Control of hyperglycaemia with insulin or sulphonylureas to the nearly normoglycaemic level did not result in the expected beneficial effect on the high cardiovascular morbidity and mortality in type 2 diabetes in two long-term interventional studies (a-t 2002; 33: 17-8) (1,2). Metformin (e.g. GLUCOPHAGE) is the only oral hypoglycaemic agent for which there are limited data for a reduction in mortality and the rate of myocardial infarction in type 2 diabetes as monotherapy in overweight patients (a-t 1998; No. 10: 88-90) (3). According to the recently published PROactive* study the glitazone pioglitazone (ACTOS; a-t 2000; 31: 103) is also said to have a beneficial effect on macrovascular complications (4). "Antidiabetic drug protects heart and brain", titled a headline in the Ärztezeitung (5).

5,238 patients with type 2 diabetes, an average age of 62 years, and HbA1c over 6.5%** despite diabetes therapy took part in the study sponsored by Lilly and Takeda, the suppliers of pioglitazone. Patients with heart failure NYHA II to IV, ischaemic ulcers, gangrene or rest pain in the legs or hepatic impairment (GPT over 2.5 times the upper limit of normal) were excluded. The exclusion criteria covered also monotherapy with insulin. The participants had a high cardiovascular risk: 47% had a previous history of myocardial infarction, 19% of stroke, 31% of coronary revascularisation and 14% an acute coronary syndrome, and 20% suffered from intermittent claudication. At the start of the study 84% of the patients were taking metformin or sulphonylureas, alone or in combination, and some (30%) in addition to insulin. The others used different combinations and 4% were exclusively on diet. After randomisation, one group took up to 45 mg pioglitazone daily in addition to the usual medications, and the other group took placebo. All medications, including antihypertensive and lipid-lowering drugs, had to be kept optimally adjusted during the course of the study. HbA1c had to be lowered below 6.5% (4).

The HbA1c fell in 2.9 years from a median 7.9% initially by 0.8% with pioglitazone and by 0.3% with placebo. The optimisation of the concomitant therapy remained inadequate: although diabetes with concomitant atherosclerotic diseases is a clear indication for simvastatin (e.g. ZOCOR) since publication of the Heart Protection Study (a-t 2002; 33: 83-4) only 55% of the participants were taking a statin at the end of the PROactive study conducted between May 2001 and January 2005. The information about blood pressure control, which is of prognostic importance in diabetics, and particularly about the antihypertensive therapy is insufficient. However, in the pioglitazone group, the blood pressure fell more than on placebo (systolic by a median 3 mmHg vs. 0 mmHg; p = 0.03) (4).

There is no effect of pioglitazone on the primary endpoint (death, myocardial infarction including silent infarct, stroke, acute coronary syndrome, amputation above the ankle or revascularisation of coronary or leg arteries; 19.7% of patients on pioglitazone vs. 21.7% on placebo; hazard ratio [HR] 0.90; 95% confidence interval [CI] 0.80-1.02) although the event rate is more than 40% higher than initially calculated. The overall mortality was also not affected (6.8% vs. 7.1%; HR 0.96; 95% CI 0.78-1.18). A secondary combined endpoint consisting of death, myocardial infarction and stroke is said to be significantly reduced (11.6% vs. 13.6%; HR 0.84; 95% CI 0.72-0.98; Number needed to treat = 50). On the other hand, the incidence of heart failure (in 10.8% vs. 7.5% of the patients; p < 0.0001; Number needed to harm [NNH] = 30) and hospital admissions because of heart failure (5.7% vs. 4.1%; NNH = 63; p < 0.007) increased significantly (4).

Oedema without heart failure was observed in 22% of patients on pioglitazone compared to 13% on placebo and symptomatic hypoglycaemia occurred in 28% vs. 20%. Body weight increased in the pioglitazone arm by an average of 3.6 kg, while in contrast it fell by 0.4 kg in the placebo arm. With regard to serious adverse events there was no difference overall. A possible risk signal is the higher incidence of malignant bladder tumours on pioglitazone (14 [0.5%] vs. 6 with placebo [0.2%]). Acute hepatotoxicity was not observed with the drug (4).

