Lead Exposure Across the Life Course and Age at Death

Abstract

Background:

Lead (Pb) exposure has been associated with an increased risk of all-cause mortality, even at low levels. Little is known about how the timing of Pb exposure throughout life may influence these relationships. Quantifying the amount of Pb present in various tissues of the body provides measurements of exposure from different periods of life. These include bone, tooth enamel, which is the hard outer layer of the crown, and tooth cementum, which is the calcified connective tissue covering the tooth root. The purpose of the study was to examine Pb exposure at multiple periods throughout life, including childhood (enamel), adulthood (cementum), and later life (bone), and to estimate their associations with age at death.

Methods:

208 skeleton donors (born 1910-1960) from an ongoing case-control study were included in this study. Pb was measured in tibia (shin), bone using X-Ray Florescence and in teeth using Laser-Ablation Inductively Coupled Plasma Mass Spectroscopy. After excluding unusually high measurements (>2sd), this resulted in a final sample of 111 with all exposure measures. Correlations across measures were determined using partial Spearman correlations. Associations between Pb exposure and age at death were estimated using Multivariable Linear Regression.

Results:

Pb measures across exposure periods were all significantly correlated, with the highest correlation between cementum and tibia measures (r = 0.61). Donors were largely female (63.0%), White (97.3%), and attended some college (49.5%). Single exposure models found that higher tooth cementum Pb (−1.27; 95% CI: −2.48, −0.06) and tibia bone Pb (−0.91; 95% CI: −1.67, −0.15) were significantly associated with an earlier age at death. When considered simultaneously, only cementum Pb remained significant (−1.51; 95% CI: −2.92, −0.11). Secondary analyses suggest that the outer cementum Pb may be especially associated with an earlier age at death.

Conclusions:

Results suggest that higher Pb exposure is associated with an earlier age at death, with adulthood as the life period of most relevance. Additional studies using Pb exposure measures from different life stages should be conducted.

Graphics Abstract

1. Introduction

Exposure to lead (Pb) has long been viewed as a serious risk to the health of individuals and our ecosystems.¹ Common human exposure sources have included drinking water pipes, paint, and leaded gasoline.² At the highest concentrations (e.g., 100-120 μg/dL of blood), Pb may result in imminent death, while exposure to lower concentrations has been associated with an increased mortality risk.³ Studies using blood as a biomarker of Pb exposure have consistently found that higher Pb is associated with an increased risk of death in the next two decades following exposure.⁴ The nonlinear relationship found in these studies suggests that even at low concentrations (e.g., 3.6-10 μg/dL) Pb exerts an effect on mortality.⁵ Bone Pb, a measure of cumulative exposure, is also related to an increased risk of mortality in older individuals.^6,7 The analysis of Pb concentrations within teeth has been used to provide a history of early life exposure; however, these measures are much less frequently used in the epidemiology literature.⁸ To date, no study has simultaneously examined the health effects of Pb exposure at different life stages in the same individual. Therefore, the most influential exposure periods remain unknown.

To date little work has been done to document the long-term consequences on adult health of Pb exposure from earlier in life. One study conducting a death certificate search on Pb-poisoned children decades later found an increased risk of mortality in adulthood due to cardiovascular disease compared to rates in the general United States adult population.⁹ However, this work was conducted in a sample with very high Pb exposures that are not representative of the Pb concentrations of the general population. Additional work in both humans and animals suggests that exposures from earlier life may only manifest in impairment at older ages, including those related to cognitive decline and Alzheimer’s-like pathology.^10–12 Given these long term health consequences and the fact that Pb can be highly correlated throughout the life course, studies of later life Pb exposure could be confounded by earlier exposures.^13,14

Studying the potential long-term effects of Pb is made difficult by the limited window captured by commonly used biomarkers. While blood Pb concentrations reflect exposure over the few months immediately prior to blood collection, bone concentrations represent average exposure from years to decades before assessment depending on bone type, making bone measures critical for documenting the effects of chronic exposure.^15,16 A recent study among adults found that the half-life of Pb in tibia (shin bone), made up mostly of cortical bone, is between 7-26 years depending on the age of the participant.¹⁶ Unlike bone, tooth enamel grows incrementally over a defined period and does not remodel. Therefore, Pb accumulation in enamel reflects exposure during enamel formation, which in permanent (adult) teeth occurs from birth to about 4-10 years of age depending on the type of tooth (e.g., incisor, premolar, etc.).¹⁷ After completion of the tooth crown (enamel), mineralization of the tooth root continues into adulthood. The tooth is anchored in the jaw through fibers (periodontal ligaments) that connect the cementum to the surrounding gum. Cementum is a calcified layer that starts to form after the crown is completed and when the root forms. Cementum deposition continues in concentric layers through adulthood, like the growth rings of a tree.¹⁸

