The application of biochemical markers to detect heavy alcohol use in women has shown disappointing results until now. We evaluated carbohydrate-deficient transferrin (CDT) by the CDTcct® method and γ-glutamyltransferase (GGT) in a large cohort of alcohol-using perimenopausal women studied primarily for osteoporosis.
CDT and GGT were measured in 431 women aged 46 to 54 years, who were selected from a large cohort (n= 8503) of pre-, peri-, and postmenopausal women. Their alcohol intake was known from questionnaires and face-to-face interviews. Three groups were constructed for statistical analysis: those drinking on average less than 7 alcoholic drinks per week (n= 103), those drinking 7 to 34 per week (n= 280), and those drinking at least 35 per week (n= 48).
The mean values of CDT and GGT of the three groups increased with an increasing alcohol intake, but there was a poor correlation between CDT and GGT in the complete study group (r= 0.3). The specificities of CDT and GGT were comparable, 83% and 78%, respectively. The sensitivities for CDT and GGT were 30% and 50%, respectively. A logistic regression model could assign, overall, 77% of the women correctly in relation to their alcohol intake: 43% of the women drinking at least 35 drinks per week and 92% of the women drinking less than 7 drinks per week.
The test characteristics of both GGT and CDT are not good enough to be used as biochemical markers for detecting heavy alcohol use in women. The use of a logistic regression model offers an advantage, because both numeric values of CDT and GGT are taken into account instead of arbitrary cutoff values.