Lead Exposure, Glucocorticoids, and Physiological Stress Across the Life Course: A Systematic Review

Abstract

The biological pathways linking lead exposure to adverse outcomes are beginning to be understood. Rodent models suggest lead exposure induces dysfunction within the hypothalamic-pituitary-adrenal (HPA) axis and glucocorticoid regulation, a primary physiological stress response system. Over time, HPA axis and glucocorticoid dysfunction has been associated with adverse neurocognitive and cardiometabolic health, much like lead exposure. This systematic review utilized PRISMA guidelines to synthesize the literature regarding associations between lead exposure and downstream effector hormones of the HPA axis, including cortisol, a glucocorticoid, and dehydroepiandrosterone (DHEA), a glucocorticoid antagonist. We additionally determined the state of the evidence regarding lead exposure and allostatic load, a measure of cumulative body burden resultant of HPA axis and glucocorticoid dysfunction. A total of 18 articles were included in the review: 16 assessed cortisol or DHEA and 3 assessed allostatic load. Generally, the few available child studies suggest a significant association between early life lead exposure and altered cortisol, potentially suggesting the impact of developmental exposure. In adulthood, only cross sectional studies were available. These reported significant associations between lead and reduced cortisol awakening response and increased cortisol reactivity, but few associations with fasting serum cortisol. Two studies reported significant associations between increasing lead exposure and allostatic load in adults and another between early life lead exposure and adolescent allostatic load. The paucity of studies examining associations between lead exposure and allostatic load or DHEA and overall heterogeneity of allostatic load measurements limit conclusions. However, these findings cautiously suggest associations between lead and dysregulation of physiological stress pathways (i.e., glucocorticoids) as seen through cortisol measurement in children and adults. Future research would help to elucidate these associations and could further examine the physiological stress pathway as a mediator between lead exposure and detrimental health outcomes.

Graphical Abstract:

Introduction

No level of lead exposure is considered safe as even small amounts of lead have been associated with deleterious health and developmental outcomes (Ruckart et al., 2021). Despite worldwide prevention efforts, global estimates suggest that 1 in 3 children, or around 800 million, have elevated blood lead levels (BLLs) (UNICEF & Pure Earth, 2020). Children are particularly vulnerable to lead exposure due to frequent hand to mouth activities, overall oral curiosity, and increased absorption rates compared to adults (World Health Organization, 2021). In children, lead exposure has been associated with a multitude of negative outcomes such as lower IQ scores (Heidari et al., 2022), increased aggression (Tlotleng et al., 2022), and elevated blood pressure (Zhang et al., 2012). Likewise, exposure during adulthood has been associated with harmful health outcomes including reduced cognitive function (Sasaki & Carpenter, 2022) and increased risk for cardiovascular and diabetic kidney disease (Wan et al., 2021). Effects may persist across the life course, given childhood lead exposure has been associated with decreased brain volume in adults (Cecil et al., 2008).

Understanding the biological pathways that may drive lead’s influence on neurocognitive and cardiometabolic outcomes is crucial to developing interventions to mitigate risk. Rodent studies suggest the body’s physiological stress response systems, including the hypothalamic-pituitary-adrenal (HPA) axis, as a plausible biological pathway (Cory-Slechta et al., 2008). The HPA axis, in conjunction with the autonomic nervous system, is responsible for eliciting a physiological response to perceived external stressors (Smith & Vale, 2006). Dysfunction within the HPA axis may lead to irregular baseline and responsivity of glucocorticoids, which over time has been associated with detrimental neurocognitive and cardiometabolic outcomes, much like lead exposure (McEwen, 2017). In rodents, prenatal lead exposure has been associated with increased corticosterone in male offspring at 9 weeks (i.e., adolescence) (Cory-Slechta et al., 2004) and alterations in dexamethasone efficacy during an injection stress test for males and females at 5 months (i.e., adulthood), suggesting hypercortisolism and alterations to the negative feedback loop that governs cortisol release (Rossi-George et al., 2009).

In humans, cortisol, a glucocorticoid, and dehydroepiandrosterone (DHEA), a steroid hormone and glucocorticoid antagonist, are the two primary downstream effector hormones released from the adrenal cortex in response to stress and are some of the most commonly studied downstream effects of the HPA axis (Charoensri et al., 2017; Dutheil et al., 2021; Spencer & Deak, 2017). Both biomarkers can be collected via saliva, blood, hair, or urine to assess different functional aspects of the HPA axis including baseline, diurnal rhythms, and reactivity. Cortisol levels follow a diurnal rhythm, where levels peak around 30 minutes after awakening and gradually decrease throughout the day. The salivary cortisol awakening response (CAR), (Fries et al., 2009; Kobayashi et al., 2017) is a frequently used measure of HPA axis functionality in which a blunted CAR has been linked with adverse health outcomes (Kuehl et al., 2015). Cortisol can also be measured in response to an external stressor and conceptually measures the responsivity/reactivity of the HPA axis. Cortisol is collected at baseline, immediately following a stressor, and then at multiple timepoints post-stressor to assess a return to homeostasis. Altered cortisol responses, including both accelerated and blunted responses to both the initial stressor and a return to baseline, can be measured using area under the curve techniques (AUC) and are indicative of an altered HPA axis (Miller et al., 2018). DHEA has anti-glucocorticoid properties, playing a role in the negative feedback loop of the HPA axis (McNelis et al., 2013). The ratio between cortisol/DHEA can, therefore, be used as an indicator of HPA axis function in addition to individual measures of either hormone (Basson et al., 2019; Qiao et al., 2017; Wolkowitz et al., 2001).

A more cumulative measure of chronic stress on multiple biological systems, allostatic load, is thought to encapsulate long-term HPA axis dysfunction, body burden, and is a predictor of chronic disease and all-cause mortality (Castagné et al., 2018; Parker et al., 2022). In brief, allostatic load quantifies the physiological wear and tear on the body resulting from chronic stress (McEwen, 1998; Seeman et al., 1997), which can be assessed through various markers of physiological dysregulation (Rodriquez et al., 2019). There is significant variability in allostatic load index construction making it difficult to quantify. The original allostatic load index created by Seeman et al. (1997) included 10 biomarkers focusing on primary mediators of the overall stress response (i.e., serum dehydroepiandrosterone sulfate, and 12 hour urinary cortisol, epinephrine, and norepinephrine) and secondary outcomes (i.e., systolic and diastolic blood pressure, waist-hip ratio, serum high-density lipoprotein, total cholesterol, plasma glycosylated hemoglobin) (Seeman et al., 1997). Over time, researchers quantifying allostatic load have included additional biomarkers and most often include biomarkers from four distinct categories including several biological systems: (1) cardiovascular, (2) metabolic, (3) inflammatory, and (4) neuroendocrine (Beckie, 2012; Rodriquez et al., 2019; Whelan et al., 2021). These biomarkers are categorized as either “high” or “low” risk using binary scoring based on either sample distributions or clinical cut-off values, if available, and summed to create an allostatic load score.

