Table 2

Correlation coefficients (standardized beta values) derived from multiple linear regression of key metabolite to lifestyle factors.

Metabolite

Chemical shift

Sample number

Correlation to Y variable


(ppm)

number

(U-Cd)

age

sex

smoking statusa


3-HV

1.276 to 1.269

160

-0.056

-0.326***

-0.233**

-0.023


DMG

2.936 to 2.920

176

-0.127

-0.122

-0.168

-0.052


citrate (1)

2.515 to 2.590

175

0.194*

-0.078

0.126

-0.298***


citrate (2)

2.515 to 2.590

143

0.228*

-0.060

0.157

-0.203*


citrate (3)

2.515 to 2.590

108

0.251*

-0.046

0.121

n/a


Creatinine

4.073 to 4.040

178

-0.010

-0.427***

-0.395***

0.015


Creatine

3.940 to 3.930

178

-0.006

0.088

0.321***

0.150


4-DEA

1.118 to 1.108

152

-0.042

-0.370***

-0.076

0.039


aSmoking status was classified into three classes. 1 = never smoked, 2 = past smoker and 3 = current smoker. Specific exclusions were made for each metabolite identified where coinciding interferences made the relevant resonance unsuitable for integration. All six metabolites were correlated to at least one of the epidemiological factors in either the PLS/PLS-DA regression analysis or the covariance/correlation models. Asterisks denote statistical significance (ANOVA with Bonferroni post-hoc correction): * P < 0.05, ** P < 0.01, *** P < 0.001. U-Cd was log normally distributed.

citrate (1) all three classes included (never smoked, past- and current-smoker). Note, three samples excluded. citrate (2) current-smokers additionally excluded. citrate (3) past- and current-smokers additionally excluded. ANOVA, analysis of variance; 3-HV, 3-hydroxyisovalerate; DMG, dimethylglycine; 4-DEA, 4-deoxyerythronic acid; PLS, partial least squares; PLS-DA, partial least squares - discriminant analysis; U-Cd, urinary cadmium.

Ellis et al. BMC Medicine 2012 10:61   doi:10.1186/1741-7015-10-61

Open Data