The use of laser ablation with traditional inductively coupled plasma mass spectroscopy (LA-ICP-MS) is a spatially resolved sampling strategy for the reconstructing of metal exposure over time rather than bulk analysis of ground samples.^8,19,20 While many studies have been conducted quantifying Pb in human deciduous teeth,^8,21–26 including work in epidemiology,^27–30 relatively fewer studies have been conducted with human adult teeth.^31–35 A significant correlation has been found between the Pb concentration in the enamel of permanent teeth and the enamel of the same person’s deciduous tooth, suggesting this is a reliable measure of childhood exposure.³³ To our knowledge no epidemiologic study has considered cementum Pb concentrations as an exposure measure; however, other disciplines have used LA-ICP-MS to quantify metals in feline and ursine cementum and anthropologists have used cementum to estimate individuals’ ages at death.^36–38 The latter work used microscopy to count the layers of cementum, which deposit nearly annually until death. Estimates aligned with mortality data for those dying at age 35 and under but underestimated the age at death for those dying at older ages suggesting less cementum formation later in adulthood.

How the timing of Pb exposure throughout the life course affects the risk of mortality remains unknown. Several studies have found a relationship between Pb exposure and mortality in the one to two decades following exposure.⁴ Additionally, one study has examined the long-term effects of high Pb exposure in childhood on mortality effects in adulthood.⁹ To our knowledge, no study has examined the relationship between mortality in an adult population using biomarkers reflective of Pb exposures throughout the life course including childhood, adulthood, and later life. The novelty of this study is that it examines Pb concentrations within the bone and teeth of the same individuals to identify sensitive periods for the mortality effects of Pb exposure, information that can help allocate resources to protect potentially vulnerable groups. Additionally, this is one of the first epidemiologic studies to evaluate the usefulness of LA-ICP-MS derived Pb measures from adult teeth in relation to a health outcome and is the first to examine cementum using this technique.³⁹ We hypothesized that Pb exposure at each time period (childhood, adulthood, and later life) would independently be associated with an earlier age at death.

2. Materials and methods

2.1. Population.

Participants’ bodies were donated to the body donation program of the Department of Anthropology at the University of Tennessee, Knoxville (UTK) and their skeletal remains are curated in the UTK Donated Skeletal Collection. Their teeth and tibiae were analyzed for this study, which includes 208 donors with no known history of dementia. Of these, 203 tibiae were available for Pb measurement, and we randomly selected 125 for tooth analyses. These extracted teeth were predominately incisors (n = 82) and premolars (n = 31) and were free of dental restorations, such as fillings. Both enamel and cementum were measured on tooth samples. Cementum measurements were not possible on 2 of the donors. In primary analyses we excluded individuals with any notably high measures (≥2 standard deviations (sd) above or below the mean). This resulted in the exclusion of 7 individuals, with 4 being outliers for two or more measurement sites. We included these high measures in sensitivity analyses. This resulted in a total of 111 participants with non-extreme exposure measures in all three types of tissue (see Supplement S1 and S2).

Additional information about the donors’ medical and social histories was either collected from the donors themselves antemortem or from their next-of-kin post-mortem. To address concerns of confounding we included variables in our models that are thought to be related to Pb exposure and one’s age at death. All models included variables that would precede the earliest exposure measure such as year of birth, sex, self-identified race, and childhood socioeconomic status (SES). Year of birth was also included to account for the general reduction of Pb exposure that has occurred over the course of the 20^th century. Additionally, since donors had to be deceased, the maximum possible age at death is limited by their year of birth. For models primarily examining the effect of later life Pb exposures, we additionally adjusted for donor’s educational attainment as a proxy for adulthood SES, as this would likely influence exposures in later life through shaping occupational and residential opportunities. Education was not adjusted for in models primarily focused on childhood Pb as it likely serves as a mediator of the potential childhood Pb-mortality association. We dichotomized the childhood SES variable into Lower SES vs. Middle/Upper SES and the education variable into High School or less vs. At least some college for efficiency. The childhood SES variable was only added to the registration questionnaire in 2006, therefore there were a substantial number of donors missing these data. Due to the lack of racial diversity in the sample, the self-identified race variable was dichotomized to White and non-White.

