Putting teeth into the developmental origins hypothesis: A longitudinal study of early childhood malnutrition, enamel hypoplasia and adolescent health in Amazonian Bolivia

Thesis (Ph.D.)--University of Washington, 2016-06

Adult teeth may chronologically reflect early childhood experience because enamel on the permanent teeth calcifies incrementally during early childhood and is sensitive to physiological stress. Defects in the enamel do not repair after occurrence or during the life course, leaving a permanent biological mark of physiological insults that occurred during early childhood. These characteristics suggest enamel defects may serve as a useful biomarker of chronic malnutrition and thus predictor of long-term health. This dissertation sought to investigate associations between malnutrition-related early childhood exposures and dental enamel hypoplasia (EH) and to evaluate EH as a predictor of adolescent anthropometrics and biomarkers. We conducted analyses using data from 349 Tsimane’ adolescents in Amazonian Bolivia, collected between 2002-2010 and in 2015. In 2015, we examined EH in the permanent maxillary incisors and mandibular canines using digital photography from which the following measures of EH were abstracted: occurrence (any, none), extent of occurrence (<1/3, 1/3-2/3, >2/3 of the tooth surface) and estimated age at occurrence (1, 2, 3, 4 years of age). Data on malnutrition-related early childhood exposures (1-4 years of age) were collected between 2002 and 2010, including stunted growth (height-for-age z-scores, HAZ), underweight (weight-for-age z-scores, WAZ), anemia (hemoglobin), immune activation (c-reactive protein) and parasitic gastrointestinal infection (hookworm infection). Adolescent outcomes (10-17 years of age) were collected in 2015 and included anthropometrics (height, weight, body mass index (BMI)) and biomarkers (hemoglobin (Hb), glycated hemoglobin (HbA1c), white blood cell count (WBC) and blood pressure). First, we evaluated the reliability of EH measurement using digital photographs and the Modified DDE (developmental defects in the enamel) Index by investigating inter- and intra-rater reliability and evaluating the frequency of EH detection across examiners for systemic biases. Next, we investigated associations between several malnutrition-related childhood exposures and EH in the permanent central maxillary incisors using multivariate log-binomial and ordinal logistic regression as well as generalized estimating equations (GEE). Finally, we investigated EH in the permanent central maxillary incisors (as a marker of early childhood experience) in relation to anthropometrics and biomarkers in adolescence using multivariate linear regression. We further evaluated the accuracy of EH in the permanent central maxillary incisors as a marker of chronic malnutrition and utility as a predictor of subsequent health outcomes compared to growth stunting through sensitivity, specificity and receiver operating characteristics (ROC) analyses. Our evaluation of the digital photographs revealed a rough, cobblestone-like EH pattern in the tooth surface that was particularly prevalent in the study sample (92.3%) and was thus the focus of this project. EH detection was most common on the central maxillary incisors (87%) compared to the lateral maxillary incisors (63%) and mandibular canines (26%). The intra-examiner reliability for detecting EH occurrence on the central maxillary incisors was very good (mean kappa=0.77) and greater than the lateral maxillary incisors (mean kappa=0.68) and mandibular canines (mean kappa=0.49). However, the inter-examiner reliability was fair to poor, though the inter-examiner reliability was also better for the maxillary incisors (mean kappa= 0.29) than the mandibular canines (mean kappa=0.17). The study sample had a high prevalence of childhood malnutrition, demonstrated by prevalence of stunted linear growth (75.2%), anemia (56.9%), elevated immune activation (39.1%), and gastrointestinal hookworm infection (49.6%) between 1 and 4 years of age. Results indicated an association between average childhood HAZ (PR=0.98, 95% CI: 0.95, 1.00), CRP levels (PR=1.01, 95% CI: 1.00, 1.03) and presence of gastrointestinal hookworm infection (OR=0.28, 95% CI: 0.08, 0.94 for <1/3 vs. >2/3 of the tooth affected by EH) and EH, though some of the point estimates lacked statistical precision. Extent of the tooth surface affected by EH seemed to be an important measure of EH as it related to early childhood