The conclusion of the authors states: "Pioglitazone reduces the composite of all-cause mortality, nonfatal myocardial infarction, and stroke in patients with type 2 diabetes who have a high risk of macrovascular events." (4) Because the originally defined primary endpoint did not show the desired result, the alleged clinical benefit is simply deduced from a secondary endpoint. This is inadmissable for methodological reasons; when a primary endpoint is nonsignificant, the significant result of a secondary endpoint can be regarded only as explorative and not as proven (8). It obviously did not yet have the significance that is attributed to this endpoint in the Lancet article ("main secondary endpoint") in the study design previously published. It does not even appear among the secondary endpoints listed there (9). Accordingly, the combination of overall mortality, infarction and stroke seems to be an endpoint defined post hoc that is not reliably meaningful.

There are, however also clinical arguments against the conclusion of the authors: the relevant clinical cardiovascular effects of pioglitazone are not adequately represented by either the primary or the secondary combined endpoint. Heart failure, one of the most dangerous risks of the glitazones and at the same time a feared cardiac complication of diabetes, which has a particularly poor prognosis here, is missing. For incomprehensible reasons, it was not recorded as an endpoint but only among the adverse events. Pioglitazone increases the heart failure rate by 3.3%. This increase markedly outweighs the risk reduction of 2% documented in the secondary combined endpoint. It remains unclear to what extent the increased heart failure rate can be attributed to the combination with insulin, which is contraindicated in Germany (because of the increased risk). From subgroup analyses, which are only obtainable from an independent commentary as part of the conference presentation it can be concluded that pioglitazone lacks any beneficial effect in statin users on either the primary or the secondary combined endpoint (10).

The manipulative handling of study data by the authors, apart from the methodologically improper overassessment of a secondary endpoint that was obviously defined afterwards, is also documented by the fact that a second secondary endpoint, predefined in the protocol but obviously negative, i.e. cardiovascular mortality, was not evaluated at all. Only the absolute numbers were communicated, from which rates of 4.9% (pioglitazone) vs. 5.2% are calculated.

	Pioglitazone (ACTOS) has no benefit compared to placebo in the PROactive study as a supplement to existing hypoglycaemic medication in patients with type 2 diabetes and high cardiovascular risk.
	Pioglitazone has no significant effect on the primary endpoint, a composite of death, myocardial infarction, stroke, amputation or revascularisation.
	In an inadmissable manner, the authors proclaim a secondary endpoint of death, myocardial infarction and stroke with a nominally significant reduction in the event rate, which was not mentioned at all in the published design and therefore was apparently defined later to prove the benefit of pioglitazone. Thus there is a suspicion for manipulation of data.
	The absolute risk reduction of 2% in the combined endpoint is more than compensated by the increase in the heart failure rate of more than 3% on pioglitazone.
	Cardiovascular and overall mortality remain unaffected.
	A clinical benefit of pioglitazone is not proven by the PROactive study. Not even the long-term safety of the hypoglycaemic has been adequately demonstrated because of the higher rate of heart failure.

(R = randomised study)

1

UGDP: Diabetes 1982; 31 (Suppl. 5): 1-81

2

UK Prospective Diabetes Study Group: Lancet 1998; 352: 837-53

3

UK Prospective Diabetes Study Group: Lancet 1998; 352: 854-65

4

DORMANDY, J.A. et al.: Lancet 2005; 366: 1279-89

5

Ärztezeitung of 12 Oct. 2005

6

Diabetes Control and Complications Trial Research Group: N. Engl. J. Med. 1993; 329: 977-86

7

Heart Protection Study Collaborative Group: Lancet 2002; 360: 7-22

8

FREEMANTLE, N.: BMJ 2005; 331: 836-8

9

CHARBONNEL, B. et al.: Diabetes Care 2004; 27: 1647-53

10

YKI-JÄRVINEN, H.: European Association for the Study of Diabetes, Athens, Sept. 2005;
http://www.easd-lectures.org/athens/index.php?menu=view&id=101

PROactive = PROspective pioglitAzone Clinical Trial In macroVascular Events

According to the test used in the DCCT (Diabetes Control and Complications Trial) (6) or equivalent values if other test methods were used