Aleitamento materno e metabolismo da glicose em mulheres no puerpério: avaliação da mediação via biomarcadores associados à resistência à insulina, inflamação ou disfunção endotelial
Arquivos
Data
2024-04-26
Autores
Oliveira, Julia Martins de 
Orientadores
Pititto, Bianca de Almeida
Tipo
Tese de doutorado
Título da Revista
ISSN da Revista
Título de Volume
Resumo
Introdução: A incidência de diabetes mellitus gestacional (DMG) vem aumentando exponencialmente nos últimos anos. É bem conhecido que o DMG é um importante fator de risco e preditor para o desenvolvimento de diabetes mellitus tipo 2 (DM2) mais tardiamente na vida da mulher. Estudos populacionais têm verificado que o aleitamento materno (AM) é capaz de diminuir o risco futuro de DM2. Observa-se que tal benefício é proporcional ao tempo de AM e, por motivos ainda não bem compreendidos, persiste por vários anos depois de cessado o AM. A fisiopatologia deste fenômeno ainda é pouco compreendida. Alguns biomarcadores tem sido relacionados na literatura aos mecanismos de resistência à insulina (RI), inflamação subclínica e disfunção endotelial - prolactina, adiponectina, copeptina, E-selectina, lipopolissacarídeos (LPS), zonulina e aminoácidos de cadeia ramificada (BCAA). Deste modo, questionamos se estes biomarcadores poderiam ser mediadores da associação entre AM e tolerância à glicose no período puerperal. Objetivo: Avaliar a associação do AM com os marcadores do metabolismo glicídico e de RI (glicemia jejum, glicemia de 2 horas, insulina de jejum, HOMA-IR, HOMA-Beta, índice TyG, razão Tg/HDL) e testar o papel mediador dos biomarcadores. Material e Métodos: 95 mulheres (45 com diagnóstico de diabetes mellitus gestacional) maiores que 18 anos e com IMC ≥ 25/kg/m2 que faziam seguimento nos ambulatórios de pré-natal da UNIFESP-EPM foram acompanhadas do primeiro/segundo ou terceiro trimestre de gestação até dois a seis meses após o parto, durante os anos de 2018 a 2020. As participantes responderam a questionários padronizados e foram submetidas a avaliação clínica ao longo da gestação e no período puerperal. No dia da avaliação puerperal (realizada entre dois e seis meses após o parto), foi realizada coleta de exames laboratoriais. Ao final do estudo as pacientes foram divididas em dois grupos de acordo com o status de AM no momento da visita pós-parto (grupo AM com n=44 e grupo não-AM com n=51). Os dados clínicos e laboratoriais foram comparados entre os grupos AM e não-AM usando o teste t de Student e Mann Whitney (variáveis contínuas de distribuição normal ou não, respectivamente) ou teste do qui-quadrado (variáveis categóricas) com o auxílio do programa Statistical Package for the Social Sciences®, versão 16.0 (SPSS Incorporation, 2000). Para identificação do conjunto mínimo de fatores de confusão para uso nos ajustes das análises multivariadas, usamos a metodologia do DAG (Direct Acyclic Graphics) pela plataforma DAGGITY®. Foram realizadas análises de regressão linear com ajuste multivariado e análise de mediação com auxílio do programa Statistics and Data Science®, versão 17.0 (STATA). O P foi considerado estatisticamente significativo se <0,05. Resultados: No Artigo 1 verificamos que os grupos AM (n=44) e não-AM (n=51) eram homogêneos em relação à prevalência de DMG, IMC pré-gestacional, ingesta calórica diária, atividade física e perda ponderal no pós-parto. O grupo AM tinha níveis significativamente maiores de prolactina em relação ao grupo não-AM [47.8(29.6-88.2) vs. 20.0(12.0-33.8) ng/dL, p<0.001]. Também encontramos níveis menores de média (DP) ou mediana (IIQ) de glicemia de jejum [89.0(8.0) vs. 93.9(12.6) mg/dL, p=0.04], triglicerídeos (TG) [92.2(37.9) vs. 122.4(64.4) mg/dL, p=0.01], relação TG/HDL [1.8(0.8) vs. 2.4(1.6), p=0.02], índice TyG [8.24(0.4) vs. 8.52/90.53), p=0.005], insulina de jejum [8.9(6.3-11.6) vs. 11.4(7.7-17.0)μU/mL, p=0.048] e HOMA-IR [2.0(1.3-2.7) vs. 2.6(1.6-3.9), p=0.025] no grupo AM. A análise de Regressão linear com ajustes para potenciais fatores confundidores