As HPA-axis and glucocorticoid dysfunction are associated with long-term adverse neurocognitive and health outcomes, understanding relationships between lead exposure and glucocorticoid function will help to elucidate potential biological pathways. Currently, no synthesis exists summarizing what is known about the relationship between lead exposure and glucocorticoids in humans across the life course. While dysregulation of physiological stress pathways such as the HPA axis, overtime, induces cumulative measures of stress such as allostatic load, few studies exist testing associations between lead and allostatic load. Therefore, the aims of this review were threefold: (1) To comprehensively synthesize the empirical research on the association between lead exposure and primary effector hormones (i.e., glucocorticoids and antagonists) of the HPA axis across the life course in humans; (2) To determine the state of the literature associating lead exposure with allostatic load; and (3) To identify methodological limitations and gaps in the literature to make recommendations for future research.

Methods

Data sources and search methodology

We followed Cochran’s Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) checklist (Moher et al., 2009). We searched databases PubMed, PsychINFO, Web of Science, and Embase in November of 2023. Search terms included ‘lead exposure’, ‘blood lead levels’, ‘allostatic load’, ‘chronic stress’, ‘physiological stress’, ‘toxic stress’, ‘cortisol’ and ‘dehydroepiandrosterone’ (Supplemental Table 1). We created free text strings and included subject headings (i.e., MeSH terms) when available in databases.

Literature screening

After the initial search, two reviewers systematically reviewed studies first by title and abstract and then by full text against inclusion and exclusion criteria. Screening was completed in the online program Covidence (Covidence, 2022). To be included in the review, studies had to be: 1) human studies ranging from prenatal to adulthood 2) an original study with lead exposure measured via a physiological biomarker (e.g., blood, bone, cord measurement); 3) outcomes assessed include cortisol or DHEA/DHEA sulfate and as a secondary outcome any studies including allostatic load as an outcome; 4) reported in English as a peer-reviewed publication (i.e., Published paper, dissertations/thesis). There were no limitations on years or geographic location. We excluded studies that were conducted in animal models and any study in which lead exposure was measured via contamination source (e.g., water, pollution, soil etc.) and not a physiological biomarker. Studies including lead as a mediator in a relationship or intervention designs were excluded. Studies examining oxidative stress, self-reported stress, and any other outcome were excluded. Finally, studies published as case studies, editorials, poster presentations, opinion pieces were excluded (Supplemental Table 2).

Data extraction

Data extracted included research design, age at exposure and outcome assessment, study location, lead exposure measure and mean/median levels, biomarker outcomes (i.e., cortisol/DHEA biomarkers or allostatic load), and major findings. Studies were first separated by whether individual biomarkers or allostatic load were assessed. Next, studies were separated into those conducted in childhood (<10 years old), early to late adolescence (11–18 years old), and adults (>18 years old). Quality was assessed using the National Institute of Health’s (NIH) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (National Heart, 2021).

Results

Search and selection results

The initial search, including the English and human studies criteria, yielded 518 papers. Duplicates were removed (n=57) for a yield of 461 unique papers to be screened. After article screening by the two reviewers, 18 papers were available for inclusion, 16 of which included cortisol or DHEA measurements and three of which included an allostatic load outcome (Figure 1 PRISMA; one paper included both cortisol and allostatic load). Reasons for exclusion while screening full text studies included not assessing a predetermined physiological stress outcome (n=11) or examining a self-reported stress outcome (n=1), not assessing a true lead exposure biomarker (n=2), including an intervention (n=1), and being an opinion or editorial piece (n=5). Table 1 includes characteristics of included studies, Table 2 major findings of included studies, and Table 3 provides quality grading results.

Figure 1.

PRISMA Diagram

Table 1.

Study characteristics in included studies, organized by outcome and life course stage