2.2. Bone Pb Measurement

Pb was measured in donors’ tibia bones with X-Ray Fluorescence (XRF) using the Thermo Fisher Niton XL3t GOLDD+ portable XRF device (Thermo Fisher, Billerica, MA, USA). The device used custom calibration and settings using a 50 kV and 40 μA x-ray tube and a silver and iron filter. Calibration was conducted using plaster-of-Paris samples ranging from 0 to 100 μg/g dry bone. Corrections were made based on size of the bone by normalizing to the silver anode Compton Scattering peak. Bone measurements were all made on bare tibia bone at the mid-shaft location and measurements were normalized to grams of bone mineral to account for potential bone differences between individuals. The average uncertainty for the measurements in this study was 2.0 μg/g (standard deviation, sd=0.51) bone mineral, markedly lower than previous studies that had measurements through soft tissue.^40,41 The tibia measurements reflect Pb exposure in the years to decades prior to death based on recent evidence suggesting that the half-life of Pb in the tibia for the ages represented in this study is between 13-26 years.¹⁶

2.3. Tooth Pb Measurement

LA-ICP-MS analysis used a New Wave Research 213 laser ablation system with a 10 cm² sample chamber, interfaced with an Agilent 7900 ICP-MS system. Laser power was 75% with a frequency of 20 Hz, with an average fluence of 5 J/cm² Ablation patterns were a series of 50 μm round spots with a dwell time of 7 seconds running along the surface of the tooth. We ablated the exterior surface of the tooth (n = 125) on the vertical midline through the tooth on the buccal surface of incisors, and randomly of molars, from the occlusal edge to the cervical margin using a spacing of 200 μm for the enamel (or erupted portion of the tooth) and 500 μm spacing between spots in the cementum (below the gum line). This method is minimally invasive and non-destructive of the tooth samples. Standard reference materials used for the analysis were pressed pellets of NIST 1486 (bone meal) (2 tons for 5 minutes) and a Pb-doped goat bone material from the Wadsworth Center (New York State Department of Health) NYS RM 05-0.⁴² Scan speeds, spot size and dwell settings for ablation patterns of the reference materials were identical to the samples. Both standards were analyzed before and after each specimen.

LA-ICP-MS data were analyzed and processed using the Iolite 4 software package.⁴³ The first and last second of data from each ablation spot was trimmed to eliminate the effect of wash-in and wash-out on the data and avoid any surface contamination of the tooth. Counts per second Pb data were then background corrected, ratioed to calcium (Ca), and converted to μg/g using NIST 1486 as a single point calibration as a primary standard. While prior studies have used the ratio of Pb to Ca to internally normalize the amount of tooth tissue ablated,⁴⁴ we further quantified data into a standardized μg/g measurement that additionally normalized measurements across time where the LA-ICP-MS relative sensitivity (mass bias) for Pb and Ca can vary. We used a Pb-doped goat bone standard as a secondary check. As described above, different sections of adult teeth develop at different life stages, with Pb concentrations in those sections reflecting exposures during that development period. Thus, we averaged all our spot measurements in enamel and cementum separately to reflect exposures in early childhood and over adulthood, respectively. To further refine the exposure period reflected by cementum in secondary analyses, we separated each cementum spot measure into two—the first and second halves of the laser ablation dwell time. This provides a measure of the outer and inner cementum, respectively, with the inner cementum laid down earlier in adulthood than the outer cementum.

2.4. Statistical Analyses

Partial Spearman correlations across the various exposure metrics were calculated, adjusting for batch in which the teeth and bone measurements occurred. To assess the primary association between each Pb exposure metric and age at death, we first used generalized additive models (GAMs). Penalized splines using Restricted Maximum Likelihood were included for the effect of Pb in all GAMs to allow for potential non-linear relationships between the Pb measure(s) and age at death. All GAMs suggested linearity, so results from linear regression models are reported throughout this paper as adjusted beta coefficients and 95% confidence intervals (CI) for years difference in age at death per sd difference in the relevant Pb exposure metric. In some cases, models for later Pb exposure measures were adjusted for earlier Pb exposure measures as these earlier exposures could introduce confounding of the associations between the later measure and age at death. To account for missing data, all models were run using 100 imputations from multiple imputation by chained equations, with models predicting missing values including variables contained in the fully adjusted models, as well as year of death.⁴⁵ To account for the additional uncertainty of using multiple imputation, variance was pooled using Rubin’s rules.⁴⁶ Importantly none of the Pb exposures independently predict missingness of childhood SES suggesting these methods could yield unbiased estimates of the true effects of Pb on age at death. For all tests, we considered significance at the 95% level of confidence. All analyses were performed in R version 4.1.1.

