Exposición pre y postnatal a mercurio y presión arterial en preescolares

Background and aims According to the World Health Organisation, mercury is one of the top ten chemicals or groups of chemicals that raise concerns about public health. High systolic blood pressure is the main risk factor for illness and death from any cause, and can be modified. Research in animal models has found increases in systolic blood pressure after exposure to methylmercury, the most toxic chemical form of mercury. In the adult population, a non-linear dose-response relation has been found between postnatal mercury exposure and the risk of arterial hypertension, with a change in trend around 3 µg/g of hair. To date no studies have been conducted to assess the effects of prenatal mercury exposure on blood pressure during adulthood. Previous studies in children have shown inconsistent results for both prenatal and postnatal exposure. The general aim of this thesis is to evaluate the relation between prenatal and postnatal mercury exposure and blood pressure in 4-year-old children in three cohorts from the Childhood and Environment (INMA) Project, as well as the role of possible confounders, effect modifiers and the shape of this relation. Methodology This doctoral thesis consists of two studies. First, a systematic review of the existing scientific literature on the relation between chronic mercury exposure and blood pressure in the paediatric age was conducted. Subsequently, the relation between prenatal and postnatal mercury exposure and blood pressure at 4 years of age was evaluated in children from the INMA Project. For the systematic review we searched for observational studies, in English, French and Spanish, published until 31 January 2019, in six databases (PubMed, Embase, Web of Science, Scopus, Lilacs and Índice Médico Español), using a previously defined search strategy. Free-access doctoral theses were also searched in the www.oatd.org repository. Studies were selected on the basis of eligibility criteria described elsewhere. The risk of bias of the primary studies was evaluated using an ad hoc instrument. The population included in the second study in this thesis consisted of 1295 4-year-old children who participated in the Asturias, Sabadell and Valencia cohorts of the INMA Project. Mercury exposure was assessed in umbilical cord blood (prenatal exposure) and hair at the age of 4 years (postnatal exposure), except in the Asturias cohort, where only data on prenatal exposure were available. Blood pressure was determined during a visit at the age of 4 years using oscillometric devices according to pre-established protocols in each cohort. Blood pressure was measured once in the Asturias and Sabadell cohorts and up to three measurements were recorded in the Valencia cohort. Physical examinations, questionnaires, scales and clinical records were used to collect information on anthropometric, sociodemographic, dietary, lifestyle and clinical characteristics at different times throughout the follow-up from the prenatal stage. Prenatal exposure to polychlorinated biphenyls (PCB 138, PCB 153 and PCB 180) was measured in the mother's blood in the first trimester of gestation. After logarithmic transformation of the mercury exposure and blood pressure variables, multiple linear regression models were constructed to assess the relation between prenatal and postnatal mercury exposure and blood pressure at 4 years of age. Models were adjusted for anthropometric, sociodemographic, environmental and dietary intake variables. The interaction of sex, fish consumption (by mother and child) and prenatal exposure to polychlorinated biphenyls in the relation between mercury and blood pressure was examined. The shape of the relation was evaluated using generalised additive models. A sensitivity analysis was performed with data from repeated blood pressure measurements from the Valencia cohort. Results The systematic review of the scientific literature identified nine articles related to five cohort studies, one case-control study and two cross-sectional studies. Mercury was analysed in different matrices and periods of exposure. The earlier articles, published between 1996 and 2012, included populations with an exposure level considered as very high for the general population (between 30 and 35 µg/L of umbilical cord blood and/or 4-8 µg/g of hair). Articles published between 2014 and 2018 assessed populations with substantially lower levels of exposure (around an order of magnitude lower). Four articles assessed prenatal exposure, two simultaneously considered prenatal and postnatal exposures, and three evaluated postnatal exposure. Blood pressure was measured according to different protocols. The association between mercury and blood pressure was adjusted for different covariates in each study. Four articles found a significant positive association between chronic mercury exposure and blood pressure in children or adolescents. Of these four, three of them assessed prenatal exposure. One of the six articles evaluating interaction by sex found some effects of prenatal mercury exposure on blood pressure only in adolescent boys. Three articles (two studies) examined the shape of the relation, finding a positive linear relation until prenatal exposures of approximately 10 µg/L of cord blood which flattened out at higher levels of exposure. In the assessment of the relation between mercury exposure and blood pressure in 1295 4-year-old boys and girls from the INMA Project (second study in this thesis), it was found that the geometric averages (95% confidence interval) of