postpartum depression risk stratification

Abstract

We commend Chen et al recently published a study in World Journal of Psychiatry, for investigating maternal thyroid-stimulating hormone (TSH)/thyroid peroxidase antibody (TPOAb), associations with postpartum depression (PPD) and adverse pregnancy outcomes (APOs) in their retrospective analysis. However, the reliance on universal biomarker thresholds (TSH ≥ 3 mIU/L, TPOAb ≥ 7 U/mL) as independent APO predictors warrants scrutiny. Thyroid physiology exhibits significant inter-individual variability during pregnancy, and fixed thresholds may overlook subclinical dysfunction in women with TSH or TPOAb levels below these cutoffs. The study’s exclusion of participants with overt thyroid dysfunction introduces a selection bias, potentially underestimating the true risk spectrum. Moreover, the use of single-timepoint measurements at 6-7 weeks postpartum ignores dynamic fluctuations in thyroid autoimmunity across gestation, which recent evidence links to the course of PPD symptoms. We urge future studies to adopt individualized, trimester-specific reference ranges and serial biomarker assessments to better capture nonlinear relationships between thyroid dysfunction, PPD, and APOs. This approach will help optimize clinical risk prediction models for diverse populations.

Key Words: Postpartum depression; Adverse pregnancy outcomes; Risk stratification; Thyroid thresholds

Core Tip: Chen et al identified thyroid-stimulating hormone ≥ 3 mIU/L and thyroid peroxidase antibody ≥ 7 U/mL as independent predictors for adverse pregnancy outcomes (APOs) in postpartum depression (PPD). While this underscores the thyroid-PPD link, reliance on universal, fixed thresholds is problematic. Pregnancy induces profound, individualized thyroid physiological changes, making static cut-offs potentially misleading for risk stratification. Future research must prioritize trimester-specific reference ranges and longitudinal biomarker profiling to capture dynamic thyroid-autoimmune interactions and refine personalized prediction models for PPD and APOs.

TO THE EDITOR

We read with interest the retrospective study by Chen et al[1] published in World Journal of Psychiatry, which investigated the associations between maternal thyroid-stimulating hormone (TSH), thyroid peroxidase antibody (TPOAb), postpartum depression (PPD), and adverse pregnancy outcomes (APOs)[1]. The authors reported significantly elevated serum TSH and TPOAb levels in 83 PPD patients compared to 80 controls, with positive correlations to depression scale scores, and identified TSH ≥ 3 mIU/L and TPOAb ≥ 7 U/mL as independent predictors for APOs[1]. This work valuably highlights the role of thyroid function in perinatal mental health. However, the methodological reliance on universal, fixed biomarker thresholds for risk stratification raises significant concerns regarding physiological validity and clinical generalizability, which we wish to discuss.

STUDY OVERVIEW AND DISCUSSION

Chen et al[1] conducted a case-control study comparing 83 PPD and 80 control women at 6-7 weeks postpartum[1]. Their key findings include higher mean TSH and TPOAb in the PPD group (data presented graphically), significant positive correlations between these markers and Edinburgh Postnatal Depression Scale (EPDS)/Self-Rating Depression Scale (SDS) scores (TSH-EPDS: r = 0.470, P < 0.001; TPOAb-SDS: r = 0.334, P = 0.002), and a notably higher overall APO incidence in the PPD group (27.71% vs 7.50%, P < 0.001)[1]. Multivariable analysis identified multiparity (≥ 3), PPD diagnosis, TSH ≥ 3 mIU/L [odds ratio (OR) = 3.333], and TPOAb ≥ 7 U/mL (OR = 3.958) as independent APO predictors[1].

The study’s strength lies in its direct examination of thyroid markers within a PPD clinical context. However, the application of fixed thresholds (TSH ≥ 3 mIU/L, TPOAb ≥ 7 U/mL) is a critical limitation. We acknowledge that the use of fixed thresholds in retrospective analyses or clinical settings with limited resources can provide valuable preliminary data; however, their limitations for precise, individualized risk prediction necessitate more refined methods for future research. Pregnancy necessitates a major physiological adaptation of the hypothalamic-pituitary-thyroid axis. Leading guidelines, including the 2017 American Thyroid Association guidelines, strongly recommend using population-based, gestation-specific reference ranges for TSH during pregnancy, as the upper reference limit typically decreases in the first trimester[2]. Similarly, the 2014 European Thyroid Association guidelines advocate for the use of trimester-specific reference intervals[3]. Applying a fixed, non-pregnancy-adjusted TSH threshold (3 mIU/L) to postpartum women risks misclassification, as gestational physiology alters thyroid reference ranges. For instance, a TSH of 2.5-3.0 mIU/L might be within the normal non-pregnant range but could represent relative hypothyroidism in early pregnancy for a particular individual based on her own pre-pregnancy baseline or population-derived trimester-specific norms[4]. Similarly, while TPOAb positivity is a recognized risk factor for thyroid dysfunction and APOs[5], the clinical significance of the specific fixed cutoff of 7 U/mL reported by Chen et al[1], particularly in relation to APO risk, requires validation. This value (7 U/mL) is notably below the common clinical positivity threshold (e.g., 9 IU/mL used in many assays), and its specific predictive value for APOs requires further validation in the context of pregnancy[6-8].