2.5. Sensitivity Analysis

To evaluate the robustness of results, several sensitivity analyses were performed. First, the fully adjusted models were re-run with the inclusion of the tooth type to account for potential minor differences in timing associated with the Pb exposure or potential analytic differences in measurements.⁴⁴ Second, because of the large amount of missing data for childhood SES and concerns that multiple imputation methods yield wide confidence intervals, we ran the fully adjusted models only among donors with SES data (n = 59) and in the full sample using a missing indicator. Third, each tooth time period average exposure was recalculated after excluding any individual spot measurement that exceeded two standard deviations from the mean of all spots in that region of the tooth (i.e. enamel or cementum) or had an uncertainty exceeding two standard deviations of the mean uncertainty for that region. This was intended to reduce the effects of potentially spurious individual spot measurements. Finally, we also ran the models including the seven donors with very high Pb exposure values.

3. Results

Pb measures across exposure periods were variably correlated with one another. The highest correlation was between the cementum (adulthood) and tibia (late life) measures (partial ρ = 0.61). The correlations between the other pairs of measurements (enamel—cementum; enamel—tibia) were both 0.35. Correspondingly, levels of the cementum (adulthood) and tibia (late life) Pb measures tended to increase with increasing quartile of the enamel (early childhood) exposure (Table 1).

Table 1:

Descriptive Statistics by Quartile of Childhood (Enamel) Pb Exposure

Enamel Pb Quartile	1st Quartile	2nd Quartile	3rd Quartile	4th Quartile	p-value
n	28	28	27	28
Exposure
Tooth Enamel Pb	25.2 (11.3)	56.4 (8.8)	86.3 (12.5)	181.9 (66.6)	<0.001
Tooth Cementum Pb	39.1 (25.8)	42.5 (16.4)	67.5 (80.6)	68.3 (46.1)	0.04
Tibia Bone Pb	14.5 (13.3)	22.34 (15.1)	21.7 (8.2)	27.7 (13.4)	0.003
Demographics
Male (%)	7 (25.0)	12 (42.9)	10 (37.0)	12 (42.9)	0.47
Year of Birth	1936 (14.4)	1932 (8.2)	1930 (9.0)	1926 (8.8)	0.004
Education (%)					0.87
High School or Less	8 (28.6)	9 (32.1)	6 (22.2)	8 (28.6)
At Least Some College	13 (46.4)	15 (53.6)	15 (55.6)	12 (42.9)
Missing	7 (25.0)	4 (14.3)	6 (22.2)	8 (28.6)
Childhood SES (%)					0.46
Lower	5 (17.9)	5 (17.9)	3 (11.1)	4 (14.3)
Middle	11 (39.3)	8 (28.6)	10 (37.0)	5 (17.9)
Upper	0 (0.0)	4 (14.3)	2 (7.4)	2 (7.1)
Missing	12 (42.9)	11 (39.3)	12 (44.4)	17 (60.7)
White (%)	28 (100.0)	26 (92.9)	26 (96.3)	28 (100.0)	0.29
Sample Type
Tooth Type (%)					0.02
Canine	0 (0.0)	0 (0.0)	1 (3.7)	4 (14.3)
Incisor	19 (67.9)	23 (82.1)	21 (77.8)	12 (42.9)
Molar	1 (3.6)	0 (0.0)	0 (0.0)	0 (0.0)
Premolar	8 (28.6)	5 (17.9)	5 (18.5)	12 (42.9)
Batch 2 (%)	5 (17.9)	7 (25.0)	6 (22.2)	8 (28.6)	0.81

Numbers reflect mean and (sd) unless otherwise specified. Tooth enamel Pb corresponds to childhood. Tooth cementum Pb corresponds to adulthood. Tibia bone Pb corresponds to later life.

Donors were born between 1910 and 1960 and died at a median age of 76. The majority of the donors in this sample were female (63.0%) and White (97.3%). Nearly half of the donors attended at least some college (49.5%). Among those with childhood SES data (n = 59), 71.2% reported middle or upper SES. Consistent with decreasing temporal trends of Pb in the environment, higher enamel Pb levels were found among those with earlier years of birth (Table 1). Other variables did not show very consistent patterns, although the percentage of males trended upwards at higher enamel Pb levels. These distributions were similar by quartiles of the other Pb measures (Supplement S3, S4). Additionally, the average Pb concentration was higher in males compared to females for enamel (88.6 vs. 86.7 μg/g), cementum (65.5 vs. 47.6 μg/g), and tibia (26.6 vs. 18.6 μg/g).

