Response to Amanzada on GGT paper

Dear Editor,

Amanzada and colleagues suggest that the ratio of GGT to ALT activities may be a better predictor than GGT alone for virological response to pegylated interferon and ribavirin in chronic hepatitis C (HCV). Accordingly, we have re-analyzed the data from the HALT-C trial, substituting GGT/ALT quintile for GGT quintile in all analyses. For the 1319 patients, the quintile distributions for GGT/ALT were 0.07–0.61, 0.62–0.97, 0.98–1.50, 1.51–2.47, and 2.48–22.0. The responses to treatment at week 12 (EVR), week 20, and week 72 (SVR) in the HALT-C lead-in phase are shown in the figure using the same format as Figure 1 in the published paper.(1) The GGT/ALT ratio predicted treatment response modestly better than GGT alone, especially for SVR. Per quintile increase in GGT/ALT, the odds ratio (OR) for failure to achieve SVR was 1.85 (95% confidence interval (CI) 1.61–2.13) compared with an odds ratio of 1.51 (95% CI 1.33–1.71) for GGT. Importantly, ALT/GGT performed no better than GGT alone in predicting clinical outcomes. For example the hazard ratio (HR) per quintile increase in GGT/ALT was 1.11, slightly lower than the HR of 1.14 for GGT alone.

Figure.

Association of quintile of GGT to ALT activity with probability of virological response during the lead-in phase of HALT-C. EVR, early virological response; SVR, sustained virological response

As noted in their letter, the authors have previously reported an odds ratio for SVR of 26.7 for IL28b rs12979860 genotype CC homozygotes versus other genotypes (CT or TT) among patients with genotype 1 infection and GGT/ALT <0.70.(2) This result was based on achievement of SVR in 45 of 52 (87%) patients with CC genotype and 42 of 78 (54%) patients with either CT or TT genotype. These SVR response rates result in an OR of 5.5, not 26.7. In HALT-C, using the same cut-off of 0.70 for GGT/ALT ratio, we found an OR of SVR for CC versus other genotypes of 1.56 (95% CI of 0.91–2.67) among patients with low GGT/ALT. Only 26% of HALT-C patients (who were all non-responders to prior treatment) had a GGT/ALT ratio of <0.70 compared to 49% of their patients.

Thus we found modest improvement in predicting virological response for GGT/ALT over GGT alone, little value of rs12979860 at low GGT/ALT, and no added value of GGT/ALT for clinical prognosis. Furthermore, we are unaware of a biological basis for the use of this ratio.