mercury in umbilical cord blood and in hair at 4 years were 9.2 μg/L (8.8-9.7) and 1.0 μg/g (0.9-1.0) respectively. The geometric mean (95% confidence interval) of the first valid measure of systolic/diastolic blood pressure (SBP/DBP) at age 4 years was 100.4 (100.0-101.1)/63.2 (62.5-64.0) mmHg. In the Valencia cohort the geometric mean of the first, second and third SBP/DBP measurements was 101.5 (100.2-102.8)/64.1 (62.9-65.4) mmHg, 99.0 (97.5-100.5)/61.8 (60.3-63.3) mmHg and 96.3 (94.5-98.1)/58.0 (56.3-59.8) mmHg respectively. After adjusting for covariates, prenatal mercury exposure was inversely related to SBP (β=(-0.0001), CI95% (-0.0140;0.0138)) and directly related to DBP (β=0.0040, CI95% (-0.0175;0.0255)), without reaching statistical significance. Postnatal mercury exposure was inversely and non-significantly related to SBP (β=(-0.0078), CI95% (-0.0194;0.0038)) and DBP (β=(-0.0028), CI95% ((-0.0206);0.0150)). Simultaneous assessment of prenatal and postnatal mercury exposure reinforced the inverse relation between prenatal mercury and SBP (β=(-0.0062), CI95% (-0.0232);0.0108)) and DBP (β=(-0.0072), CI95% (-0.0340;0.0195)), as well as the inverse relation between postnatal mercury and SBP (β=(-0.0081), CI95% (-0.0216;0.0054)) and DBP (β=(-0.0085), CI95% (-0.0297;0.0128)). Maternal fish consumption during gestation as well as prenatal exposure to polychlorinated biphenyls acted as confounders in the relation between prenatal mercury and blood pressure. No effect modification was found for mercury on blood pressure by sex, maternal fish consumption, child fish consumption or prenatal exposure to polychlorinated biphenyls. In terms of the shape of the relation, the best fit was a linear model, which was consistent with the absence of any association between prenatal and postnatal mercury and the systolic and diastolic blood pressure at 4 years of age. A subsequent sensitivity analysis found a modest significant inverse association between prenatal mercury exposure and systolic (β=(-0.0349), CI95% (-0.0688;-0.0009)) and diastolic blood pressure (β=(-0.0663), CI95% (-0.1248;-0.0078)). This inverse association was maintained by further adjusting for postnatal mercury exposure, and was found to be marginally significant with SBP (β=(-0.0335), CI95% (-0.0704;0.0034)) and significantly so with DBP (β= (-0.0675), CI95% (-0.1308;-0.0043)). Postnatal mercury exposure was inversely and non-significantly related to SBP (β=(-0.0095), CI95% (-0.0346;0.0156)) and DBP (β=(-0.0155), CI95% (-0.0571; 0.0261)). Further adjustment for prenatal mercury exposure weakened the relation between postnatal mercury and systolic (β=(-0.0027), CI95% (-0.0324;0.0269)) and diastolic blood pressure (β=0.0006, CI95% (-0.0503; 0.0514)). In general, in the sensitivity analysis, the imprecision of the effect estimates increased as the statistical power of the models was considerably reduced. It was found that, even though there is an association, the effect size is small and its clinical significance in childhood is uncertain. Conclusions The children in our study had elevated levels of mercury in umbilical cord blood and hair at 4 years of age, at intermediate levels among previous studies that evaluated the association between mercury and blood pressure. The levels of prenatal exposure in our study are relatively high compared to other recent exposure assessments, especially compared to other European populations. The results of our study support the idea that the use of the first measure or the average of the second and third measures of blood pressure is an important factor to be considered in studies that use blood pressure as an outcome variable, and may alter the direction of the associations studied and therefore the healthcare decisions. Increasingly lower blood pressure measurements were obtained in repeated measurements within the same visit in 4-year-old children, consistent with previous evidence. The results of our study would not support the existence of an association between prenatal and postnatal mercury exposure and blood pressure at 4 years of age. The results obtained in the sensitivity analysis seem to suggest that the foetal period would be the most vulnerable to exposure, although they should be considered with extreme caution due to the imprecision of the measures and lack of statistical power. Our study suggests that consumption of fish by the mother during gestation and prenatal exposure to polychlorinated biphenyls confounded the relation between prenatal mercury and blood pressure. The child's consumption of fish in the past year was not related to his or her blood pressure at 4 years of age. No effect modifications by sex were found in any of the relations studied. The shape of the relation was compatible with the absence of any association. There are still few studies that evaluate mercury exposure early in life, across a wide range of exposures, and its relation with blood pressure during the paediatric age. Studies published to date have shown inconsistent results. The designs are very heterogeneous, making it difficult to compare them and to draw conclusions about a possible association. Further research is warranted and should takes into consideration aspects such as improved assessment of mercury exposure and blood pressure measurement, the inclusion of new risk factors identified as possible confounders of the association, a comprehensive approach to environmental exposures and genetic and epigenetic aspects, as well as the monitoring of existing birth cohorts at different times of development, such as adolescence and the young adult stage.