The effects of discontinuing pioglitazone in patients with nonalcoholic steatohepatitis


  • Potential conflict of interest: Nothing to report.


A pilot study of a 48-week course of pioglitazone demonstrated significant improvements in the biochemical and histological features of nonalcoholic steatohepatitis (NASH). The aim of the study was to assess the effects of stopping pioglitazone. Twenty-one patients with NASH were treated with pioglitazone (30 mg/day) for 48 weeks and underwent baseline and end-of-treatment evaluation including liver biopsy. Thirteen patients were followed for at least 48 weeks after stopping therapy and 9 underwent repeat liver biopsy. Statistical comparisons were made to evaluate whether discontinuation of pioglitazone resulted in a reversal of improvements seen on therapy. Stopping pioglitazone was associated with subsequent elevation in serum alanine aminotransferase levels (from 34 ± 13 to 70 ± 39 IU/l), decrease in adiponectin (from 9.7 ± 9.1 to 5.1 ± 4.5 μg/ml), worsening insulin sensitivity (HOMA Index: from 2.9 ± 1.8 to 5.5 ± 5.4), and increase in total hepatic fat (from 30% ± 32% to 71% ± 33%) despite no change in average body weight compared to the end of treatment. Repeat liver biopsy in 9 patients revealed significant worsening of parenchymal inflammation (from 1.2 ± 0.7 to 2.9 ± 1.1) and steatosis (from 0.9 ± 0.6 to 2.1 ± 1.3) but no change in fibrosis (from 1.1 ± 1.2 to 1.2 ± 1.3). NASH was again present on liver biopsy in 7 patients. Conclusion: These findings suggest that long-term therapy with pioglitazone may be necessary to maintain improvements in disease activity in patients with NASH, although weight gain during treatment may ultimately limit its beneficial effects. (HEPATOLOGY 2007.)

Nonalcoholic steatohepatitis (NASH) is an inflammatory liver disease associated with progressive liver injury which can eventually lead to cirrhosis, end-stage liver disease, and hepatocellular carcinoma.1 The etiology of NASH is thought to be related to insulin resistance and the metabolic syndrome.2 NASH is closely associated with obesity and thus may represent the most prevalent chronic liver disease in the United States.3, 4

Several strategies aimed at improving insulin resistance have been employed in clinical trials for NASH, one being controlled weight loss using diet and exercise. This approach has had limited success in those patients who are able to lose and maintain weight loss in the long term. The improvement in histology with weight loss is primarily in hepatic steatosis rather than inflammation and fibrosis.5–7 Thus far, pharmacological therapy aimed at weight loss has been disappointing, and histological data on the effects of weight-loss medications are lacking.8 A more sustained option is bariatric surgery which has been shown to promote marked weight loss, decrease insulin resistance, and improve liver histology in patients with NASH.9 However, for most patients with NASH, an invasive surgical procedure with major metabolic effects is not an appropriate option.

Drug therapy primarily aimed at improving insulin resistance presents a possible option for patients with NASH who are unable to lose weight or maintain weight loss or who are not suitable candidates for bariatric surgery. The thiazolidinediones, pioglitazone and rosiglitazone, are agonists to the nuclear peroxisome proliferator activated receptor-gamma (PPARγ) and have been shown to improve insulin sensitivity in patients with type 2 diabetes. Pilot studies of these two agents in both diabetic and non-diabetic patients with biopsy-proven NASH have demonstrated a decrease in serum aminotransferase levels, an improvement in insulin resistance, a decrease in liver fat by magnetic resonance imaging (MRI) and marked improvements in the full spectrum of histological findings characteristic for NASH.10–12 Therapy with either agent was associated with significant weight gain.

Unlike maintained weight loss, short-term therapy with these agents is not expected to result in a durable decrease in insulin resistance when therapy is discontinued. However, it is unclear if the beneficial effects of these agents are based primarily on decreasing insulin resistance, because the PPARγ agonists have anti-inflammatory, anti-fibrotic as well as other metabolic activities.13, 14

The aim of the current study was to characterize the biochemical, metabolic, and histological changes that occur after pioglitazone is discontinued in a cohort of patients who demonstrated significant improvement in these parameters during therapy.


ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; DEXA, dual-energy x-ray absorptiometry; IRB, Institutional Review Board; MRI, magnetic resonance imaging; NASH, nonalcoholic steatohepatitis; PPARγ, peroxisome proliferator activated receptor-gamma.