Domains	Study Characteristics	Study Design	Lead Measurement Method	Lead Levels	Stress Measures	Covariates
Cortisol and DHEA Child
(Cai et al., 2019)	Chinese children (ages 3 to 6 years; n=574) from Guiyu (mean age 4.79 years; n= 358; recycling town) and Haojiang (mean age 4.62 years; n= 216; town with no e-waste recycling activity)	Cross-sectional	Blood lead levels	Guiyu = Median 4.88 mcg/dl Haojiang= Median 3.47 mcg/dl	Serum (venous) cortisol (early morning, fasting).	Adjusted for sex, height, weight, current use of medications, and monthly household income.
(Gump et al., 2008)	Children from Oswego Children’s Study (mean age 9.5 years; n=91 females and 78 males)	Longitudinal	Prenatal lead (cord blood) and postnatal blood lead levels (mean age 2.6 years)	Range ≤1–13.10 mcg/dl Prenatal quartiles= 1st: ≤ 1 mcg/dl 2nd: 1.1–1.4 mcg/dl 3rd : 1.5–1.9 mcg/dl 4th: 2.0–6.3 mcg/dl Postnatal quartiles= 1st: 1.5–2.8 mcg/dl 2nd: 2.9–4.1 mcg/dl 3rd: 4.2–5.4 mcg/dl 4th: 5.5–13.1 mcg/dl	Salivary cortisol during stress protocol (baseline; 21 and 40 after the start of the task, and in recovery 60 minutes after task)	Adjusted for paternal height and weight, maternal pre-pregnancy weight and height, maternal weight gain during pregnancy, maternal reported illness during pregnancy, obstetric complications (using the Ballard and a measure of optimality), head circumference at birth, birth weight, gestational age, maternal substance use during pregnancy (e.g., cigarettes, alcohol), the Home Observation for Measurement of the Environment (HOME), the Clinical Epidemiological Studies–Depression (CES-D; Radloff 1977) inventory, the Four Factor Index of Social Status (Hollingshead 1975) to measure socioeconomic status (SES), and body mass index (BMI; weight (kilograms)/height (meters).
(Tamayo y Ortiz et al., 2016)	PROGRESS cohort study in Mexico City (12 months; n=255) (18–24 months; n=150)	Longitudinal	Maternal pregnancy blood lead levels (2nd and 3rd trimester), Maternal bone lead (tibia) collected 1 month postpartum.	For 12 month infants: 2nd trimester blood lead (3.5 μg/dl); 3rd trimester blood lead (3.7 μg/dl); Tibia lead 5.6 (μg/g) For 18–24 month infants: 2nd trimester blood lead (3.9 μg/dl); 3rd trimester blood lead (4.2 μg/dl); Tibia lead 4.9 (μg/g)	4 saliva cortisol samples from infants for 2 days at home (early morning: before eating or diaper changing, mid-morning: between 1100–1300, mid-afternoon: between 1500–1700, and night: at least 30 minutes after dinner and before bed).	Adjusted for sex, maternal age at delivery, years of schooling, pre-pregnancy BMI.
Adolescent
(Castiello et al., 2020)	INMA cohort in the province of Granada, Southern Spain (boys ages 15–17; n=133)	Cross-sectional	Urine lead (first void)	Geometric mean= 0.24 mcg/g (creatinine adjusted)	Serum dehydroepiandrosterone and cortisol collected between 1700–1900.	Adjusted for urinary creatinine, age, season of blood sampling, hour of blood extraction, area of residence (urban or semi-urban/rural), waist to height ratio, pubic hair growth stage, and urinary metals simultaneously (As, Cd, Hg, Ni, Cr, Mn).
Adult
(Braun et al., 2014)	Ongoing prospective birth cohort in Mexico City (mean age 28 years; blood lead analysis n=873–918 women and bone lead n=568 –594 women)	Cross-sectional	Blood lead levels (2nd trimester). Tibia (cortical bone) and patella (trabecular bone) at 1 month postpartum.	Blood lead 2nd trimester mean= 3.7 mcg/dl Maternal tibia mean = 2.7 mcg/g Maternal patella mean= 4.6 mcg/g	Diurnal salivary cortisol collected 5x/day for 2 days during the week or weekend. Using passive drool technique collected upon awakening, 45 after waking, 4 hours after waking, 10 hours after waking, and at bedtime. Collected between gestational weeks 14–35; mean 19.7 weeks)	Adjusted for maternal age, marital status, years of education, BMI, gestational age at time of cortisol collection, parity, and smoking status during pregnancy.
(Butts et al., 2014)	The Study of Metals and Assisted Reproductive Technologies (SMART);52 women (35.7 years) and 28 men (37.9 years) completing IVF cycle.	Cross-sectional	Blood lead levels	Median 0.70 mcg/dl	Urine cortisol- all procedures completed between 900–1300	Adjusted for sex, age, BMI, and blood Hg and Cd biomarkers.
(Fortin et al., 2012)	Male participants (mean age 46.4 years; n= 76)	Cross-sectional	Blood lead and tibia bone lead	Blood lead median= 4 mcg/dl Bone lead median =13 mcg/g	Serum cortisol in response to and during recovery from a stressor (TSST) 1) baseline 2) during stress test at 5, 10, 15, and 20 minutes and 3) during recovery at 40, 60 and 75 minutes. Conducted in the early afternoon ~1225 hours.	Adjusted for age, education, social inhibition, negative affect, depression (CDI-II), and baseline cortisol (only for reactive models).
(Gustafson et al., 1989)	Moderately exposed male workers in Sweden (mean age 36; n= 25 exposed and n= 25 matched controls not occupationally exposed)	Case-control	Blood lead levels	Mean 1.9 mcmol/L exposed group (39 mcg/dl); Mean 0.2 mcmol/L non-exposed matched controls (4.14 mcg/dl)	Serum cortisol	No covariates.
(Kamińska et al., 2020)	Women in Poland (total mean age 53.23; n= 233) per-menopausal women as risk for metabolic syndrome (mean age 53.06; n= 64) or with metabolic syndrome (mean age 54.49; n=47) or control (mean age 52.84; n=122)	Cross-sectional	Blood lead levels	Mean 7.25 mcg/dl	Serum cortisol	No covariates in ANOVA.
(Klimenko et al., 2016)	40 participants (20 men and 20 women) TIA patients; plus additional 10 male and 10 female controls (mean age 57.2 years)	Case-control	Plasma lead	Male TIA: 0.1 mcg/dl Female TIA: 0.05 mcg/dl Male control: 0.04 mcg/dl Female control: 0.05 mcg/dl	Serum cortisol	No covariates in correlations.
(Ngueta et al., 2018)	Postmenopausal women (mean age 59 years; n=65)	Cross-sectional	Blood lead levels	Mean 2.41 mcg/dl	Salivary cortisol 5x/day for 3 consecutive week days: upon awakening, 30 minutes after awakening, 1400, 1600, and at bedtime. Samples averaged across 3 days. Salivary cortisol during induced stress test, TSST, collected 2 times at baseline, one post- 10 minute anticipatory period, and one sample 15 minutes after, and additional samples every 10 minutes (n=9 samples).	Adjusted for age, gender, waist-hip ratio, smoking status and income levels.
(Owsianowska et al., 2020)	Perimenopausal women (aged 44–65; n=233)	Cross-sectional	Blood lead levels	3 groups; Group 1 <5mcg/dl, Group 2 5–10 mcg/dl, and Group 3 >10 mcg/dl. Overall mean not reported.	Serum cortisol collected between 0700–0900 fasting.	No covariates in ANOVA.
(Pérez-Cadahía et al., 2008)	Healthy individuals that participated in the cleaning of coastal areas affected by the ‘Prestige’ oil spill (mean age 32 years; n=179)	Cross-sectional	Blood lead levels	Mean 59.13 (μgl^-1)	Plasma cortisol collected between 0800–0900 fasting.	No covariates- univariate models.
(Choi et al., 2022)	Mother-child pairs in CHECK cohort in Korea (mean age 33.4 years; n= 182)	Cross-sectional	Urinary lead levels	Mean 4.37 (mcg/L)	Urinary free cortisol and specific gravity	Adjusted for maternal age, pre-pregnancy BMI, gestational age (<39 weeks or ≥39 weeks, parity, delivery mode, infant sex, smoking during pregnancy (active and passive), drinking during pregnancy, and income. Also adjusted for specific gravity.
Zhang et al., 2023	Chinese adults (mean age 56.66 years; control area n=154, exposed area n=314)	Cross-sectional	Blood lead levels	Mean 13.37 (ng/mL)	Serum cortisol	Adjusted for age, sex, BMI, smoking, and alcohol consumption.
(Souza-Talarico et al., 2017)*	Brazilian adults (mean age 65.9 years; n=126)	Cross-sectional	Blood lead levels	Mean 2.1 mcg/dl	Saliva cortisol CAR (upon awakening, 30 min after wakening, 14:00 h, 16:00 h and at bedtime). Collected on two non-consecutive workdays.	Adjusted for age, gender, time of awakening, SES, Geriatric Depression Symptoms GDS, and Perceived stress scale (PSS) score.
Allostatic Load Adolescent
(Halabicky et al., 2023)	Mother-child pairs in the ELEMENT cohort (mean age at adolescent follow up 13.91 years; n=391)	Longitudinal	Bone and blood lead levels	Maternal tibia bone mean 9.07 (mcg/g) Maternal patella bone mean 11.41 (mcg/g) Maternal 1st trimester blood mean 5.44 (mcg/dl) Maternal 2nd trimester blood mean 4.89 (mcg/dl) Maternal 3rd trimester blood mean 5.40 (mcg/dl) Child 12 month blood mean 4.56 (mcg/dl) Child 24 month blood mean 4.71 (mcg/dl) Adoelscent blood mean 3.12 (mcg/dl)	Allostatic Load Index: cardiovascular (systolic blood pressure, diastolic blood pressure, total cholesterol, HDL, and triglycerides), metabolic (body mass index [BMI], homeostatic model assessment for insulin resistance [HOMA-IR], and waist circumference), immune (high sensitivity-CRP, IL-6, insulin like growth factor-1 [IGF1], and TNF-α), and neuroendocrine biomarkers (DHEAS, and plasma cortisol). ‘High risk’ was considered as values above the 75th percentile of sample values for each biomarker except HDL, cortisol, and DHEAS. For HDL, values below the 25th percentile were considered ‘high risk’. For cortisol and DHEAS, values below the 10th and above the 90th percentile were considered ‘high risk’.	Adjusted for age, sex, concurrent Pb, maternal years of schooling, and smoking during pregnancy
Adults
(Obeng-Gyasi et al., 2021)	NHANES (1999–2008) (ages 20–85 years; n=28,852)	Cross-sectional	Blood lead levels	Median 1.55 mcg/dl	Allostatic Load Index: Cardiovascular (SBP, DBP, triglycerides, HDL cholesterol, total cholesterol), inflammatory (CRP), and metabolic systems (BMI, hemoglobin A1C, albumin, creatinine clearance). AL markers were divided into quartiles based on their distribution within the database. High-risk was considered to be the top 25% in the distribution for all biomarkers apart from albumin, creatinine clearance, and HDL cholesterol, for which the bottom 25% of the distribution was considered to have the highest risk.	Adjusted for gender, BMI, smoking, alcohol consumption, country of birth, and income.
(Souza-Talarico et al., 2017)	Brazilian adults (mean age 65.9 years; n=126)	Cross-sectional	Blood lead levels	Mean 2.1 mcg/dl	Allostatic load index: Neuroendocrine (dehydroepiandrosterones), metabolic (glucose, triglycerides, total cholesterol, HDL),and anthropometric (BMI). Quartile ranges were created based on the clinical reference ranges for each biomarker. The index was calculated by summing the number of values equal or above the 75th percentile for total cholesterol, triglycerides, glucose and body mass index. For HDL and dehydroepiandrosterones values falling below the 25th percentile were considered “high risk” values. AL ranged from 0–5.	Adjusted for age, gender, time of awakening, SES, Geriatric Depression Symptoms GDS, and Perceived stress scale (PSS) score.