STRENGTHS AND LIMITATIONS

The study’s retrospective design using clinically collected data offers real-world insights. The significant correlations between thyroid markers and depression scores reinforce the biological plausibility of the thyroid-PPD link, which may involve neuroendocrine and immune pathways[9]. A recent systematic review also highlighted the potential role of neuroinflammation and HPA axis dysregulation in this association[10].

Nevertheless, several limitations constrain the interpretation of the fixed thresholds. First, the exclusion criteria included “thyroid dysfunction”, but the definition used is not specified. If overt dysfunction was excluded based on standard clinical criteria, the cohort inherently represents a population with “normal” thyroid function by those criteria. Identifying TSH ≥ 3 mIU/L as a risk factor within this "normal" group suggests subclinical dysfunction, but the threshold itself may not be optimal for this specific postpartum context. Second, thyroid autoimmunity is dynamic. TPOAb levels fluctuate throughout pregnancy and into the postpartum period. Therefore, a single measurement after delivery cannot accurately represent the autoimmune activity during critical phases, such as placental development or the early postpartum window when postpartum thyroiditis typically occurs[11]. Furthermore, the 6-7 weeks postpartum timepoint is particularly relevant for assessing postpartum thyroiditis, but it may not fully reflect gestational autoimmune activity or its potential impact on earlier pregnancy outcomes. Third, the pathophysiological link between these specific threshold levels and APOs remains mechanistically underexplored. Thyroid autoimmunity is associated with an increased risk of miscarriage and preterm birth. The proposed mechanisms for this association primarily involve placental dysfunction, which can be triggered by autoimmune inflammation or subsequent subclinical hypothyroidism[12,13]. The direct contribution of a TPOAb level of 7 U/mL-a value typically considered very low or negative in most clinical assays-to APO risk requires further biological validation.

FUTURE RESEARCH DIRECTIONS

Building on the findings of Chen et al[1], future research should pursue a more nuanced and personalized assessment of thyroid-related risk. We propose several key directions for improving thyroid function assessment in pregnancy. First, trimester-specific and individualized reference ranges for TSH should be adopted, utilizing data from local iodine-sufficient populations and considering pre-pregnancy baselines when available[2]. This approach would help distinguish women with genuinely high-normal TSH levels who may be at elevated risk. Second, implementing longitudinal biomarker profiling through serial measurements of TSH, free thyroxine, and TPOAb across trimesters and into the postpartum period is essential. This could be operationalized through prospective cohort studies with serial sampling at each trimester and postpartum time point. Evidence indicates that monitoring the longitudinal trajectory of thyroid function is more predictive of postpartum thyroid dysfunction and associated mood disorders than relying on isolated measurements. For instance, a positive TPOAb status during pregnancy has been consistently linked to an increased risk of subsequent PPD[14]. Third, comprehensive risk models should integrate dynamic thyroid parameters with established psychosocial risk factors. This combined approach can significantly enhance the prediction of PPD and APOs. In line with this, recent large-cohort studies have further demonstrated the critical importance of utilizing population- and trimester-specific reference intervals for accurate thyroid status assessment during pregnancy, rather than relying on fixed thresholds[15], and have quantified the substantial misclassification that occurs when non-pregnant reference ranges are applied to pregnant populations[16]. Analytical approaches like machine learning could model these complex, non-linear interactions by integrating serial thyroid parameter trajectories with other clinical and psychosocial variables to improve predictive accuracy. Finally, future research must elucidate the mechanistic pathways linking low-titer TPOAb and subtle thyroid hormone deficiency to placental dysfunction and fetal neuroinflammation. This will provide the biological evidence needed to substantiate the observed epidemiological associations.

CONCLUSION

Chen et al[1] provide valuable data reinforcing the connection between thyroid markers, PPD, and APOs. However, the use of fixed, universal thresholds for TSH and TPOAb may oversimplify the complex and individualized physiology of pregnancy. To advance towards precision medicine in perinatal psychiatry and obstetrics, subsequent research must transition from static cut-offs to dynamic, trimester-adjusted assessments. This approach will enhance the accuracy of risk prediction, optimize timely intervention, and ultimately improve maternal and neonatal outcomes.