In the base models, which only adjusted for year of birth and batch number, the single exposure models all suggested that higher Pb exposure may be associated with an earlier age at death, though only cementum Pb was statistically significant (Table 2). Further adjusting for sex, self-identified race, childhood SES—and for the donor’s education in analyses of cementum and tibia Pb—showed similar results. In fact, these adjustments strengthened the association between tibia Pb and age at death and made it statistically significant. For both the cementum (−1.27; 95% CI: −2.48, −0.06) and tibia (−0.91; 95% CI: −1.67, −0.15), a one standard deviation increase in exposure to Pb was associated with an approximately one year earlier age at death on average. In adjusted analyses restricted to the 111 donors with all three Pb measures, the adjusted associations with enamel and cementum Pb were stronger than in the larger sample, while the association with tibia Pb was weaker, and when additionally adjusted for the two earlier life Pb measures, totally null (−0.01; 95% CI: −0.99, 0.96). The association with cementum Pb remained significant even when further adjusted for enamel Pb (−1.51; 95% CI: −2.92, −0.11) (Figure 2 and Table 2).

Table 2:

Difference (95% confidence interval) in age at death (years) per standard deviation higher Pb exposure.

	Childhood	95% CI		Adulthood	95% CI		Late Life	95% CI
	n= 121			n= 123			n= 199
Base- Single	−0.45	(−1.55, 0.64)		−1.83	(−2.88, −0.77)		−0.58	(−1.34, 0.17)
Adjusted- Single	−0.41	(−1.24, 0.42)		−1.27	(−2.48, −0.06)		−0.91	(−1.67, −0.15)
	n = 111			n = 111			n = 111
Base- Final	−0.37	(−1.56, 0.83)		−1.90	(−3.05, −0.75)		−0.61	(−1.86, 0.64)
Adjusted- Final	−0.48	(−1.35,0.38)		−1.61	(−2.95, −0.27)		−0.50	(−1.35, 0.35)
Multi Exposure	NA	NA		−1.51	(−2.92, −0.11)		−0.01	(−0.99, 0.96)

Base models are adjusted for year of birth and batch. Adjusted models are additionally adjusted for sex, self-identified race, and childhood SES, as well as educational attainment in the cementum and tibia models. The Multi Exposure model additionally adjusts for prior Pb exposure(s). Enamel Pb corresponds to childhood. Cementum Pb corresponds to adulthood. Tibia Pb corresponds to later life.

Fig. 2.

Adjusted age at death by adulthood (cementum) Pb exposure – with and without maximum point.

Secondary analyses that split the cementum measure into the inner and outer portions found that the effect was stronger for Pb exposure in the outer portion and weaker for the inner portion, though both measures were inversely associated with age at death. In models examining each half of the cementum individually, a 1 sd (49.7 ug/g) higher inner cementum Pb was associated with a 1.09 (95% CI: −2.22, 0.04) years earlier age at death adjusting for year of birth, sex, childhood SES, batch, and enamel Pb, while the same relative increase in outer cementum Pb was associated with 1.26 (95% CI: −2.41, −0.10) years earlier age at death adjusting for the same factors and education. Similar results were found in fully adjusted models (with and without adjustment for education) that simultaneously considered both inner and outer cementum, though the estimates were attenuated, with inner cementum Pb associated with a 0.62 (95% CI: −1.94, 0.70) year earlier age at death and outer cementum with a 0.92 (95% CI: −2.28, 0.44) year earlier age at death.

3.1. Sensitivity Analysis

Adjustment for tooth type did not substantively change the reported associations of any of the single exposure models, though it did weaken all associations in the models adjusting for multiple Pb exposures. Among those with all Pb measures, in adjusted analyses—including for any earlier life Pb measures—the findings were largely robust across sensitivity analyses (Figure 3), including the fact that cementum Pb is the strongest association in each model (S5, S6). Results were similar in magnitude in analyses that accounted for the missingness in the childhood SES data either by missing indicator or restriction to those without missing childhood SES data, with the missing indicator method yielding estimates that were somewhat stronger and with narrower confidence intervals. The complete case models yielded weaker estimates.

Fig. 3.

Effect estimates of adulthood Pb controlling for childhood Pb by senstivity analysis.