^*Also included an allostatic load index

Table 2.

Major findings from included studies, organized by outcome and life course stage

Studies	Study Location	Study Design	Major Findings
Cortisol and DHEA Child
(Cai et al., 2019)	China	Cross-sectional	Multiple linear regression: ln-transformed serum cortisol was negatively associated with the ln-transformed blood lead levels in unadjusted and adjusted models (β=−0.145, 95% CI: −0.272, −0.003, P= 0.045.
(Gump et al., 2008)	New York, United States	Longitudinal	Prenatal and postnatal lead levels, both as continuous and quartile variables, were not associated with baseline cortisol levels in adjusted models. In adjusted models, prenatal lead was significantly associated with cortisol reactivity at 21 min [t (1, 147) = 2.97, p < 0.005], 40 min [t (1, 141) = 2.82, p < 0.01], and 60 min [t (1, 138) = 1.98, p < 0.05]. Similarly, in adjusted models, postnatal lead was significantly or marginally significantly associated with cortisol reactivity at 21 min [t (1, 108) = 2.20, p < 0.05], 40 min [t(1, 103) = 1.84, p < 0.10], and 60 min [t (1, 102) = 2.31, p < 0.05].
(Tamayo y Ortiz et al., 2016)	Mexico City, Mexico	Longitudinal	In the higher lead group (greater than 10 mcg/dl), 2nd trimester lead was associated with decreased change in total diurnal infant cortisol stratified by infant age group (only for 12-month infants and not for 18 – 24 month infants; β = −0.51 (95% CI; −0.85, −0.18) % change −40 (−57, −16).
Adolescent
(Castiello et al., 2020)	Granada, Southern Spain	Cross-sectional	No significant associations of Pb with cortisol or DHEA.
Adult
(Braun et al., 2014)	Mexico City, Mexico	Cross-sectional	Women in the highest quintile of blood lead concentrations had a reduced CAR compared to the 1st quintile group (β −0.06; 95%CI: −0.12, 0.0) compared to women in the lowest quintile. Lead concentrations were not associated with cortisol AUC, though those in the two highest quintiles of patella lead had greater AUC compared to lower quintiles. Tibia/patella lead concentrations were not associated with diurnal cortisol slopes, but diurnal slopes were suggestively flatter among women in the highest patella and tibia lead quantiles compared to women in the lowest quantiles.
(Butts et al., 2014)	California, United States	Cross-sectional	No significant associations between lead and cortisol.
(Fortin et al., 2012)	New Jersey, United States	Cross-sectional	Linear regression model: Both blood (β= −18.5 95%CI −30.8, −4.0) and bone lead (β= −10.5 95%CI −19.3, −0.8) were significantly, negatively associated with baseline cortisol. When adjusting for baseline cortisol, only one significant association was reported between tibia lead and increased cortisol response at 15 minutes after baseline (β= 12.2 95%CI 2.9, 22.4). There were no other significant associations between blood or bone lead and cortisol responses, though tibia lead seemed to be positively associated with cortisol at other time points while failing to reach significance.
(Gustafson et al., 1989)	Sweden	Case-control	Cortisol in the ‘exposed’ group was significantly greater than the matched control group (p=0.04).
(Kamińska et al., 2020)	West Pomerania Province, Poland	Cross-sectional	Significant correlation between blood lead and blood cortisol in women with MetS (R=0.36 p=0.02).
(Klimenko et al., 2016)	Russia	Cross-sectional	Lead was not significantly correlated with cortisol (R=0.304)
(Ngueta et al., 2018)	Montreal, Canada	Cross-sectional	No significant associations between blood lead and baseline or reactive cortisol when treating lead as a continuous and dichotomized variable.
(Owsianowska et al., 2020)	Poland	Cross-sectional	Cortisol levels did not significantly differ between the 3 groups (p= 0.11).
(Pérez-Cadahía et al., 2008)	Spain	Cross-sectional	No significant association between Pb and cortisol in unadjusted models.
(Choi et al., 2022)	South Korea	Cross-sectional	Marginal associations between urinary lead and free cortisol in unadjusted (p= 0.093) and adjusted models (p= 0.058)
Zhang et al., 2023	China	Cross-sectional	In the adjusted model, blood Pb was significantly associated with decreasing serum cortisol (p trend all <0.01). Compared to the lowest quartile, those in the fourth quartile of lead had significantly lower cortisol (95% CI −24.06, −6.87). After adjusting for age and sex, blood Pb was significantly correlated with cortisol (r= −0.174; p < 0.01).
(Souza-Talarico et al., 2017)*	Brazil	Cross-sectional	Multiple linear regression: Blood lead was positively associated with CAR (β=0.791; p=0.007) and AUC total (β= 0.889; p < 0.001). Blood lead was positively correlated with DHEA (r=0.180; p=0.049).
Allostatic Load Adolescent
(Halabicky et al., 2023)	Mexico City	Longitudinal	In adjusted Poisson regressions, for females, maternal tibia bone lead was significantly associated with an increasing allostatic load in adolescence (95% CI 1.001, 1.03). For males, fourth quartile maternal patella bone lead was significantly associated with an increasing allostatic load (β= 1.33) compared to the first quartile of bone lead (95%CI 1.01, 1.74). Childhood lead exposure was associated with a decreasing allostatic load in adolescence for third quartile 24 month blood lead compared to the first quartile (95%CI 0.65, 0.92).
Adult
(Obeng-Gyasi et al., 2021)	United States	Cross-sectional	Linear regression: in models with dichotomized blood lead levels (50%), greater blood lead levels were significantly positively associated with allostatic load compared to low blood lead levels (β= 0.255 p=0.0001). In logistic regression models, dichotomized blood lead levels were significantly, positively associated with odds of having an AL above 3 (β= 1.52; p=0.0001) and an AL above 4 (β= 1.73; p=0.0001).
(Souza-Talarico et al., 2017)	Brazil	Cross-sectional	Blood lead was positively associated with AL index (β=0.204; p=0.032).

