Thiazolidinediones (glitazones) (BNF, section 220.127.116.11)
No drugs in diabetes have caused more controversy or been more scrutinized than the glitazones. Starting in 2007, the glitazone controversy, especially concerning rosiglitazone, caused a major rethink not only of the fundamental reasons for treating hyperglycemia in type 2 diabetes, culminating in the trials reporting in 2008 and 2009, but at least in the USA major changes in approval of antihyperglycemic agents, which are now required to demonstrate cardiovascular safety before licensing. This is not the place to recount the details of the controversy, but a meta-analysis of studies with rosiglitazone published in 2007 had suggested that both myocardial infarction and cardiovascular deaths were increased with this agent. Further studies and analyses confirmed that cardiovascular deaths were not increased, and the recent major glycemia trials confirmed overall the cardiovascular safety of rosiglitazone though other studies, for example the aborted RECORD study, are less persuasive. The PROactive study using pioglitazone (2005) had shown that major cardiovascular events were significantly reduced with this agent, but the primary outcome, a composite of cardiovascular events and arterial interventions, was unchanged. Perhaps underlying all this was the reversal of the widespread expectation that because of the mode of action of the glitazones, cardiovascular events would be consistently reduced by these agents. More generally, short-term studies in humans or experimental animals that demonstrate improvements in markers of cardiovascular risk cannot be translated into clinical benefit until definitively proven in large-scale RCTs. Redefining the position of the glitazones in the management of type 2 diabetes has been difficult because of the long-term resonances of these studies and expectations, and also history.
The prototype glitazone, troglitazone, was introduced very briefly in the UK during 1997, but withdrawn because of rare but serious hepatic side-effects. It was withdrawn in the USA in 2000. Both the successor compounds, rosiglitazone and pioglitazone, have no adverse hepatic effects. On the contrary, glitazones consistently improve mildly abnormal liver function test results (see below). Although the non-response rate is significant overall they are no less potent antihyperglycemic agents than any other agents. Nevertheless, after detailed re-evaluation, rosiglitazone was withdrawn in Europe in late 2010 because of continued anxiety over increased myocardial infarction rates, and although it remains marketed in the USA, new use is restricted to those where all other medication options, including pioglitazone, have been exhausted.
Glitazones are agonists of the peroxisome proliferator-activated receptor (PPAR)-γ, acting at nuclear receptors to stimulate various insulin-sensitive genes. They have multiple actions in insulin-sensitive tissues (muscle, liver and adipose tissue), for example increasing glucose uptake, decreasing gluconeogenesis and glycogenolysis, and increasing fatty acid uptake, lipogenesis and differentiation of adipocytes. Like metformin, they require insulin for their glucose-lowering effects, and are antihyperglycemic rather than hypoglycemic, so used in monotherapy carry a low risk of hypoglycemia. Direct and indirect markers of insulin sensitivity are consistently improved, but despite good experimental evidence for improved J3-cell function, in the ADOPT study after 5 years there was no significant difference in J3-cell function compared with glibenclamide or metformin. Although those with phenotypic features of insulin resistance might be thought to benefit from a glitazones, clinical response cannot be predicted clinically or on the basis of any simple laboratory tests. A carefully observed therapeutic trials in individual patients is therefore important, as with any agent. Onset of glucose lowering is slow and does not maximize until 4–5 months, so pioglitazone should not be used in symptomatically hyperglycemic patients.
A large portfolio of consistently demonstrated biochemical benefits (including significant reductions in high-sensitivity CRP, white count, various other inflammatory markers, including the plaque-disrupting matrix metalloproteinases, and microalbuminuria) has not translated in large-scale RCTs to a reduction in cardiovascular events. However, smaller studies in well-defined groups with advanced vascular disease are more encouraging.
- A post-hoc analysis of PROactive found that the risk of recurrent stroke was reduced by about 50%, although there was no benefit in primary prevention of stroke.
- Coronary stent restenosis in both diabetic and non-diabetic subjects is reduced with both agents over the medium term in small studies (a bigger study is in progress). However, a large-scale intravascular ultrasound study (APPROACH) found that rosiglitazone had no clear advantage over glipizide in reducing the volume of coronary atheroma over 18 months.
- CIMT is rapidly reduced, independent of glycemic benefits, again with both agents, though the CHICAGO study with pioglitazone hinted that this effect attenuated with time.
- BARI 2D (2009): cardiovascular outcomes improved in coronary bypass patients treated with metformin/rosiglitazone compared with insulin.
Currently, however, these potentially intriguing uses are largely contraindicated.
Pioglitazone and insulin
A licensed combination of clinical importance. In PROactive, addition of pioglitazone reduced the likelihood of requiring permanent insulin treatment by 50% over 3 years. There was also a consistent 0.5% (5 mmol/mol) HbA1c benefit in the pioglitazone-treated group, including the 30% who were already using insulin; 9% of patients were able to discontinue insulin. The need for multiple insulin injections was reduced as was the number of coadministered oral agents. Not unexpectedly, those in poor glycemic control and needing higher doses of insulin benefited most from the addition of pioglitazone . While reducing insulin doses or even discontinuing insulin are not considered by academics to be important outcomes, patients are likely to hold a different view.
This is an important consideration. The glitazones consistently show more stable glycemic control than sulfonylureas and, to a lesser degree, metformin, up to 5 years. In the PROactive study, triple therapy (sulphonylurea, metformin and pioglitazone) was well tolerated and delivered stable glycemic control over 3 years.
Many small studies have hinted at improvement in liver histology with glitazone treatment, but in a large 2-year study, while the antioxidant vitamin E (800 IU daily) improved overall histology, pioglitazone 30 mg daily did not. Importantly, neither agent seemed to improve fibrosis scores, though transaminases, as frequently seen in clinical practice, fell with both. Pioglitazone does not seem to consistently improve all histological outcomes in this important condition, and it is unlikely to be licensed for it .
- Weight gain: as with other agents, weight gain, largely adipose, can be rapid in the first year but may continue up to 5 years. In ADOPT, the mean weight difference between the rosiglitazone and metformin groups at the end of the study was nearly 7 kg. Although weight gain may be peripheral rather than abdominal, waist and hip circumferences both increase. The weight gain is broadly related to the glycemic benefit. Ankle oedema is common and occasionally troublesome.
- Exacerbation of heart failure: glitazones are contraindicated in any degree of heart failure, and adjudicated cases of heart failure and hospitalization for heart failure, though not mortality, are increased. Normochromic anemia with a decrease in hemoglobin (Hb) of about 1 g/dL, sometimes greater, may be seen. It may be due to a mild bone marrow effect. Since anemia occurs early in diabetic nephropathy, and significant renal impairment (eGFR > 30 mL/min) is not a contraindication to glitazone treatment, the degree of anemia may be significant in some patients.
- Fractures: PPAR-γ inhibits bone formation by diverting mesenchymal stem cells to the adipocyte lineage, and may also stimulate osteoclasts. Distal limb fractures are increased in premenopausal and postmenopausal women taking glitazones compared with those taking sulfonylureas or metformin, and men may also be at increased risk. Typical osteoporotic spine and femoral neck fractures are not increased.
- Macular oedema: there have been some case reports associating thiazolidinediones with macular oedema. Cohort studies have found a slightly increased risk with both drugs, but there was no association in a cross-sectional analysis of the ACCORD study . Nevertheless, be alert to the possibility, especially in patients with known retinopathy.