4. Discussion

Overall, we found associations between higher Pb exposures and an earlier age at death, a finding that aligns with several previous studies.⁴ The most consistent evidence was for the association with cementum Pb, for which a one standard deviation higher Pb concentration was associated with about 1.5 years earlier age at death (95% CI: −2.92, −0.11). In all analyses, enamel Pb was not significantly associated with age at death, although it trended that way. Tibia Pb showed a significant association with age at death in the single exposure model. However, this association was entirely attenuated after adjusting for enamel and cementum Pb.

Each of the biomarkers in this study corresponds to different exposure periods reflective of when the tissue mineralized. For enamel, this corresponds to childhood, specifically the period between birth and up to 5-9 years depending on tooth type.¹⁷ This measure reflects the shortest window across the biomarkers in this study. Cementum on the other hand represents the longest period from around adolescence through death.³⁸ The inner cementum corresponds roughly to adolescence and earlier adulthood and the outer cementum represents later adulthood through death. Therefore, we conceptualized the full cementum as measuring adulthood Pb exposure. Finally, the tibia measure captures Pb exposure occurring in the final decades of life, which we refer to as later life exposure. While the cementum and tibia measures have some overlap, the tibia measure is more representative of the last decade of life exposure because bone tissue undergoes turnover throughout life, whereby osteoclasts break down the bone, re-releasing minerals into the blood stream, and osteoblasts replace it with new tissue.¹⁶ Therefore, Pb in bone is weighted towards reflecting exposures closer to the time of measurement, and no evidence exists that exposures from childhood would be present in this sample. Of course, if an individual’s earlier external Pb exposures were higher earlier in adulthood than later, it could still be that the earlier exposures represent the largest contribution to the individual’s bone Pb. Cementum on the other hand does not turn over, so the measure is more strictly an accumulated signal of the total Pb exposure accrued throughout the time of cementum formation during a person’s adulthood. When these biomarkers are included in the same model, the cementum measure more faithfully reflects the adulthood period alone as the Pb exposure from later life is controlled for with the tibia measure. As a result, we have elected to refer to cementum Pb as adulthood Pb throughout this discussion.

These results broadly agree with previous findings; however, no known study has considered Pb exposure in multiple life periods simultaneously and their association with death, making direct comparisons difficult. Prior studies that examined exposures during single time periods considered mortality rate.^6,9 An increase in mortality rate translates into an earlier age at death and is the outcome used in these analyses. Only one study has focused on the effects of early life Pb exposure on mortality in adulthood, finding that high acute Pb exposures in childhood were associated with earlier age at death after these incidents, with a particularly elevated occurrence of deaths from cardiovascular disease, blood disorders, and subsequent events of Pb poisoning, a result that differs from our findings.⁹ However, this discrepancy could arise from the difference in Pb exposures, namely, early life acute versus chronic exposures throughout the lifespan, being assessed across the two studies. Further, those exposed to Pb in early life may also be more likely to have additional exposures later in life that were not captured in the earlier study. Several studies that analyzed blood and/or bone Pb levels in adulthood have suggested that higher Pb exposure is related to increased risk of cardiovascular and all-cause mortality.^5,6,47–50 While the biospecimen used in this study for analyses of Pb exposure during adulthood —tooth cementum—was different from that used in those other studies, tooth cementum Pb concentrations also reflect exposures throughout adulthood. Our results using this measure suggest a similar adverse association between adult Pb exposure and mortality. We also measured bone Pb, as has been done in the earlier studies, but the different timing of the measurements means that our tibia measures may reflect different exposures than those in the previous studies, which may explain the difference in results.^6,7 In our current study, bone Pb concentrations were measured post-mortem; therefore, exposure is estimated to correspond to decades immediately before death, which in our sample occurred at age 76 on average. Alternatively, a prior study measured bone Pb in living adults, at a mean age of 67.3, meaning those measures reflected exposure in the decades before that.⁶ Thus, measurements that capture the individuals middle to later adulthood are more comparable to the period being captured by the cementum measures in our current study.

We did not find significant evidence for an association between early childhood Pb exposure and age at death, although the direction of the association was towards an earlier age at death with higher Pb exposure. It is possible that there is no association at these levels of childhood Pb exposure, although as mentioned above, high level exposure in childhood has been associated with mortality later in life.⁹ The lack of significant association found in this study could stem from the fact that low levels of Pb are detrimental only with chronic, perhaps decades long, exposure. Since our enamel measure only corresponds to a handful of years, it is possible that our biomarker captures too short of a window to find a significant effect on mortality. This aligns with prior research finding a stronger association with mortality for bone Pb, a measure of averaged exposure over a much longer period of time, than blood Pb, which is a measure of short term exposures.⁶ Here, the enamel measure reflects a few years during childhood, while our cementum exposure represents an extended period of time through adulthood. Still, it is noteworthy that Pb over even just a few years in childhood trended towards a negative association with age at death (−0.48, 95% CI: −1.34, 0.38). This suggests the exposure period could in fact be a somewhat more sensitive one for later age at death but would require a larger study to identify this association more robustly.

