1 Introduction

Peroxisome proliferator-activated receptor γ (PPARγ) activators, or thiazolidinediones, are used as insulin sensitizers in the treatment of diabetes. They also have a range of properties that may be beneficial in heart failure. For example, pre-treatment with the PPARγ activator rosiglitazone reduced infarct size in normal rats [1] and reduced ischemic injury in isolated, perfused normal, diabetic [2] and obese Zucker rat hearts [3]. These protective effects were associated with increased GLUT4 glucose transporter protein [2,3] and normalisation of insulin resistance and glucose uptake [3]. As insulin resistance also occurs in chronic heart failure [4,5], and the failing heart undergoes a switch in substrate preference from fatty acids to glucose [6,7], rosiglitazone treatment may improve cardiac metabolism in heart failure. Other studies suggest that PPARγ activators have anti-hypertrophic properties in isolated cardiomyocytes independent of the hypertrophic stimulus [8], and in mice in vivo with pressure-overload hypertrophy due to aortic banding [9]. In addition, rosiglitazone pre-treatment improved ischemia/reperfusion-induced myocardial contractile dysfunction [1], and may [10] or may not [2] be a positive inotrope in isolated perfused rat hearts. Finally, PPARγ activators inhibit lipopolysaccharide-induced expression of TNF-α [11], which has been implicated in the pathophysiology of heart failure.

PPARγ activators are currently contraindicated in heart failure due to an increased incidence of fluid retention and oedema [12,13]. However, information on the effects of PPARγ activators in heart failure models is urgently needed, as clinical trials are already underway in heart failure. Consequently, we undertook this study hypothesising that, on balance, rosiglitazone treatment would be beneficial, or at least benign, in a rat model of chronic heart failure following myocardial infarction (MI). Rosiglitazone-treated rats were compared with untreated rats, with captopril-treated ones, as a gold standard for heart failure therapy in this model, and with captopril and rosiglitazone combined to determine if potential beneficial effects were additive.

2 Methods

This investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institute of Health (NIH publication No. 85-23, revised 1996) and to the Home Office Guidance on the Operation of the Animals (Scientific Procedures) Act, 1986 (HMSO).

2.1 Pilot study

Male Wistar rats were randomised to 3 mg/kg/day rosiglitazone in feed or untreated control (n = 5 per group). After 14 days, LV hemodynamics, and hematocrits were obtained.

2.2 Experimental myocardial infarction

Male Wistar rats (200 g) were obtained from a commercial breeder (Harlan, UK) and kept under controlled conditions for temperature, humidity and light, with rat chow and water available ad libitum. Animals were anesthetized with 1.5–2% isoflurane, intubated and ventilated at 90 breaths per minute, stroke volume 2.5 ml. A thoracotomy was performed in the fourth intercostal space, and a 5/0 Prolene suture tied around the left anterior descending coronary artery a few millimeters from its origin. In sham-operated rats, the ligation suture was not placed in the heart. Intramuscular buprenorphine 0.2 mg/kg was given for pain relief.

In the first 4 h after surgery, approximately 40% of infarcted animals died due to ventricular fibrillation. As drug treatment was not assigned or started until +6 h, these deaths were not included in the mortality figures. Deaths occurring within 36 h were excluded as being surgically related.

2.3 Drug dosing

Surviving rats were randomly assigned to four infarcted groups; untreated (UT, n = 21); rosiglitazone 3 mg/kg/day in chow (R, n = 28); captopril 2 g/l in drinking water (C, n = 15); and captopril+rosiglitazone combined (C+R, n = 17), or were sham operated (n = 12). Post-surgical mortality at 24 h was approximately 40% in all infarcted groups. Only animals surviving 8 weeks with an infarct size of >25% were included in the final analysis. Treatment was started 6 h post-surgery. Food consumption was similar at 31±3 g/day in all groups.