^*Also included an allostatic load index

Table 3.

Quality grading of included studies

Study	1. Clear research question	2. Clear population	3. Participation rate 50%	4. Similar subject recruitment	5. Sample size justification	6. Exposure before outcome	7. Sufficient timeline	8. Varying exposure levels	9. Exposure measures clearly defined	10. Exposure assessed more than once	11. Outcome measures defined and valid	12. Outcome assessors blinded	13. Loss to follow-up less than 20%	14. Key confounding variables included
Cortisol and DHEA
(Braun et al., 2014)	Y	Y	CD	Y	N	N	N	Y	Y	N	Y	CD	N	Y
(Butts et al., 2014)	Y	Y	Y	Y	Y	N	N	N	Y	N	Y	CD	NA	Y
(Cai et al., 2019)	Y	Y	Y	Y	N	N	N	Y	Y	N	Y	N	NA	Y
(Castiello et al., 2020)	Y	Y	Y	Y	N	N	N	N	Y	N	Y	CD	Y	N
(Fortin et al., 2012)	Y	Y	Y	Y	N	N	N	Y	Y	N	Y	CD	NA	Y
(Gump et al., 2008)	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	CD	Y	Y
(Gustafson et al., 1989)	N	Y	Y	N	N	N	N	N	Y	N	Y	CD	NA	N
(Kamińska et al., 2020)	Y	Y	CD	Y	N	N	N	N	Y	N	Y	CD	NA	N
(Klimenko et al., 2016)	Y	Y	CD	Y	N	N	N	N	Y	N	Y	CD	NA	N
(Ngueta et al., 2018)	Y	Y	CD	Y	N	N	N	Y	Y	N	Y	CD	N	Y
(Owsianowska et al., 2020)	Y	Y	CD	Y	N	N	N	Y	Y	N	Y	CD	NA	N
(Pérez-Cadahía et al., 2008)	Y	Y	CD	Y	N	N	N	N	Y	N	Y	Y	NA	N
(Tamayo y Ortiz et al., 2016)	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	CD	Y	Y
(Zhang et al., 2023)	Y	Y	Y	Y	N	N	N	Y	Y	N	Y	CD	Y	Y
(Choi et al., 2022)	Y	Y	Y	Y	N	N	N	N	Y	N	Y	CD	Y	Y
Allostatic Load
(Halabicky et al., 2023)	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	NR	N	Y
(Obeng-Gyasi et al., 2021)	Y	Y	NA	Y	N	N	N	Y	Y	N	N	NR	NA	Y
(Souza-Talarico et al., 2017)	Y	Y	Y	Y	N	N	N	Y	Y	N	Y	N	NA	Y

Quality assessment

Study quality was evaluated using the NIH Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. The tool consists of a 14 item checklist to assess the quality of items such as sample size and attrition, temporal order of exposure and outcome and length of time to outcome assessment, measurement of exposure and outcomes, blinding, and confounding variable adjustment. All 14 questions were answered as “yes,” “no,” “not applicable” or “not reported/could not be determined.” The final studies (N=18) that met the criteria for inclusion were independently assessed using the NIH tool by two authors, with any disagreements being resolved through discussions until a consensus was reached (Table 3). Some common weakness identified across the studies were the lack of a sample size justification (N=16), lack of temporal order in which the exposure was assessed before the outcome and lack of sufficient timeframe to reasonably see an effect or the exposure on the outcome (i.e. cross-sectional; N=15), and lead exposure being assessed only one timepoint (N=15).

Cortisol and DHEA

Children

Three studies, one cross-sectional and two longitudinal, examined associations between lead and cortisol in children. The cross-sectional study reported significant negative associations between BLLs (Guiyu Group= Median 4.88 mcg/dl; Haojiang Group= Median 3.4 mcg/dl) and serum fasting cortisol for children ages 3–6 years old in adjusted models (β= −0.145; 95% CI: −0.272, −0.003; p= 0.045) (Cai et al., 2019). The two longitudinal studies examined salivary cortisol post-stressor and in diurnal patterns. Prenatal (F[1,144] = 0.24, p >0.25) and postnatal lead levels (F[1,103] = 1.70, p>0.15) were not significantly associated with baseline cortisol in adjusted models (Gump et al., 2008). However, prenatal lead (range ≤1–6.3 mcg/dl) was significantly associated with an increasing cortisol response at 21 (t [1, 147] = 2.97; p < 0.005), 40 (t [1, 141] = 2.82; p < 0.01), and 60 minutes (t [1, 138] = 1.98; p < 0.05) post stressor in adjusted models. Similarly, postnatal lead (range ≤1–13.1 mcg/dl) was significantly associated with increasing cortisol reactivity at 21 (t [1, 108] = 2.20; p < 0.05) and 60 minutes (t [1, 102] = 2.31; p < 0.05). Diurnal patterns were also shown to be influenced by lead exposure. Prenatal exposure (i.e., blood lead at the second trimester; mean= 3.5 mcg/dl) was significantly associated with a 40% decreased change in total diurnal infant cortisol (95 % CI −57, −16) for 12-month-old infants in adjusted models, though not for 18–24-month-old infants (95 % CI −0.85, −92) (Tamayo y Ortiz et al., 2016).

