Associations of Coffee Consumption With Liver Injury Markers
Associations of Coffee Consumption With Liver Injury Markers
Participant characteristics by categories of total coffee consumption are presented in Table 1. In this population, 79 % of individuals were consuming at least one type of coffee. Amounts of coffee consumed however were relatively modest with a median total coffee intake of 0.50 servings/day [interquartile range (IQR), 0.04–0.79]. Approximately 66 % of the population consumed caffeinated coffee while decaffeinated coffee intake was less common with only 28 % of individuals consuming this type. Ethnicity, level of education and smoking and alcohol consumption behaviours differed significantly across categories of total coffee consumption (Table 1). The highest category of total coffee consumption had the lowest proportion of participants with impaired glucose tolerance, however S I, BMI, waist circumference, family history of diabetes and energy expended did not differ significantly across categories of total coffee consumption. Energy intake, percent of energy from saturated fat and regular soda consumption differed significantly across categories where the highest energy intakes and saturated fat consumption were found in the highest categories of coffee consumption. Caffeinated and decaffeinated coffee consumption were inversely correlated (Spearman r = −0.20, p <0.0001). ALT and fetuin-A were inversely correlated with S I (ALT Spearman r = −0.27, p <0.0001; fetuin-A Spearman r = −0.16, p <0.0001), and positively correlated with each other (Spearman r = 0.11, p = 0.0057).
Caffeinated coffee consumption was inversely associated with ALT, AST and NAFLD liver fat score, and, in the case of ALT, there was a stronger magnitude of association across staged multivariate models (Table 2, model 3: ALT β = −0.08, p = 0.0111; AST β = −0.05, p = 0.0155; NAFLD liver fat score β = −0.05, p = 0.0293). Caffeinated coffee consumption however, showed no significant association with fetuin-A (Table 2, model 3: β = 0.04, p = 0.17). There were no significant associations of decaffeinated coffee consumption with markers of liver injury.
Associations of coffee with markers of liver injury were not modified by age, sex, ethnicity, BMI or S I (all p > 0.05).
Multivariate linear regression analyses were also conducted with total coffee as the exposure variable. For the liver enzymes, results of regression analyses with total coffee consumption were consistent with those of caffeinated coffee consumption, with significant associations of total coffee intake with both ALT and AST (Additional file 1: Table S1, http://www.biomedcentral.com/1471-230X/15/88/additional model 3: ALT β = −0.07, p = 0.0177; AST β = −0.05, p = 0.0131). Total coffee consumption was not however, associated with NAFLD liver fat score (Additional file 1: Table S1, http://www.biomedcentral.com/1471-230X/15/88/additional model 3: β = −0.04, p = 0.11) or fetuin-A, (Additional file 1: Table S1, http://www.biomedcentral.com/1471-230X/15/88/additional model 3: β = 0.02, p = 0.47).
When the heaviest alcohol consumers were excluded, the significant inverse association of caffeinated coffee with ALT, AST and NAFLD liver fat score not only remained, but, in the case of ALT, there was a stronger magnitude of association compared to the initial analysis (Table 3, model 3: ALT β = −0.11, p = 0.0037; AST β = −0.05, p = 0.0330; NAFLD liver fat score β = −0.06; p = 0.0298). With additional adjustment for S I in a fourth (mechanistic) model, the significant inverse association of caffeinated coffee with ALT remained, however the association with AST and NAFLD liver fat score was attenuated to non-significance (Table 3, model 4: ALT β = −0.08, p = 0.0400; AST β = −0.03, p = 0.20; NAFLD liver fat score β = −0.03, p = 0.27). Associations of caffeinated coffee with fetuin-A (Table 3, model 4: β = 0.03, p = 0.44), and decaffeinated coffee with all liver markers remained unchanged from the original analysis (data not shown).
Results
Participant Characteristics
Participant characteristics by categories of total coffee consumption are presented in Table 1. In this population, 79 % of individuals were consuming at least one type of coffee. Amounts of coffee consumed however were relatively modest with a median total coffee intake of 0.50 servings/day [interquartile range (IQR), 0.04–0.79]. Approximately 66 % of the population consumed caffeinated coffee while decaffeinated coffee intake was less common with only 28 % of individuals consuming this type. Ethnicity, level of education and smoking and alcohol consumption behaviours differed significantly across categories of total coffee consumption (Table 1). The highest category of total coffee consumption had the lowest proportion of participants with impaired glucose tolerance, however S I, BMI, waist circumference, family history of diabetes and energy expended did not differ significantly across categories of total coffee consumption. Energy intake, percent of energy from saturated fat and regular soda consumption differed significantly across categories where the highest energy intakes and saturated fat consumption were found in the highest categories of coffee consumption. Caffeinated and decaffeinated coffee consumption were inversely correlated (Spearman r = −0.20, p <0.0001). ALT and fetuin-A were inversely correlated with S I (ALT Spearman r = −0.27, p <0.0001; fetuin-A Spearman r = −0.16, p <0.0001), and positively correlated with each other (Spearman r = 0.11, p = 0.0057).
Regression Analyses
Caffeinated coffee consumption was inversely associated with ALT, AST and NAFLD liver fat score, and, in the case of ALT, there was a stronger magnitude of association across staged multivariate models (Table 2, model 3: ALT β = −0.08, p = 0.0111; AST β = −0.05, p = 0.0155; NAFLD liver fat score β = −0.05, p = 0.0293). Caffeinated coffee consumption however, showed no significant association with fetuin-A (Table 2, model 3: β = 0.04, p = 0.17). There were no significant associations of decaffeinated coffee consumption with markers of liver injury.
Interaction and Sensitivity Analyses
Associations of coffee with markers of liver injury were not modified by age, sex, ethnicity, BMI or S I (all p > 0.05).
Multivariate linear regression analyses were also conducted with total coffee as the exposure variable. For the liver enzymes, results of regression analyses with total coffee consumption were consistent with those of caffeinated coffee consumption, with significant associations of total coffee intake with both ALT and AST (Additional file 1: Table S1, http://www.biomedcentral.com/1471-230X/15/88/additional model 3: ALT β = −0.07, p = 0.0177; AST β = −0.05, p = 0.0131). Total coffee consumption was not however, associated with NAFLD liver fat score (Additional file 1: Table S1, http://www.biomedcentral.com/1471-230X/15/88/additional model 3: β = −0.04, p = 0.11) or fetuin-A, (Additional file 1: Table S1, http://www.biomedcentral.com/1471-230X/15/88/additional model 3: β = 0.02, p = 0.47).
When the heaviest alcohol consumers were excluded, the significant inverse association of caffeinated coffee with ALT, AST and NAFLD liver fat score not only remained, but, in the case of ALT, there was a stronger magnitude of association compared to the initial analysis (Table 3, model 3: ALT β = −0.11, p = 0.0037; AST β = −0.05, p = 0.0330; NAFLD liver fat score β = −0.06; p = 0.0298). With additional adjustment for S I in a fourth (mechanistic) model, the significant inverse association of caffeinated coffee with ALT remained, however the association with AST and NAFLD liver fat score was attenuated to non-significance (Table 3, model 4: ALT β = −0.08, p = 0.0400; AST β = −0.03, p = 0.20; NAFLD liver fat score β = −0.03, p = 0.27). Associations of caffeinated coffee with fetuin-A (Table 3, model 4: β = 0.03, p = 0.44), and decaffeinated coffee with all liver markers remained unchanged from the original analysis (data not shown).
Source...