Patients and Methods

The patient cohort for this study was derived from a group of 21 patients with well-characterized and biopsy-proven NASH who were treated in an open-label trial with pioglitazone (Actos, Takeda Pharmaceuticals North America Inc., Lincolnshire, IL) in a dose of 30 mg per day orally for 48 weeks. All details of this study were approved by the NIDDK Institutional Review Board (IRB), and patients gave written informed consent. An analysis of the anthropometric, biochemical, and histological changes seen in the first 18 of these patients has been published.11

After patients completed the 48-weeks of pioglitazone therapy, they were followed at 3-monthly intervals in the outpatient clinic as part of the initial treatment protocol. At each visit, routine clinical evaluation and fasting metabolic and hepatic tests were obtained. A separate serum sample was stored at −70°C for specialized testing including cytokine and adiponectin levels (B-Bridge International, Sunnyvale, CA).

Forty-eight weeks after pioglitazone was discontinued all patients were offered the opportunity to enroll in a separate long-term study of pioglitazone using lower initial doses (15 rather than 30 mg/day). This protocol was also reviewed and approved by the NIDDK IRB. Patients who gave written informed consented to participate in the long-term pioglitazone trial underwent a repeat liver biopsy, and metabolic profiling including oral glucose tolerance testing, whole-body dual-energy x-ray absorptiometry (DEXA) for assessment of percent body fat, and hepatic MRI for assessment of liver volume and percent fat.11

Liver biopsies were read and scored using a scale of 0 to 4 for parenchymal inflammation, ballooning cellular injury, steatosis, and fibrosis.11 The pathologist was blinded to both patient identity and sequence of biopsies. A summary score (NASH Activity Index) consisting of parenchymal inflammation, ballooning cellular injury, and steatosis (0-12) was used as the histological definition of NASH, a score of ≥4 with at least 1 point from each of the 3 parameters being required for the diagnosis.

Statistical comparisons were made using the Wilcoxon ranked test (α < 0.05) to compare pre-pioglitazone to end of pioglitazone therapy parameters. Similar analysis was used to compare end of pioglitazone therapy to end of follow-up parameters using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA).


Patient Cohort.

Of the 21 patients enrolled, 20 completed 48 weeks of therapy. One patient discontinued treatment at 3 weeks due to dizziness but underwent the 48-week evaluation including liver biopsy. Of the 20 patients who completed therapy, 13 completed a further 48 weeks of follow-up after stopping therapy. These 13 patients included 7 men, 11 non-Hispanic whites, 1 Hispanic white, and 1 Asian. The mean age of this follow-up cohort was 41.5 years (range: 27-63 years). Nine of the 13 patients consented to enroll in a study of long term pioglitazone therapy and underwent a repeat liver biopsy approximately 48 weeks after stopping the initial course of therapy (Fig. 1). The biochemical data for the cohort of 13 patients who were followed for 48 weeks are shown in Table 1. Of the 8 patients who did not complete 48 weeks of follow-up, 3 completed 24 weeks, 4 completed 12 weeks and 1 patient had no followup. Three of the 8 patients moved out of the region, 1 was started on pioglitazone by her primary care physician and 4 decided not to continue in further studies. The baseline and end-of-pioglitazone treatment biochemical, metabolic and histological features of the 13 patients with and the 8 without 48 weeks of follow up were similar (data not shown).

Figure 1.

Derivation of the patient cohort included in the study.

Table 1. Comparison of Selected Biochemical and Insulin Resistance Parameters Before and After 48 Weeks of Treatment with Pioglitazone and Again 48 Weeks After Stopping Pioglitazone
N = 13(1) Baselinea(2) 48 Weeks on Pioglitazonea(3) 48 Weeks off pioglitazoneaPb 1 vs. 2Pb 2 vs. 3
  • a

    Mean ± S.D,

  • b

    Wilcoxon Ranked Sum Test

ALT (IU/l)75.7 ± 34.734.0 ± 12.769.5 ± 38.7<0.001<0.001
AST (IU/l)47.5 ± 28.129.0 ± 7.748.3 ± 25.2<0.001<0.01
Free Fatty Acids (Eq/L)624.3 ± 205.8545.8 ± 202.4548.8 ±
HOMA4.9 ± 4.52.9 ± 1.85.5 ± 5.4<0.01<0.01
BMI (kg/m2)31.4 ± 6.232.7 ± 6.932.7 ±
Adiposity by DEXA (%)34.8 ± 7.738.2 ± 8.036.6 ±
Liver Volume (cm3)2435 ± 6202043 ± 3022222 ± 3470.070.15
Percent Liver Fat54.8% ± 26.6%30.1% ± 32.2%71.2% ± 33.2%0.070.21
Adiponectin-Males (μg/ml)2.2±0.54.0±1.61.7±1.3<0.05<0.05
Adiponectin-Females (ug/ml)4.1±1.416.4±9.87.8±4.3<0.05<0.05

Changes in Serum Aminotransferase Levels.