exposures, particularly for average HAZ and hookworm infection. Greater extent of EH on the tooth surface was also associated with adolescent outcomes, including shorter height (-0.14 HAZ, 95% CI: -0.24, -0.03 and -1.35 cm, 95% CI: -2.21, -0.50), lower weight (-0.98 kg, 95% CI: -1.73, -0.23), lower Hb (-0.36 g/dL, 95% CI: -0.59, -0.13), lower HbA1c (-0.04 %A1c, 95% CI: -0.08, -0.00), and higher WBC count (0.74 109/L, 95% CI: 0.35, 1.14) but not BMI-for-age z-score or blood pressure. EH extent was associated with anemia (PR=1.08, 95% CI: 1.00, 1.18) and elevated WBC count (PR=1.12, 95% CI: 1.01, 1.26) based on public health and clinically-relevant thresholds. When evaluated against growth stunting as the “gold standard” (HAZ< -2.0: prevalence = 62.3% in childhood and 34.1% in adolescence), EH had high sensitivity (93% in childhood, 96% in adolescence) and low specificity (11% in childhood, 10% in adolescence). EH extent was a more accurate marker of childhood stunting (AUC 0.56) than EH occurrence (AUC 0.52) due to increased specificity (0.36 vs. 0.11). The addition of EH extent to a set of markers for childhood stunting (gender and adolescent HAZ) only slightly improved the AUC (0.77 vs. 0.76, p=ns). The AUC for adolescent WBC count was greater for EH extent (AUC 0.61) than for childhood HAZ (AUC 0.58, p=ns) and adolescent HAZ (AUC 0.59, p=ns). Addition of EH extent to sets of markers for adolescent health outcomes (age, gender, childhood HAZ and adolescent HAZ) improved the AUC for nearly all outcomes. In conclusion, we detected an EH pattern that was nearly ubiquitous in the study sample, but the rough, cobblestone-like hypoplastic pattern does not fit the typical linear/grooved pattern described in the overwhelming majority of the malnutrition literature. The pattern does not provide evidence in support of a systemic cause. Intra-examiner reliability results suggest that digital photography is a reproducible method for capturing EH, particularly for the central maxillary incisors. The inter-examiner reliability results bring into question the reliability of the digital photography method, but may be explained by systemic biases between the examiners, the subjective measures included in the Modified DDE Index, insufficient examiner training, and the very high prevalence of EH in the study sample. Improvements in examiner training and the measurement index used to classify EH would likely improve inter-examiner reliability. We provided evidence in support of a relationship between early childhood chronic malnutrition (HAZ), immune activation (CRP), parasitic infection (helminth infection) and EH and between EH extent and several adverse anthropometric and biomarker measures, including shorter height and lower weight, lower hemoglobin and greater WBC count. Given that chronic malnutrition and adverse health outcomes are associated with increased mortality, our findings are in line with the bioarchaeological findings. EH extent also seemed to capture a childhood exposure relevant to adolescent HAZ, hemoglobin and WBC count outcomes above and beyond that of childhood anthropometrics. Although not a strong proxy measure for chronic malnutrition, EH extent may be an important measure for predicting adolescent health outcomes. EH extent may serve as a useful proxy measure of childhood experience among adolescents in settings where childhood stunting data is not available. Furthermore, EH extent may capture childhood exposures relevant to adolescent health outcomes, particularly WBC count, that are not captured by childhood or adolescent HAZ and may thus be a useful addition to the “toolkit” of chronic malnutrition markers. This project makes a unique contribution to the existing literature because it prospectively recorded multiple early childhood exposures (beyond height or stunted growth) and had the minimum follow-up time necessary for full eruption of the permanent dentition to demonstrate an association between malnutrition-related childhood exposures, EH in the permanent dentition and adverse adolescent health outcomes. Subsequent work that builds on this project will be directed toward improving measurement of EH, including further characterization of the spectrum of enamel defects observed in the human dentition, systematically investigating EH etiology across populations and further developing EH as a useful predictor of long-term health by evaluating additional health outcomes, associations in more populations and employing advanced methodology.