identificados pelo DAG (paridade, escolaridade, IMC pré-gestacional, DMG, via de parto, ganho ponderal durante a gestação e peso ao nascer) evidenciou uma associação estatisticamente significativa entre o AM e a glicemia de jejum [-6,37 (-10.91 a -1.83), p=0.006), o HOMA-IR [-0.27 (-0.51 a -0.04), p=0.024), o índice TyG [-004 (-0.06 a -0.02), p=0.001) e a relação TG/HDL [-0.25 (-0.48 a -0.01), p=0.038). A análise de mediação evidenciou que a prolactina não mediou as associações independentes entre AM e marcadores de metabolismo glicídico e RI, evidenciadas nos resultados da regressão linear. No Artigo 2 analisamos os grupos AM (n=44) e não-AM (n=51), considerando os biomarcadores de inflamação subclínica e disfunção endotelial (adiponectina, copeptina, E-selectina, LPS, zonulina e BCAA) como potenciais mediadores na relação entre amamentação e metabolismo glicídico e RI. Neste trabalho, verificamos que apenas os níveis de LPS foram estatisticamente diferentes entre os grupos, sendo que níveis mais baixos foram vistos no grupo AM: [6.8 (4.2-10.6) vs. 9.2 (6.9-11.5) ng/mL, p=0.048]. Nas análises de regressão linear com ajustes para fatores de confundimento (paridade, escolaridade, IMC pré-gestacional, ganho ponderal na gestação, DMG e via de parto), novamente evidenciamos associação inversa com significância estatística entre AM e glicemia de jejum [-6,30 (-10.71 a -1.89), p=0.005), HOMA-IR [-0.28 (-0.50 a -0.05), p=0.017], índice TyG [-004(-0.06 a -0.01), p=0.001] e razão TG/HDL [-0.23 (-0.46 a -0.01), p=0.001]. Criamos uma variável latente com uso do programa estatístico STATA® e a denominamos “inflamação subclínica” (IS). Nessa metodologia, a variável não é diretamente aferida, mas é inferida a partir dos 6 biomarcadores (adiponectina, copeptina, E-selectina, LPS, zonulina e BCAA). O uso desta variável torna mais robusta a análise estatística no sentido de compreender se inflamação subclínica pode mediar a melhora nos marcadores de metabolismo glicídico e RI observada no grupo AM. A IS não mediou, pela análise estatística de mediação, a melhora observada nos marcadores de RI e metabolismo glicídico mostrada no grupo de AM. Conclusão: O AM foi associado a melhora nos marcadores de metabolismo glicídico e RI após ajustes para fatores de confundimento, mesmo com um curto prazo de dois a seis meses após o parto em mulheres com e sem DMG. A prolactina e os biomarcadores relacionados à inflamação subclínica e disfunção endotelial não foram mediadores da associação vista entre AM e melhora nos marcadores de metabolismo glicídico e RI.
Introduction: The incidence of gestational diabetes mellitus (GDM) has been increasing in the last years. GDM is a well-known risk factor and a predictor for the future development of type 2 diabetes mellitus (T2DM). Populational studies have found that breastfeeding (BF) can reduce the future risk of T2DM and this phenomenon appears to be proportional to the extent of BF. For reasons that are not yet well understood, this benefit also persists for several years after weaning. The pathophysiology of this phenomenon is still poorly understood. A series of biomarkers have been related in the literature to insulin resistance (IR), subclinical inflammation and endothelial dysfunction. We questioned whether some of these biomarkers could be mediators of the association between BF and glucose tolerance and IR in the postpartum period. Objective: To evaluate the association between breastfeeding (BF) and markers of glucose metabolism and insulin resistance (fasting blood glucose, 2-hour blood glucose, fasting insulin, HOMA-IR, HOMA-Beta, TyG index, and Tg/HDL ratio) and to test the mediation role of biomarkers related to mechanisms of insulin resistance, subclinical inflammation and endothelial dysfunction (prolactin, adiponectin, copeptin, E-selectin, lipopolysaccharides - LPS, zonulin and branched-chain amino acids - BCAA). Material and Methods: 95 women (45 diagnosed with gestational diabetes mellitus - GDM) with age over 18 years and a BMI ≥ 25/kg/m2 who were being followed up at the UNIFESP-EPM prenatal clinics participated in this study. They were followed from the first/second or third trimester until two to six months after birth. They answered structured questionnaires and underwent clinical evaluation throughout pregnancy and postpartum. On the day of the puerperal assessment (carried out between two and six months after birth), laboratory tests were collected. At the end of the study, patients were divided into two groups according to their breastfeeding status at the time of the postpartum visit (BF group with n=44 and non-BF group with n=51). Clinical and laboratory data were compared between the BF and the non-BF groups using the Student and Mann Whitney t-test or chi-square test. Linear regression analysis with multivariate adjustment and mediation analysis were also performed. Statistical programs used were the Statistical Package for the Social Sciences®, version 16.0 (SPSS) Incorporation, 2000) and the Statistics and Data Science® program, version 17.0 (STATA). P was considered statistically significant if <0.05. Results: In the first paper, we found that the BF and non-BF groups were homogeneous concerning the prevalence of gestational diabetes mellitus (GDM), pre-pregnancy BMI, daily caloric intake, physical activity, and postpartum weight loss. The BF group had significantly higher prolactin levels compared to the non-BF group [47.8(29.6-88.2) vs. 20.0(12.0-33.8) ng/dL, p<0.001]. We also found lower fasting blood glucose levels [89.0(8.0) vs. 93.9(12.6) mg/dL, p=0.04], triglycerides (TG) [92.2(37.9) vs. 122.4(64.4) mg/dL, p=0.01], TG/HDL ratio [1.8(0.8) vs. 2.4(1.6) mg/dL, p=0.02], TyG index [8.24(0.4) vs. 8.52/90.53), p=0.005], fasting insulin [8.9(6.3-11.6) vs. 11.4(7.7-17.0) μU/mL, p=0.048] and HOMA-IR [2.0(1.3-2.7) vs. 2.6(1.6-3.9), p=0.025] in the BF group. Linear regression analysis with adjustments for confounding factors (parity, scholarity, pre-gestational BMI, GDM, mode of delivery, weight gain during pregnancy and birth weight) showed a statistically significant association between BF and fasting blood glucose [-6.37 (-10.91 to -1.83), p=0.006), HOMA-IR [-0.27 (- 0.51 to -0.04), p=0.024), TyG index [-004 (-0.06 to -0.02), p=0.001) and the TG/HDL ratio [-0.25 (-0.48 to -0.01), p=0.038). The mediation analysis showed that prolactin did not mediate the results from the linear regression analysis. In the second article, we analyzed the same population and conducted additional statistical analysis with the other biomarkers of interest (adiponectin, copeptin, E-selectin, LPS, zonulin, and BCAA). In this work, we verified that only LPS levels were different between the groups, with lower levels being seen in the BF group: [6.8 (4.2-10.6) vs. 9.2 (6.9-11.5) ng/mL, p=0.048]. An inverse association with statistical significance after adjustments for confounding factors (parity, scholarity, pre-gestational BMI, weight gain during pregnancy, GDM and mode of delivery), was found between BF and fasting blood glucose [-6.30 (-10.71 to -1.89), p=0.005), HOMA-IR [-0.28 (-0.50 to -0.05), p=0.017], TyG index [-004(-0.06 to -0.01), p=0.001] and TG/HDL ratio [-0.23 (-0.46 to -0.01), p=0.001]. We created a latent variable using the STATA® statistical program and called it “subclinical inflammation” (SI). The variable is not directly measured but is inferred from the six biomarkers: adiponectin, copeptin, E-selectin, LPS, zonulin, and BCAA. Using this variable makes the statistical analysis more robust in understanding whether subclinical inflammation can mediate the improvement in markers of insulin resistance observed in the BF group. SI did not mediate the improvement observed in markers of insulin resistance in the BF group. Conclusion: Two to six months of BF was associated with improvement in markers of insulin resistance after adjusting for confounding factors. The biomarkers evaluated did not mediate the association between BF and improvement in markers of insulin resistance.