In the follow up cohort of 13 patients, serum alanine and aspartate aminotransferase (ALT and AST) values fell into the normal range during the first 24 weeks of pioglitazone therapy and remained normal during the rest of the 48-week course (Fig. 2). The decreases were significant for both ALT and AST levels. Stopping pioglitazone at 48 weeks was followed by increases in ALT levels within the next 12 weeks with further rises thereafter. Changes in ALT and AST values at 48 weeks off of pioglitazone were statistically significant in comparison to end-of-treatment values but were similar to pre-treatment levels (Table 1).

Figure 2.

Changes in serum aminotransferase levels during and after a 48-week course of pioglitazone.

Changes in Insulin Sensitivity.

Pioglitazone therapy was associated with significant improvement in insulin sensitivity as assessed by the HOMA index which is calculated based upon fasting glucose and insulin levels.11 Improvement in insulin sensitivity occurred within 12 weeks of starting pioglitazone and were maintained thereafter (Fig. 3). Stopping pioglitazone was followed within 12 weeks by a worsening of insulin sensitivity, and by 48 weeks, HOMA index values were significantly higher than end-of-treatment and slightly higher than pre-treatment values (Table 1). Free fatty acid levels decreased minimally during pioglitazone therapy and showed no consistent or significant changes during or after treatment.

Figure 3.

Changes in HOMA index and weight during and after a 48-week course of pioglitazone.

Despite its effect on insulin sensitivity, pioglitazone therapy was associated with weight gain and an increase in body mass index (BMI = weight in kilograms/height in meters2). Stopping pioglitazone did not lead to subsequent weight loss but was also associated with little further weight gain (Table 1 and Fig. 3). Among the 13 patients who underwent repeat testing, total adiposity as measured by whole body DEXA scan increased from 34.8% (± 7.7%) to 38.2% (± 8.0 %) (P = 0.02) during pioglitazone therapy and remained stable during the 48 weeks after stopping (36.6% ± 8.6 %) (P = NS).

Serum adiponectin levels have been shown to be low in patients with NASH and insulin resistance and to improve during pioglitazone therapy.15 Among the 13 patients with adequate follow up, adiponectin levels improved during pioglitazone therapy and decreased during follow up, returning to pre-treatment low values (Fig. 4, Table 1).

Figure 4.

Changes in serum adiponectin during and after a 48-week course of pioglitazone.

Changes in Liver Volume and Percent Fat.

Among the 13 patients undergoing repeat MRI testing, estimated hepatic volume decreased on pioglitazone from 2435 to 2043 cm3 (P = 0.07) but then increased to 2222 cm3 48 weeks after stopping medication (P = 0.15). Similarly, average percent fat in the liver as assessed by MRI decreased on pioglitazone therapy from 55% to 30% (P = 0.07) but rose to above baseline thereafter to 71% (P = 0.21).

Changes in Liver Histology.

Pioglitazone therapy was associated with significant improvements in ballooning cell injury, parenchymal inflammation, steatosis (elements of the NASH Activity Index), and fibrosis. Repeat liver biopsy in 9 patients, 48 weeks after stopping pioglitazone, revealed significant worsening in parenchymal inflammation and steatosis with a trend in worsening of ballooning cell injury (Table 2 and Supplementary Table 1). The overall NASH Activity Index score worsened significantly. In contrast, there was no associated worsening of hepatic fibrosis which remained improved (mean and standard deviation = 1.2 ± 1.3) compared to pre-treatment values (2.1 ± 1.2) (P = 0.03). All 9 patients fit the histological criteria for NASH at the start of therapy, but only one still fulfilled the histological criteria for NASH at the end of 48 weeks of pioglitazone. At the end of the 48-week follow up off pioglitazone, 7 of the 9 patients once again fulfilled the strict histological criteria for NASH.

Table 2. Comparison of Different Histological Parameters Before and After 48 Weeks of Treatment with Pioglitazone and Again 48 Weeks After Stopping Pioglitazone
N = 9(1) Baselinea(2) 48 Weeks on Pioglitazonea(3) 48 Weeks off PioglitazoneaPb 1 vs. 2Pb 2 vs. 3
  • a

    Mean ± S.D,

  • b

    Wilcoxon Ranked Sum Test and

  • c

    Fishers Exact Test, NASH Activity Index Score of ≥ 4.