2.4 Echocardiography

Echocardiography was performed at 1 week post-surgery to screen and exclude animals with very small infarcts, and was repeated at 8 weeks to determine the severity of heart failure. Animals were lightly sedated with 3.3–6.6 mg/kg fluanisone, 0.105–0.210 mg/kg fentanyl citrate (Hypnorm, Jansen Pharmaceuticals). Short-axis two-dimensional (2D) images were taken through the left parasternal window using an Agilent Sonos 5500 with 5–12 MHz transducer. Ejection fractions were calculated: where CSA is the cross-sectional area at the papillary muscle level.

2.5 LV hemodynamics

Eight weeks after surgery, animals were anesthetized with 1% isoflurane and the right carotid artery cannulated. When a stable and reproducible pressure reading was obtained, measurements were recorded in the aorta and LV using a pressure transducer (Föhr Medical Instruments, Germany), interfaced to a Powerlab/4SP chart recorder (ADInstruments, UK). Animals were killed by cervical dislocation, and organs blotted and weighed.

2.6 Infarct size determination

Infarct size was determined histologically using picrosirius red as previously described [14].

2.7 Statistical analysis

Data are expressed as mean±S.E. Statistical analysis was performed using one-way analysis of variance (ANOVA) with Bonferroni's post-hoc test, or by unpaired Student's t-test for comparing two groups. Comparisons were made between all ligated groups and sham, and between treated and untreated groups, i.e., a total of seven comparisons. Differences were considered significant at P<0.05.

The Kaplan–Meier method was used to construct survival curves, which were then compared by the Logrank test, with correction for multiple comparisons.

3 Results

3.1 Pilot study

Hemodynamic measurements in normal rats after 14 days treatment with rosiglitazone are given in Table 1. Rosiglitazone treatment significantly increased contractility and relaxation, as measured by dP/dt, without changing heart rate, systolic or diastolic pressures. Thus, rosiglitazone treatment had positive inotropic and lusitropic effects in normal rat hearts. Hematocrits were normal in both groups, indicating absence of hemodilution.

3.2 Post-myocardial infarction study

There were no significant differences in body weight gained or food consumption in any of the groups (Table 2). All infarcted groups had considerable LV hypertrophy such that there was no difference in LV/bw ratio compared with sham, despite loss of ∼40% of LV tissue due to infarction. However, only in captopril treated animals was the LV/bw ratio reduced compared with untreated infarcted rats.

Infarct sizes were similar amongst all treatment groups at approximately 40±2% (Table 2). LV systolic pressure was significantly lower in all infarcted groups compared to sham. Only untreated animals had a significantly elevated LVEDP, with treatment groups not significantly different from sham. Thus, C and C+R almost completely and R partially prevented the chronic increase of LVEDP occurring post-MI. A similar pattern was observed for +dP/dt_max with none of the treatment groups significantly different from sham, while untreated animals had significantly lower values. With regard to relaxation, only rosiglitazone attenuated the decrease in dP/dt_min observed in the untreated, C and C+R groups. All infarcted groups had significantly lower systolic blood pressure compared to sham, while only captopril treatment significantly lowered diastolic blood pressure. Rosiglitazone was the only treatment that significantly elevated ejection fraction post-MI compared with untreated animals. However, unlike captopril and C+R groups that did not have significantly different end-diastolic area from sham, rosiglitazone alone failed to prevent LV dilatation. There were no significant differences in non-infarcted posterior wall thickness or heart rate in any of the groups (Fig. 1).

Over 2 months post-myocardial infarction, rosiglitazone treatment was clearly associated with increased mortality (Fig. 2). There was zero mortality in sham, untreated and captopril groups, but 26% (6/23) of the rosiglitazone and 19% (3/16) of the C+R groups died prematurely (P = 0.03 vs. untreated). Mortality was unrelated to excessive infarct size, as animals with relatively small infarcts were also affected, and there was no evidence of overt edema.

4 Discussion

Our pilot study demonstrates for the first time that rosiglitazone was a positive inotropic agent in normal rats in vivo, and did not cause hemodilution. The mechanism for this inotropic effect has yet to be determined, but has also been described in response to rosiglitazone perfusion of isolated rat hearts [10], which suggests that it is a direct effect.

