To the Editor:
We enjoyed the well-written review by Ratziu et al.1 on the role of insulin sensitizers in nonalcoholic steatohepatitis (NASH). However, we would like to correct two minor errors regarding our study2 and share our ongoing efforts that address some of the knowledge gaps highlighted by the authors.
Table 1 in Ratziu et al.'s review states that we recruited only patients with diabetes. As reported elsewhere,3, 4 this is incorrect. We designed the study in 2002, and within the context of the emerging liver toxicity associated with troglitazone, we felt that exposure to a thiazolidinedione (TZD) should be reserved for NASH patients with type 2 diabetes mellitus (T2DM), or nondiabetic patients at risk of developing T2DM (i.e., impaired glucose tolerance [IGT]), so that the risk/benefit ratio of treatment would favor patients at least by improving glucose metabolism (the progression from IGT to T2DM is ≈6%-10% per year). Therefore, at study entry, patients were screened with an oral glucose tolerance test. Only 14% of all patients screened (n = 70) had known T2DM. Among those patients believed to have normal glucose metabolism (n = 60), 49% had IGT, and 30% were diagnosed with new-onset T2DM, whereas only 21% had normal glucose metabolism (the latter patients were excluded from the study). These results are similar to more recent work by our group.5 The proportion of patients with IGT and T2DM in each group was evenly distributed (≈50% in each): 13 and 14 in the pioglitazone arm and 12 and 9 in the placebo arm, respectively. Histology scores at baseline as well as the response to pioglitazone were similar whether patients had IGT or T2DM. The misconception that only patients with diabetes were included, and the possibility (still to be proven) that they respond better to TZDs, is frequently advocated as an explanation for the apparent better response to treatment in this study compared with more recent RCT with TZDs.6-8
A second error appears in our study2 that the nonalcoholic fatty liver disease activity score (NAS) “improved in 46% of pioglitazone-treated patients by ≥2 points versus 14% in the placebo group (P = 0.02), although this could be due to steatosis reduction, which is part of the score.” The 46% versus 14% difference did not include steatosis, but applied only to a ≥2 reduction in the combined necroinflammation score. Using the NAS (including necrosis, inflammation, and steatosis), improvement with pioglitazone occurred in 73% compared with 24% of placebo-treated patients (P < 0.001). If the PIVENS8 primary endpoint is used to analyze our data (improvement in ≥2 grades in the NAS with at least a 1-point improvement in hepatocellular ballooning and a 1-point improvement in either the lobular inflammation or steatosis score, and no increase in the fibrosis score) in patients with ballooning at baseline (all but 4), response to pioglitazone versus placebo treatment was observed in 60% versus 19%, somewhat better than the 47% versus 23% observed in PIVENS, but was highly significant in both studies (P < 0.01). If the previous outcome measure is modified to allow the less stringent criteria of no worsening (rather than improvement) of hepatocellular ballooning, there were greater differences between pioglitazone versus placebo with histological benefit in 81% versus 24% (P < 0.0001), again slightly better than the 48% versus 25% (P = 0.003) reported in PIVENS.8 These differences between PIVENS and our study may be explained by the different populations studied (Hispanics were 15% versus 45%, respectively), duration of treatment (24 versus 6 months), or pioglitazone dose (30 versus 45 mg/day).
We believe that restoration of normal adipose tissue biology with amelioration of adipose tissue insulin resistance9 and an increase in plasma adiponectin levels10 play a key role in TZD-treated NASH patients. Clearly, additional mechanisms are still poorly understood. We share the view of Ratziu et al. on the need of long-term studies targeting special populations. To this end, we have completed recruitment for a long-term study (with a follow-up of up to 3 years) of diet plus pioglitazone (45 mg/day) versus placebo in a predominantly Hispanic population with and without T2DM (UTHSCSA NASH trial; www.clinicaltrials.org #NCT0099468211). We have also started recruitment for another long-term 3-year study of placebo versus vitamin E versus combined vitamin E and pioglitazone (45 mg/day) in a Veteran Administration (VA) population with T2DM and NASH (VA NASH trial; www.clinicaltrials.org #NCT01002547). We believe that both ongoing trials will help close some of our knowledge gaps.
Ratziu et al. clearly outline the shortcomings of TZDs in NASH and identify areas where more research is needed. Their review should encourage additional work and accelerate our understanding of the role of TZDs in the management of patients with NASH.