Adolescents

Only one cross-sectional study examined the association between lead and cortisol in males. The study measured the association between urinary lead concentrations (geometric mean= 0.24 mcg/g) and serum hormones in adolescents, aged 15–17, finding no significant associations between urinary lead and cortisol (percentage decrease −1; 95% CI −6, 4) or DHEA (percentage decrease −3; 95% CI −7, 1) (Castiello et al., 2020).

Adults

Thirteen studies, including one study that also examined allostatic load outcomes discussed below (Souza-Talarico et al., 2017), examined associations between lead and cortisol or DHEA. Of the studies examining diurnal salivary cortisol patterns, all were cross-sectional and reported mixed results. One study investigated the association between BLLs in pregnant women during the 2nd trimester (mean= 3.7 mcg/dl; SD =2.7) as well as tibia (mean= 2.7 mcg/dl; SD =8.4) and patella lead (mean= 4.6 mcg/g; SD =8.6) at one month postpartum, a measure of chronic exposure, and reported no significant associations with maternal diurnal salivary cortisol indices collected between weeks 14–35 gestation (Braun et al., 2014). However, women in the highest quintile group of blood lead had reduced CAR (β −0.06; 95%CI: −0.12, 0.0) compared to the 1^st quintile group. A cross-sectional study of postmenopausal women found no significant associations between mean BLLs (mean 2.41 mcg/dl; SD=0.15) and initial salivary cortisol after adjustment for confounders (adjusted difference of 0.01 μg/dl, p = 0.65) (Ngueta et al., 2018). Ngueta et al. additionally examined reactive cortisol following the Trier Social Stress Test (TSST) and found no statistical difference in average baseline salivary cortisol secretion before TSST exposure (p=0.16), when comparing the “Low blood lead” group (< 2 mcg/dl; mean: 0.13 SD: 0.11) with “High blood lead” group (≥ 2 mcg/dl; mean 0.10, SD: 0.05). Another study reported significant positive associations between older adult (mean 65.9 years) BLLs (mean 2.1 mcg/dl) and both CAR (β=0.791; p=0.007) and total AUC (β=0.889; p=<0.001) (Souza-Talarico et al., 2017).

One final cross-sectional study examined reactive serum cortisol following the TSST in only male participants’ including both blood (median= 4 mcg/dl) and tibia bone lead (median= 13 mcg/g) measurements (Fortin et al., 2012). Both BLLs (β= −18.5; 95%CI −30.8, −4.0) and bone lead levels (β= −10.5; 95%CI −19.3, −0.8) were significantly, negatively associated with baseline cortisol. After adjustment for baseline serum cortisol and other covariates, there were no significant associations of BLLs or tibia bone lead level with the serum cortisol response to the stressor (TSST) except in one condition: 15 min after baseline, tibia bone lead levels were associated with increased serum cortisol levels (β= 12.2; 95%CI 2.9, 22.4). Tibia bone lead appeared positively associated with serum cortisol at other time points after completion of the TSST, though did not reach statistical significance (Fortin et al., 2012).

The remaining 7 studies all examined the association of lead exposure with individual measurements of cortisol or DHEA in adults, reporting mixed results. One cross-sectional study assessed associations between urine cortisol and in vitro fertilization (IVF) outcomes with a secondary objective to examine associations between lead and cortisol. When examining BLLs (median 0.70 mcg/dl), there were no significant associations between lead and urinary cortisol (Butts et al., 2014). This finding was replicated in a study of maternal urinary lead (4.37 mcg/L) and urinary free cortisol, where marginal associations between the biomarkers (p= 0.058) were observed in adjusted models (Choi et al., 2022). Another study examined the serum cortisol levels of 25 moderately exposed lead workers (mean= 39.37 mcg/dl) and 25 matched controls not occupationally exposed (mean 4.14 mcg/dl) and found significant differences in cortisol (p = 0.4) between control and exposed groups. However, no covariates were included in this study (Gustafson et al., 1989). In a study of middle-aged adults (mean age 56.6 years), BLLs (13.37 ng/mL) were significantly associated with decreasing serum cortisol in adjusted models (p trend all <0.01), with significant trends as BLLs increased (Zhang et al., 2023). BLLs were also significantly correlated with serum cortisol, after adjusting for age and sex (r = −0.174; p<0.01). Another study considering blood morphology and severity of depression in peri-menopausal women examined associations between BLLs (mean 7.25 mcg/dl; SD= 3.70 mcg/dl) and serum cortisol as a secondary analysis, reporting significant correlations between BLLs and serum cortisol in women with metabolic syndrome (r=0.36; p=0.02) (Kamińska et al., 2020). One study examined the severity of depression and peripheral blood cell count as well as vasomotor symptoms in relation to the concentration of lead in whole blood of women in the perimenopausal period. The study separated participants into 3 groups based on lead levels (group 1: <5mcg/dl, group 2: 5–10 mcg/dl, and group 3: >10 mcg/dl), finding that serum cortisol did not significantly differ between the 3 groups (p= 0.11) (Owsianowska et al., 2020). An additional study that examined serum trace elements and serum cortisol in transient ischemic attack patients found that plasma lead (~0.05 mcg/dl) was not significantly correlated with cortisol (r=0.304) (Klimenko et al., 2016). A study of individuals that participated in cleaning of coastal areas affected by the ‘Prestige’ oil spill found no significant association between BLLs (mean= 59.13μgl^−1) and cortisol in unadjusted models (β= 0.14; p>0.10) (Pérez-Cadahía et al., 2008). Finally, in older adults (mean age 65.9 years), BLLs (mean= 2.1 mcg/dl) were found to be significantly, positively correlated with DHEA sulfate (R=0.180; p=0.049) (Souza-Talarico et al., 2017).

Allostatic load

Adolescents

One longitudinal study of adolescents reported significant associations between prenatal lead exposure and longitudinal allostatic load. In adjusted models, maternal tibia bone lead (9.07 mcg/g) was significantly associated with increasing allostatic load in females (95%CI 1.001, 1.03) while the fourth quartile of maternal patella lead (11.41 mcg/g) was associated with increasing allostatic load in males compared to the first quartile (Halabicky et al., 2023). Childhood BLLs in this study, however, were associated with decreasing allostatic load in males where the third quartile of BLLs showed lower allostatic load compared to the first quartile of BLLs for males (95%CI 0.65, 0.92).

Adults

Two cross-sectional studies assessed associations between BLLs and allostatic load in adults. An assessment of NHANES data of adults ages 20–85 years found significant positive associations between ‘high blood lead’ (dichotomized at 50^th percentile/Median 1.55 mcg/dl) and an increasing allostatic load index score in adjusted models (β= 0.255, p=0.001) (Obeng-Gyasi et al., 2021). A study of 126 adults (mean age 65.9 years) similarly reported significant associations between blood lead (mean 2.1 mcg/dl) and an increasing allostatic load index (β= 0.204, p= 0.032) (Souza-Talarico et al., 2017). This study also examined associations with cortisol, reported above.