One major finding of this study is that rosiglitazone treatment did not alter LV structural dilatation or modulate the hypertrophic response. This is in contrast to previous studies suggesting that PPARγ activators have anti-hypertrophic effects in cultured cardiomyocytes [8] and in vivo in aortic constricted mice [9]. The hypertrophic signaling pathway is likely to be similar in pressure-overload and post-myocardial infarction hypertrophy. Although the primary stimulus differs, there is the same signature activation of fetal gene expression, and drugs that are effective in one model tend to be effective in both [15,16]. The lack of an antihypertrophic effect of rosiglitazone in the post-MI rat model was therefore surprising and suggested that the pro-hypertrophic signal after myocardial infarction can override the antihypertrophic PPARγ-dependent pathway.

Rosiglitazone treatment was associated with increased mortality in the post-MI model, both when given alone and in combination with captopril. It is this concordance between rosiglitazone containing groups and the lack of a singular death in the untreated or captopril groups that makes this result so striking. This could not be explained by greater infarct sizes in the animals found dead during the course of the study, and rosiglitazone actually attenuated the deterioration of both systolic and diastolic function in surviving animals.

We have been unable to detect any systematic bias that could explain these results. Of the nine animals that died, they died on eight different dates spread over a 5-month period. Similarly, their surgery was performed on six different dates over a 3-month period.

We have been unable to determine exactly how these animals have died as 24 h ECG monitoring was unavailable to us and the increased mortality was an unexpected result. However, death was evidently sudden as no animal was observed in the process of dying. Animals were often observed to be fit and well on the day of death. The most likely mode of death is either rapid decompensation leading to acute heart failure or cardiac arrhythmia.

Thus, the mechanisms leading to increased mortality by Rosiglitazone treatment remain speculative. A pro-arrhythmic effect is one possible explanation. Previous clinical heart failure trials of positive inotropic agents, such as phosphodiesterase inhibitors [17], showed increased mortality, mostly due to pro-arrhythmic effects.

Direct toxicity of rosiglitazone is unlikely. The dose employed here has been shown to improve insulin sensitivity without ill effects in previous studies in the rat [13], and there was no evidence of hemodilution in our pilot study. One possibility that we cannot rule out completely is an interaction between the anesthetic used during echocardiography and rosiglitazone in the setting of heart failure, which could account for some of the deaths. However, there is no evidence for such an interaction in the literature and the dose of anesthetic is low, producing only mild sedation.

Another possible explanation may be that rosiglitazone treatment interfered with the switch in energy substrate metabolism that occurs during the development of cardiac hypertrophy and failure, in which the heart reverts from fatty acid β-oxidation towards the more oxygen-efficient glycolytic production of ATP [6]. This switch occurs despite increased plasma free fatty acid concentrations caused by higher circulating catecholamine levels. Rosiglitazone treatment is known to lower plasma free fatty acid levels in rats [3] by increasing storage in adipose tissue. Consequently, although lowering of plasma free fatty acid concentrations should be beneficial to the heart [3], it may be that the failing heart becomes energy-starved and that a treatment that lowered the free fatty acids exacerbated this condition.

While our results should certainly be taken into consideration in the context of ongoing or planned clinical heart failure studies using PPARγ-agonists, care should be exercised in extrapolating these results to the clinical setting. This study was in normal, non-diabetic rats, and there may be differences related to species, dose and timing of treatment. For example, it is likely that diabetic patients will already be receiving a PPARγ activator prior to MI, which reperfusion studies suggest may convey benefits on immediate post-ischemic recovery [1–3].

5 Conclusions

In conclusion, we have shown that rosiglitazone is a positive inotrope in normal rat hearts in vivo. Contrary to expectations, rosiglitazone treatment did not modulate LV remodeling after myocardial infarction and was associated with increased mortality. These results suggest that rosiglitazone may not be a useful agent to prevent cardiac remodeling post-myocardial infarction in non-diabetic patients, and might even result in increased mortality in this setting.