• diagnostic tool;
  • fasting plasma glucose;
  • HbA1c;
  • liver function tests;
  • pre-analytical bias

Diabet. Med. 29, 742–747 (2012)


Background  HbA1c is currently being introduced for diagnostic purpose in diabetes. Previous studies have, however, indicated that patients with liver disease have false low HbA1c levels. We therefore investigated the correlation between HbA1c and plasma glucose in patients with different levels of increased liver enzyme concentrations.

Methods  Data from 10 065 patients with simultaneous measurement of HbA1c, venous fasting plasma glucose, alanine aminotransferase and γ-glutamyl transferase were extracted from our laboratory database. Correlations were investigated in four patient groups divided according to their liver enzyme concentrations.

Results  The correlation between HbA1c and plasma glucose was high in all groups, with r = 0.77 for men and r = 0.78 for women (P < 0.001), a correlation confirmed with multiple regression analysis (P < 0.001). However, interaction analysis revealed that linear regression lines were significantly different for men and women, with increase of both liver enzyme measurements and also, for women, with increased alanine aminotransferase. When compared with biological variation for HbA1c, only men with increased measurements of both liver enzymes had a clinically important decrease in HbA1c.

Conclusions  Increased liver enzyme concentrations do not bias the correlation between HbA1c and fasting plasma glucose. However, men with low plasma glucose and increased concentrations of both liver enzymes do have a slightly decreased HbA1c and, if the clinical suspicion is strong enough, one should consider supplement testing.