Discussion

To our knowledge, this is the first synthesis of human studies examining associations between lead exposure and biomarkers related to HPA axis function. Eighteen studies were included in the review, with 16 reporting on HPA axis biomarkers and three reporting allostatic load outcomes (one study also included cortisol). Generally, the few studies available in children suggested a significant association between early life lead exposure and altered cortisol. In adulthood, only cross sectional studies were available. These reported significant associations between lead and reduced CAR and increased cortisol reactivity, while fasting serum cortisol showed few significant associations. This review is also an initial report of studies examining associations between lead and allostatic load. While only three studies examined allostatic load directly, they suggested associations between early life and adult lead exposure and an increasing allostatic load, or cumulative body burden (Halabicky et al., 2023; Obeng-Gyasi et al., 2021; Souza-Talarico et al., 2017).

Glucocorticoids and the HPA axis

Cortisol is particularly complicated to measure given its natural diurnal rhythm and relatively slow response to stress. Cortisol can be characterized in terms of in response to a stressor, diurnal pattern, or levels circulating at a single timepoint (Khoury et al., 2015). Levels of cortisol will steadily increase in response to an acute stressor, eventually returning to homeostasis via the HPA axis negative feedback loop (Juster et al., 2016). In the presence of chronic stress, however, cortisol may be either blunted or highly reactive in response to a stressor, both of which are maladaptive (Joos et al., 2019; McEwen, 2006). Considering developmental timing, early life psychological stress, compared to cumulative life stress, seems to be most associated with altered cortisol responses indicative of a dysregulated HPA axis (Young et al., 2021). This association is consistent with the biological embedding model, which posits that early life experiences and biological dysregulation are the basis for adult disease (Shonkoff et al., 2009).

The results of this review seem to align with the biological embedding model, in that early life exposures appeared to have consistent significant associations with serum, diurnal, and reactive cortisol, though reported in few studies. The three studies in children noted significant cross-sectional and longitudinal associations between lead exposure and decreased cortisol levels collected after fasting (Cai et al., 2019), increased cortisol reactivity in response to an external stressor (Gump et al., 2008), and altered diurnal cortisol patterns (Tamayo y Ortiz et al., 2016). These results align with many other studies of very early life lead exposure and longitudinal adult health and neurocognitive outcomes, where longer-term effects are well-known (Cecil et al., 2008; Reuben, 2018; Searle et al., 2014). In addition, these studies report relatively low levels of lead, with mean BLLs ranging from 1.1–5.6 mcg/dl. There are too few studies to assess a linear trend in this relationship, however, these results point to the negative influence of early life lead on the developing HPA axis even at lower levels of exposure.

The one study of adolescents found no significant associations between a single measure of urinary lead and serum cortisol and DHEA collected in the evening, results that may be due to the limited nature of the urinary lead measure which is not well suited to estimate environmental exposure (Castiello et al., 2020; Moreira Mde & Neves, 2008). In this cross-sectional sample, early life exposure to lead was unknown, which could be a more influential time for lead to impact the HPA axis. Non-significant findings could suggest that exposure in adolescence did not influence functionality of the HPA axis. However, results could also suggest that impacts will become apparent in later adulthood. Finally, this study was limited by including only male adolescents. Additional research in male and female adolescent populations examining physiological stress biomarkers, with longitudinal study designs and multiple points of exposure assessment from early life to adolescence, would help to elucidate associations between lead and HPA axis dysfunction across the life course.

In adults, few studies were available that related lead to diurnal or reactive cortisol. One study reported a significant association between lead and cortisol response to an acute stressor (Fortin et al., 2012), while another reported no significant associations (Ngueta et al., 2018). Of two studies that reported on diurnal patterns, one found a significant association between BLLs and diurnal cortisol (Souza-Talarico et al., 2017), and the other reported no significant associations (Ngueta et al., 2018). Lack of statistical significance could be due to a non-generalizable sample of post-menopausal women and a relatively low BLLs (2.41 mcg/dl). Further research is needed with longitudinal study designs including early life exposures throughout adulthood to elucidate associations.

A number of additional studies found mixed results on associations between lead and individual measurements of cortisol via urine or serum. Studies that found significant results reported higher levels of lead exposure with means ranging from 7.25 mcg/dl (Kamińska et al., 2020) to 39 mcg/dl (Gustafson et al., 1989), potentially suggesting associations with baseline cortisol are apparent mostly at higher levels of lead exposure. These mixed or null results could also be due to the unique characteristics of the samples. In fact, these studies included participants undergoing in-vitro fertilization (Butts et al., 2014), women with metabolic syndrome (Kamińska et al., 2020), during pregnancy (Choi et al., 2022) or perimenopausal periods (Owsianowska et al., 2020), adults experiencing TIAs (Klimenko et al., 2016), or individuals impacting by the Prestige oil spill (Pérez-Cadahía et al., 2008), where researchers examined associations between lead and cortisol as often a secondary goal. These samples are likely not representative of the general population and suggest the need for further study. As well, a single measure of cortisol, via serum or urine, is not a highly valid measure of HPA axis functioning (Golden et al., 2011) and is influenced greatly by acute stressors, time of day, fasting and diet, physical activity, and a number of other confounders. While studies of individual measures of cortisol add to the literature base, future research should consider other collection protocols, such as diurnal rhythm and reactivity, to assess HPA axis function and provide a more robust picture of glucocorticoid activity.

Reported associations between lead and DHEA were limited and mixed. In adolescent males, there were no significant associations between urinary lead and serum evening DHEA (Castiello et al., 2020). In adults, however, a cross-sectional study found a significant positive correlation between BLLs and DHEA sulfate (Souza-Talarico et al., 2017). Longitudinal study of endocrine disrupting chemicals (EDCs) has shown associations between prenatal EDC exposure and adolescent DHEA sulfate, suggesting early life exposure may detrimentally influence DHEA sulfate long-term (Watkins et al., 2014). As there were so few studies examining associations between lead exposure and DHEA, especially those with exposure in early developmental years, conclusions on associations could not be made. Further study is warranted to understand how lead may influence DHEA secretion and the role in cortisol regulation.

Allostatic load

Findings from this review tentatively reflect the cross-sectional association between lead exposure and increased allostatic load in adulthood as well as early life lead exposure and adolescent allostatic load. These results were seen even at very low levels of lead exposure (~2 mcg/dl) and while controlling for multiple relevant covariates. A majority of previous research study has focused on the impact of psychosocial stressors and behavioral factors such as childhood maltreatment (Finlay et al., 2022), poverty (Brisson et al., 2020), neighborhood deprivation (Ribeiro et al., 2018), race (Geronimus et al., 2020), and health risk behaviors (Suvarna et al., 2020) on allostatic load. In the environmental health literature, few studies have examined environmental toxicants and their association with allostatic load. EDCs have been reported to influence endocrine gland and HPA axis functionality (Lauretta et al., 2019) and air pollution studies seem to be drawing connections with dysregulated HPA axes (Miller et al., 2020; Plunk & Richards, 2020). Continued research on associations between environmental toxicants on allostatic load is needed to better understand how environmental stressors influence this biological pathway. As a further benefit, allostatic load collection protocols would include individual measures of primary effector hormones, like cortisol, which could also be examined as early indicators of allostatic load development.