Parencyhmal Inflammation3.0 ± 1.11.2 ± 0.72.9 ± 1.4<0.010.02
Steatosis2.6 ± 1.00.9 ± 0.62.1 ±
Cell injury1.6 ± 0.70.4 ± 0.71.2 ±
NASH Activity Index7.1 ± 2.42.6 ± 0.96.2 ±
NASH fibrosis2.1 ± 1.21.1 ± 1.21.2 ±
Diagnostic criteria for NASH9 (100%)1 (11%)7 (78%)<0.001c0.02c
Interval between biopsies (wks) 50.6 ± 4.450.2 ± 6.3  


Forty-eight weeks of therapy with pioglitazone resulted in significant improvement in the biochemical and histological features of NASH including hepatic fibrosis.11 Discontinuation of pioglitazone resulted in an early reversion to the NASH phenotype with elevated serum aminotransferase levels, worsening insulin resistance and return of high levels of hepatic fat. Importantly, there was also a return of hepatocellular ballooning injury, inflammation and steatosis once pioglitazone was stopped in the majority of patients. In this subgroup of patients, the improvement in liver fibrosis that was seen during therapy was unchanged despite the discontinuation of therapy.

The results of this study suggest that long-term therapy with pioglitazone may be necessary to decrease hepatic injury in patients with NASH. Long-term therapy is also likely to be necessary for continuing improvement by other insulin-sensitizing agents, such as metformin.16, 17 Long-term treatment with either agent is a possibility as both are well tolerated and safe; minor common side effects being peripheral edema with pioglitazone, and diarrhea and abdominal upset with metformin.18 However, the ability to maintain a long-term response will need to be shown. Indeed, in the case of metformin improvements in serum aminotransferase levels and hepatic histology may not be sustained for even 48 weeks despite continuation of the medication.19, 20

While a beneficial response to the thiazolidinediones has been shown in patients with NASH treated for up to 48 weeks, long-term maintenance of a response cannot be assumed. Indeed, the major side effect of pioglitazone has been weight gain, which may ultimately reverse the beneficial effects of these agents against NASH. In the initial study of pioglitazone, the average weight gain during 48 weeks of treatment was 3.5 kilograms, or approximately 4% of initial body weight.11 Furthermore, the weight gain was due to increase in adiposity (increase in total body fat) and was progressive with little evidence that the rate of increase was decreasing over time (Fig. 3). A similar degree of weight gain was found in a recently published, randomized controlled trial of pioglitazone in patients with diabetes and NASH in which the average weight gain in patients receiving a 6 month-course of pioglitazone was 2.5 kilograms, most of which was an increase in fat mass.12 In contrast, placebo recipients experienced a mild weight reduction (0.5 kilograms). In the current follow-up study of 13 patients, rapid weight gain stopped once pioglitazone was discontinued, but no patient returned to their pre-treatment weight. Pioglitazone is thought to act by inducing maturation and proliferation of adipocytes, which, once present, are unlikely to regress spontaneously.21 However, it appears that with aggressive dietary restriction and behavior modification weight loss can be achieved after a course of thiazolinediones.22

Although not statistically significant, total adiposity as assessed by DEXA, the amount of hepatic steatosis as assessed by MRI and the degree of insulin resistance as assessed by HOMA were greater 48 weeks after stopping pioglitazone than before treatment was initiated. Thus, there appeared to be a redistribution of fat once pioglitazone was stopped, perhaps from peripheral (subcutaneous) to central (liver and intra-abdominal) sites. These findings suggest that the weight gain from thiazolidinedione therapy may predispose ultimately to worsening of obesity and its metabolic effects including insulin resistance, hepatic steatosis, and NASH once the medication is stopped, thus ultimately limiting any temporary benefit.

These conclusions, however, are based upon a small cohort who were studied in an open-label and non-controlled fashion. The long-term effects of the thiazolidinediones on NASH and weight gain deserve further careful prospective analysis in larger groups of patients with adequate follow-up both during and after therapy. The thiazolinediones have recently been reported to decrease the incidence of type 2 diabetes in high-risk patients with impaired glucose tolerance.23 Pioglitazone is currently under prospective investigation in a large, randomized controlled trial in adult patients with NASH being conducted by the NASH Clinical Research Network.24 Nevertheless, until results from larger cohorts of patients treated for a more prolonged period are available, the thiazolidinediones should be used with caution in patients with NASH.


The authors thank the patients who participated in these clinical studies as well as Yoon Park, R.N. of the Liver Diseases Branch and the dedicated research nurses of Clinical Center, NIH.