To the Editor: We read with interest the article by Hussein and colleagues on their experience with thiazolidinediones (TZDs).1 These agents are only approved by the Pharmaceutical Benefits Scheme as part of dual therapy. We wish to present our experience of adding TZDs to metformin and sulfonylureas — hence, triple therapy — in patients with suboptimally controlled type 2 diabetes mellitus.

The records of 28 patients (15 men, 13 women) with type 2 diabetes, for whom pioglitazone was added to maximal doses of metformin and sulfonylurea because of suboptimal control, were reviewed. Baseline patient characteristics are shown in . Mean follow-up was 9.6 months (range, 3–24 months); the average pioglitazone dose was 31.3 mg. The mean fall in the level of glycohaemoglobin (HbA1c) was 1.26% — 10 patients achieving an HbA1c level of less than 7% at last review. Four patients did not respond to therapy; none withdrew because of side effects. Eight patients whose HbA1c fell less than 0.5% after 3 months continued taking pioglitazone, achieving an average fall in HbA1c of 1.25% after a mean of 12 months follow-up. Mean weight gain was 3.35 kg (– 3.2 kg to 11.6 kg). Mean changes in HbA1c level and weight compared with baseline over 24 months are shown in . There was no correlation between these outcomes, and no baseline characteristic predicted glycaemic response.

Six studies have reported the efficacy of TZDs in triple therapy (), and show a consistent fall in HbA1c level at the expense of weight gain, with a low rate of withdrawals because of adverse effects. Our findings were similar to those of these previous reports in terms of glycaemic response and low rate of side effects. The much higher rate of side effects reported by Hussein et al1 is likely to be the result of the coprescription of TZDs with insulin in 64% of patients in their study. While fluid retention has been reported in up to 5% of patients taking TZDs as monotherapy or in combination with oral hypoglycaemics, 15% of patients using TZDs with insulin may develop significant oedema. Most reports describing precipitation of cardiac failure with TZDs have been in patients using combination therapy with insulin. It would be interesting to know what proportion of the patients who developed peripheral and pulmonary oedema in the study by Hussein et al1 were also receiving insulin. One prospective randomised trial comparing the addition of pioglitazone and bedtime insulin to maximal metformin and sulfonylurea found similar efficacy in improving glucose control, but less hypoglycaemia and improved high density lipoprotein cholesterol levels with pioglitazone.3 A study of the long-term efficacy of triple therapy found 26 of 35 patients (74%) had good control after a mean follow-up of 37 months, their HbA1c level having fallen from 8.7% to 6.9%.8

In conclusion, the experience of our unit and the published literature is that TZDs are efficacious in improving suboptimal diabetic control in patients on maximal doses of metformin and sulfonylurea. Eight individuals in our group had a significant improvement in control subsequent to minimal response after the initial 3 months of treatment, suggesting a longer trial of TZDs should be employed before classifying patients as non-responders. It is to be hoped that the regulatory authorities will allow the use of TZDs in triple therapy.

2 Changes in glycohaemoglobin (HbA1c) level and weight compared with baseline values

To the Editor: It was with interest that we read the recent article on real-life experience with thiazolidinediones by Hussein et al.1 The authors noted the absence of severe liver toxicity with the newer agents, rosiglitazone and pioglitazone, in contrast to troglitazone, which was withdrawn because of cases of hepatic failure.2 However, the sample was too small to conclude that these thiazolidinediones (TZDs) carry no hepatic risk, and they endorsed the current Pharmaceutical Benefits Scheme recommendations that liver function tests (LFTs) be monitored every 2 months. We have collected similar clinic data showing that the development of significant abnormalities in results of LFTs with TZD therapy is uncommon, and in fact, there are often improvements in LFT findings.

We reviewed the files of 166 patients with type 2 diabetes treated with TZDs between 1 August 2000 and 30 November 2002, with the aim of assessing their long-term effect on LFT results. Therapy was discontinued within 3 months in 26 patients. The reasons were non-compliance (7), therapy ineffective (8), weight gain (5), dyspnoea (1), peripheral oedema (1), malaise (1), dizziness (1), angio-oedema (1), and pre-existing LFT abnormality (1). We analysed data on the remaining 140 patients (see Box) treated for a mean of 188 ± 4 days with either pioglitazone (109 patients) or rosiglitazone (31 patients).

All LFT results improved significantly (Box). At baseline, 90 patients had abnormal findings on LFTs. These findings normalised in 43 of these patients (including one with steatohepatitis proven on biopsy); improved in 29 patients; and were unchanged in nine patients. LFT findings deteriorated in nine patients, leading to cessation of therapy in two. Most patients with normal LFT results at baseline experienced improvements of these parameters within the normal range. Three patients developed new abnormalities in their LFT findings, and therapy was stopped in one patient, leading to resolution of LFT abnormalitites. Changes in glycohaemoglobin (HbA1c) levels correlated positively with changes in activity of alkaline phosphatase (correlation coefficient [r], 0.33; P < 0.01), aspartate aminotransferase (r, 0.27; P < 0.01) and alanine aminotransferase (r, 0.29; P < 0.01).

Our findings support those of Hussein et al, that TZD therapy is usually stopped for reasons other than hepatic dysfunction. In contradistinction to early concerns about their hepatic safety, the improvements in LFT findings seen in our patients suggest that TZDs may even benefit hepatic function. In patients with diabetes, abnormal findings on LFTs are often attributed to fatty liver, which predisposes to steatohepatitis. TZDs may well alleviate or prevent steatohepatitis,3 thereby providing benefits beyond that of improved glycaemic control, and mild abnormalities in LFTs should not discourage their use in patients with diabetes.