In the included studies, there were inconsistencies in the biomarkers included in allostatic load measures, highlighting the difficulties in the synthesis of allostatic load index studies. There is general consensus that allostatic load should include biomarkers from four main categories including cardiovascular, metabolic, immune, and neuroendocrine function. Notably one included study utilized data from the NHANES and was, therefore, unable to include measures of neuroendocrine function due to limitations of the dataset (Obeng-Gyasi et al., 2021). Another study only included six biomarkers with no measures of immune function (Souza-Talarico et al., 2017), whereas the final study included 14 biomarkers inclusive of the four main categories (Halabicky et al., 2023). The studies were comparable in their construction of the allostatic load index, where “high” risk categories were created from the sample distribution and values in the top 25% were binary coded for high risk (i.e., 1) and all others coded as low risk (i.e., 0). Both studies followed previous guidance for biomarkers in which high values are considered appropriate, such as HDL and DHEA sulfate. While the three studies included in this review used similar methodologies for generating their allostatic load index, there still remains a tremendous amount of variability in allostatic load biomarkers, a point for future research.

Timing of lead exposure and physiological stress development

An important consideration of the included studies is their cross-sectional design and lack of data on previous lead exposure. All adult studies included in this review were cross-sectional and, therefore, unable to account for early life lead exposure, which may be influential in these associations. In addition, these studies relied on BLLs to assess lead exposure, which is an acute measurement of exposure and does not represent chronic exposure to lead (Centers for Disease Control and Prevention, 2017). Early life lead exposure, especially in the prenatal and early postnatal periods, can be highly detrimental due to the increased plasticity and susceptibility of children’s brains and rapidly developing organs and body systems (Ismail et al., 2017). In these developmental periods the HPA axis undergoes foundational growth making it especially susceptible to lead exposure and dysregulation. Longitudinal research accounting for effects of early life lead exposure on HPA axis dysregulation and resulting allostatic load in later life is needed to clarify these results. Furthermore, understanding how lead exposure influences the HPA axis, glucocorticoid functionality, and allostatic load during significant physiological life events as individuals age (i.e., postpartum, menopause, etc.) would help to elucidate how lead may alter physiological stress response pathways and the body’s ability to respond to stressors.

HPA axis alteration, evidenced through glucocorticoid alteration, and the eventual progression to allostatic load takes time to progress and become observable. The progression from HPA axis dysfunction to allostatic load may mediate associations between lead exposure and later neurocognitive and health outcomes. In related studies, allostatic load has been reported to mediate associations between childhood trauma and adult depression (Scheuer et al., 2018), neighborhood deprivation and multiple adult health outcomes (Prior et al., 2018), and socioeconomic position and mortality (Kim et al., 2018). Testing HPA axis dysfunction as a mediator within relationships between lead exposure and health outcomes will help to elucidate potential biological pathways for intervention. Examining associations between lead and HPA axis dysfunction in early life, before allostatic load develops, could help with early detection and intervention development to prevent later disease progression. Since allostatic load has been shown to linearly increase as individuals age (Tampubolon & Maharani, 2018), further research investigating relationships between lead exposure, glucocorticoid function, and change in allostatic load in young, mid-life, and older adults would be helpful in understanding the life course implications of lead exposure. Additional research is needed at various stages across the life course to better understand what time periods of lead exposure are most detrimental to HPA axis development and function as well as how long it may take for associations to be observable, if they do exist.

Strengths and limitations

To our knowledge, this is the only systematic review synthesizing the literature on the relationship between lead exposure and HPA axis function across the life course in humans. A large strength of this review is the systematic search strategy, which adhered to PRISMA guidelines, that was designed to capture all existing studies on this topic. The results of this review effectively synthesize what is known about the relationship between lead exposure and the HPA axis via glucocorticoid function in humans while also highlighting gaps in the literature examining human samples. We included primary effector hormones of HPA axis evidencing glucocorticoid function including cortisol and DHEA collected as individual measures, diurnal patterns, and in response to a stressor. However, there are other biomarkers used to quantify stress outside the realm of HPA axis and glucocorticoid function, such as the autonomic nervous system (ANS). Our search would have therefore excluded biomarkers such as epinephrine and norepinephrine, stress hormones within the ANS. This review was also mostly qualitative in nature, given the heterogeneity in measurement of cortisol and DHEA across the studies, difference in age periods, and overall lack of studies assessing allostatic load.

Conclusion

Of the studies available assessing associations between lead and diurnal and reactive cortisol, there appeared to be significant associations in both children and adults, though further research is needed. While not measured in these studies, dysregulated cortisol could suggest an increased allostatic load in these samples. Results seemed more consistent in childhood, which may suggest an important window of susceptibility for lead exposure during times of rapid neurodevelopment. Future research should consider diurnal and reactive cortisol as well as DHEA and the cortisol/DHEA ratio to capture different aspects of HPA axis and glucocorticoid functionality. Few studies examined DHEA or it’s sulfate, limiting conclusions. There are too few studies available to make conclusions about the association between lead and allostatic load. However, the three available studies do cautiously suggest a positive association between lead and increased allostatic load in adults and longitudinally in adolescents. Additional research is needed to standardize construction of allostatic load measures across the life course. HPA axis and glucocorticoid alterations may be a plausible biological mechanism through which lead in part exerts effects on later adverse neurocognitive and health outcomes, a potential area of future research.

Given the global rates of lead exposure as well as that altered HPA axis, glucocorticoid function, and increased allostatic load are associated with chronic health conditions, the association of lead to HPA axis function represents a significant public health concern. Additional research that considers gaps highlighted in this review can be beneficial for public health policy and interventions. First, further research in this area could improve policymakers and clinicians’ understanding of the impact of lead exposure on HPA axis function and the resulting impacts on biological systems. Furthermore, additional research can help establish clearer benchmarks for intervention development targeting potential biological pathways between lead exposure and adverse health outcomes. Understanding these relationships could establish criteria for early detection of HPA axis dysfunction, via glucocorticoid measurements, which could guide clinical practice. Such research will provide support for researchers and clinicians transitioning these research findings into policy to improve the health of those exposed to lead. Overall, continued research is needed on the part of environmental health researchers to consider the biological mechanisms responsible for neurocognitive and health outcomes resulting from lead exposure, with particular attention paid to developmental periods that may be the most susceptible time for lead to influence HPA axis, glucocorticoid functionality, and physiological stress long-term.

Data availability Statement:

The data that support the findings of this study are available from the corresponding author, OMH, upon reasonable request.