The additional possibility that selection bias could have contributed to the findings should be considered. Our study sample consisted of eligible controls for a dementia study and thus participants were selected for being older to match cases of dementia, primarily a disease affecting older adults. If childhood Pb exposure is associated with age at death at even earlier ages than those in our current study, then this would have created a downward bias on our study’s estimate of the association with childhood Pb and could have contributed to the lack of significance for our findings. Recent simulation studies examining this type of differential survival bias suggests that the effects of this bias are most concerning in the presence of an unmeasured effect modifier (or multiple ones) such as childhood nutrition or home environment enrichment.^51,52 Unfortunately, a lack of data on such modifiers limited our ability to explore this possibility.

Our findings for later life Pb concentrations were somewhat mixed. In the full sample, we saw an association between high Pb concentrations and an earlier age at death. However, this estimate was much weaker in the analyses restricted to those with all three Pb measures, and completely disappeared when we adjusted for Pb exposure at earlier life periods. This suggests that more acute Pb exposure in the years immediately prior to death may be less relevant for mortality than chronic Pb exposure throughout adulthood, the correlation with which may have accounted for the findings when late life Pb is considered alone. Our results suggest that using a bone measure alone still captures part of the effect of adulthood Pb exposure and should therefore still be used in the absence of available cementum data. The tibia measure reflects the latter time periods of exposure and therefore is a less accurate measure of chronic exposure whereas measures derived from cementum are complementary and better capture lifetime exposures. It is not unreasonable that Pb exposure throughout adulthood is more relevant than that immediately before death, given the chronic nature of many of the health effects of Pb,⁴ and the fact that Pb in cementum is deposited with cementum layers over a very long time period. Despite there being some overlap in the timing represented in our study in measures derived from the tibia and those from cementum, cementum remained significantly associated with an earlier age at death even after including tibia Pb in the same model. This suggests that cementum better captures the more relevant measure for the mortality effects of Pb exposure, perhaps because cementum deposition is less influenced by behavioral and biological changes in late life compared to bone turnover and captures a longer exposure period. It is unlikely these results are due to differences in how Pb concentrations were measured in cementum (i.e., LA-ICP-MS) and tibia (i.e., portable XRF) as a recent analysis has found a high correlation between bone measures using both methods (r > 0.9).⁵³ To date, little epidemiologic work has used LA-ICP-MS to assess bone Pb.⁴

Beyond the relative amount of time represented by the cementum measurement, findings related to exposures during adulthood align with previous results suggesting that this period is influential for cardiovascular health and mortality.⁶ This may stem from the fact that individuals are less capable of withstanding Pb exposure as they age, particularly because Pb is detrimental to many of the body’s repair pathways.^3,50 The mortality effects of adulthood Pb likely arise through its disruptions to the nervous system, cardiovascular system, and the renal system.⁴ Changes to organ function contribute to several different health outcomes including renal dysfunction, anemia, and cognitive decline in adults.^3,10 Cardiovascular disease, the leading cause of death for adults in the United States, has also been linked to Pb exposure, with several studies suggesting that higher Pb exposure is associated with increased blood pressure, altered cardiac function, and increased cardiovascular mortality.⁴ These results suggest that continued monitoring of Pb exposure in adults, like that already done in children, should be considered. Additionally, more research should be conducted to develop strategies to broadly reduce Pb in the general population and understand how nutrition or other lifestyle factors may mitigate the health effects of adults already exposed to Pb.

Intriguingly, the comparison between data from the outer layers of cementum (later adulthood) and deeper layers (earlier adulthood) of cementum, suggested that later adulthood Pb exposure was more relevant for mortality. Since the time periods averaged in this analysis by each half of the cementum would be roughly the same, this suggests that later adulthood is a more relevant exposure window than earlier adulthood. These findings align with recent work identifying the later adulthood period as one of particularly steep declines in cardiovascular health.⁵⁴ These declines, and their potential downstream effects on mortality are likely due in part to Pb exposure during this period.⁴ However, the exact timing covered by these halves cannot be verified without histology and would likely vary by tooth type and exact location on the tooth. Therefore, it is possible that the outer half captures a larger period of time, and its relative importance in our models is due to the total amount of years of exposure captured rather than the specific years captured. While of theoretical interest in terms of honing in on potential sensitive windows for Pb exposure, these secondary results should be interpreted with caution as they cannot be verified with histology. Additional work should be done to better understand the precise timing of these measurements and their relation to mortality.