Introduction: The incidence of gestational diabetes mellitus (GDM) has been increasing in the last years. GDM is a well-known risk factor and a predictor for the future development of type 2 diabetes mellitus (T2DM). Populational studies have found that breastfeeding (BF) can reduce the future risk of T2DM and this phenomenon appears to be proportional to the extent of BF. For reasons that are not yet well understood, this benefit also persists for several years after weaning. The pathophysiology of this phenomenon is still poorly understood. A series of biomarkers have been related in the literature to insulin resistance (IR), subclinical inflammation and endothelial dysfunction. We questioned whether some of these biomarkers could be mediators of the association between BF and glucose tolerance and IR in the postpartum period. Objective: To evaluate the association between breastfeeding (BF) and markers of glucose metabolism and insulin resistance (fasting blood glucose, 2-hour blood glucose, fasting insulin, HOMA-IR, HOMA-Beta, TyG index, and Tg/HDL ratio) and to test the mediation role of biomarkers related to mechanisms of insulin resistance, subclinical inflammation and endothelial dysfunction (prolactin, adiponectin, copeptin, E-selectin, lipopolysaccharides - LPS, zonulin and branched-chain amino acids - BCAA). Material and Methods: 95 women (45 diagnosed with gestational diabetes mellitus - GDM) with age over 18 years and a BMI ≥ 25/kg/m2 who were being followed up at the UNIFESP-EPM prenatal clinics participated in this study. They were followed from the first/second or third trimester until two to six months after birth. They answered structured questionnaires and underwent clinical evaluation throughout pregnancy and postpartum. On the day of the puerperal assessment (carried out between two and six months after birth), laboratory tests were collected. At the end of the study, patients were divided into two groups according to their breastfeeding status at the time of the postpartum visit (BF group with n=44 and non-BF group with n=51). Clinical and laboratory data were compared between the BF and the non-BF groups using the Student and Mann Whitney t-test or chi-square test. Linear regression analysis with multivariate adjustment and mediation analysis were also performed. Statistical programs used were the Statistical Package for the Social Sciences®, version 16.0 (SPSS) Incorporation, 2000) and the Statistics and Data Science® program, version 17.0 (STATA). P was considered statistically significant if <0.05. Results: In the first paper, we found that the BF and non-BF groups were homogeneous concerning the prevalence of gestational diabetes mellitus (GDM), pre-pregnancy BMI, daily caloric intake, physical activity, and postpartum weight loss. The BF group had significantly higher prolactin levels compared to the non-BF group [47.8(29.6-88.2) vs. 20.0(12.0-33.8) ng/dL, p<0.001]. We also found lower fasting blood glucose levels [89.0(8.0) vs. 93.9(12.6) mg/dL, p=0.04], triglycerides (TG) [92.2(37.9) vs. 122.4(64.4) mg/dL, p=0.01], TG/HDL ratio [1.8(0.8) vs. 2.4(1.6) mg/dL, p=0.02], TyG index [8.24(0.4) vs. 8.52/90.53), p=0.005], fasting insulin [8.9(6.3-11.6) vs. 11.4(7.7-17.0) μU/mL, p=0.048] and HOMA-IR [2.0(1.3-2.7) vs. 2.6(1.6-3.9), p=0.025] in the BF group. Linear regression analysis with adjustments for confounding factors (parity, scholarity, pre-gestational BMI, GDM, mode of delivery, weight gain during pregnancy and birth weight) showed a statistically significant association between BF and fasting blood glucose [-6.37 (-10.91 to -1.83), p=0.006), HOMA-IR [-0.27 (- 0.51 to -0.04), p=0.024), TyG index [-004 (-0.06 to -0.02), p=0.001) and the TG/HDL ratio [-0.25 (-0.48 to -0.01), p=0.038). The mediation analysis showed that prolactin did not mediate the results from the linear regression analysis. In the second article, we analyzed the same population and conducted additional statistical analysis with the other biomarkers of interest (adiponectin, copeptin, E-selectin, LPS, zonulin, and BCAA). In this work, we verified that only LPS levels were different between the groups, with lower levels being seen in the BF group: [6.8 (4.2-10.6) vs. 9.2 (6.9-11.5) ng/mL, p=0.048]. An inverse association with statistical significance after adjustments for confounding factors (parity, scholarity, pre-gestational BMI, weight gain during pregnancy, GDM and mode of delivery), was found between BF and fasting blood glucose [-6.30 (-10.71 to -1.89), p=0.005), HOMA-IR [-0.28 (-0.50 to -0.05), p=0.017], TyG index [-004(-0.06 to -0.01), p=0.001] and TG/HDL ratio [-0.23 (-0.46 to -0.01), p=0.001]. We created a latent variable using the STATA® statistical program and called it “subclinical inflammation” (SI). The variable is not directly measured but is inferred from the six biomarkers: adiponectin, copeptin, E-selectin, LPS, zonulin, and BCAA. Using this variable makes the statistical analysis more robust in understanding whether subclinical inflammation can mediate the improvement in markers of insulin resistance observed in the BF group. SI did not mediate the improvement observed in markers of insulin resistance in the BF group. Conclusion: Two to six months of BF was associated with improvement in markers of insulin resistance after adjusting for confounding factors. The biomarkers evaluated did not mediate the association between BF and improvement in markers of insulin resistance.
Descrição
Citação
OLIVEIRA, Julia Martins de. Aleitamento materno e metabolismo da glicose em mulheres no puerpério: avaliação da mediação via biomarcadores associados à resistência à insulina, inflamação ou disfunção endotelial. 2024. 111 f. Tese (Doutorado em Endocrinologia e Metabologia) – Escola Paulista de Medicina, Universidade Federal de São Paulo, São Paulo, 2024.