Despite the strengths of this study, including the use of Pb biomarkers at different life stages, there are several limitations. First, childhood SES may be an important confounder,^14,55,56 and there was a high percentage of donors for whom we did not have this data. Our choice to handle missing data through multiple imputation relies on the assumption that no unmeasured variables predict missingness. While this assumption is ultimately unverifiable, we did not find any variable that significantly predicted missingness after including year of death in the model. This suggests that missingness is, at least largely, a result of whether or not the donor died prior to UTK systematically collecting these data rather than demographic or behavioral factors. Therefore, the missingness was well accounted for using multiple imputation methods. We also found generally similar findings across several different methods for addressing missing values, which suggests that the residual confounding of this covariate likely cannot fully explain away our results. Additionally, donor education, an indicator of adulthood SES, was included in the models of adulthood, the measure that most consistently found significant results. This limits the possible effect of unmeasured confounding to pathways not mediated through educational attainment, which may be minimal.⁵⁷ However, it is possible that education does not fully capture adulthood SES, particularly for females (the majority of our sample), due to a lack of educational opportunities for individuals at this time. This would result in some individuals being misclassified as low SES. Including additional covariates that were unavailable for this cohort such as family income could mitigate this potential bias. Second, there were some individuals with extreme Pb levels in the different Pb measures and we excluded these individuals in the main analyses. This is mostly because even if those are correct exposure estimates, the few people at those levels would likely be subject to strong selection bias (i.e., be unrepresentative of people exposed to those levels). Nonetheless, our findings were similar when including those individuals. Third, we were also limited by the lack of racial diversity, with only 3 donors not identifying as White, reducing the generalizability of our results. In fact, the main analyses utilizing multiple imputation excluded the non-White donors due to missing childhood SES data. Alternative methods found no difference in estimates when including or excluding the non-White donors; however, additional work in more racially diverse samples should be conducted to examine these effects in other racial groups who, at least in the United States, share an undue burden of Pb exposure.⁵⁸ Fourth, we used a study sample selected to be older, so generalizing effects of Pb on death at earlier ages is compromised. Fifth, there is some overlap in the exposure windows reflected by cementum and bone Pb. This could have somewhat impacted our ability to detect significant findings with our bone measurement. However, the cementum measurement consistently remaining significant even when including the bone measure suggests that, at least for this sample, the cementum measure is a better predictor of mortality. Finally, our sample is rather small, which limited our power to examine potential interactions between Pb exposure and a host of other environmental exposures, lifestyle factors, and key sociodemographics (e.g., race, sex, SES). As these interactions may be relevant for understanding the Pb-mortality relationship and creating interventions to mitigate it, additional work should be done in larger and more racially and socioeconomically diverse samples that allow for this exploration.

5. Conclusions

This study provides evidence for a negative relationship between exposure to Pb and age at death. Our choice of biomarkers leveraged different time resolution achieved for enamel, cementum, and bone, allowing for the exploration of higher granularity with the enamel measures, representing childhood, and longer cumulative averages with the cementum and bone representing adulthood and late life respectively. With this, we identified adulthood—and in particular later adulthood—as an especially relevant exposure period for this association—as opposed to childhood and late in life years closer to death. However, examining this question in a sample that includes younger individuals or those from a population without general declining Pb exposure is needed. Although it is surprising that childhood Pb was not significantly associated with earlier age at death in this study, there are a myriad documented adverse health effects of childhood Pb exposures.⁴ Further, given the short period of exposure averaged by our enamel Pb measure, the fact that the association trended negative was notable and further exploration of this in a larger sample would be valuable. Our findings do suggest that exposure timing across the life course may well be relevant for mortality—and possibly other health effects. Taking advantage of different exposure matrices can make this kind of assessment of exposures in different life stages possible. More work should be done leveraging these different measures simultaneously to tease out sensitive periods of Pb exposure on different health states.

Fig. 1.

Timing of Pb exposure.

Highlights:

• Exposure to Pb in childhood, adulthood, and late life were negatively associated with age at death

• Only adulthood remained significant after accounting for earlier exposure(s)

• First epidemiologic study to examine tooth cementum Pb in relation to a health outcome

• Measures in different parts of teeth can assess the total health impact of Pb