Author + information
- Received December 26, 2002
- Accepted January 9, 2003
- Published online April 16, 2003.
- W.H.Wilson Tang, MD*,* (, )
- Gary S. Francis, MD, FACC*,
- Byron J. Hoogwerf, MD† and
- James B. Young, MD, FACC*
- ↵*Reprint requests and correspondence:
Dr. W. H. Wilson Tang, Department of Cardiovascular Medicine, Cleveland Clinic Foundation, 9500 Euclid Avenue, Desk F25, Cleveland, Ohio 44195, USA.
Objectives We sought to define the characteristics of fluid retention after thiazolidinedione (TZD) initiation in patients with established heart failure (HF).
Background Fluid retention associated with the use of TZD is commonly attributed to exacerbation of HF, which has led to the proscription of these potentially useful agents in patients with chronic HF.
Methods We examined 111 consecutive diabetic patients with chronic systolic HF who were treated with TZD from January 1999 to June 2001. A retrospective chart review was performed to determine the incidence of fluid retention in this cohort. Physical signs of fluid retention were compared between TZD users and an age- and gender-matched control group of diabetic, non-TZD users with chronic HF who had fluid retention. Baseline clinical and echocardiographic data were compared between TZD users with and without fluid retention.
Results Nineteen TZD users (17.1%) developed fluid retention, which reversed after drug withdrawal and presented predominantly as peripheral and not central edema. Comparing patients in the upper and lower tertiles of weight gain, more female patients and insulin users developed TZD-related fluid retention. However, there were no differences in the baseline New York Heart Association functional class or echocardiographic severity of cardiac dysfunction.
Conclusions Although fluid retention after treatment with TZD in diabetic patients with chronic systolic HF occurs, the mechanism is undefined. Fluid retention related to TZD tends to be peripheral and is usually reversible after drug withdrawal. No direct association between the risk of fluid retention and the baseline degree of severity of HF was observed.
Thiazolidinediones (TZDs) are novel insulin-sensitizing agents that are being widely used as first- and second-line agents in patients with type II diabetes mellitus (1). These agents primarily improve insulin sensitivity and can have potent, beneficial effects on endothelial function even before the complications of overt diabetes ensue (2). In animal studies, TZD has been shown to inhibit cardiac hypertrophy (3,4)and reduce infarct size after coronary artery ligation (5). Thiazolidinedione is likely to have an expanding therapeutic role, particularly in the growing number of cardiac patients with established chronic heart failure (HF) and impaired insulin sensitivity (6).
Fluid retention is known to be an adverse effect of TZD therapy and has anecdotally been attributed to exacerbation of HF (7). In a retrospective, population-based cohort analysis of information from health insurance claims, patients receiving TZD had a 12.4% adjusted risk of developing HF, compared with 8.4% in the control group, after 36 months of follow-up (8). The characteristics of TZD-induced fluid retention and the underlying mechanism remain poorly defined. Data suggest that edema formation is independent of an improvement in glycemic control (9). Although a direct negative inotropic effect has not been excluded, many cases of TZD-induced HF could represent peripheral edema unrelated to worsening cardiac function, similar to the edema seen in patients taking dihydropyridines. Nevertheless, case reports of pulmonary edema occurring in patients taking TZD have led to the proscription of these potentially useful agents for diabetic patients with chronic HF (10–13). Therefore, we reviewed our experience of TZD use in patients with established chronic HF and examined the role of TZD in the appearance of edema and/or worsening HF. Our objectives were: 1) to determine the incidence and characteristics of fluid retention in patients with established chronic systolic HF after TZD therapy; and 2) to determine whether the severity of underlying HF predicted the development of fluid retention after TZD initiation.
The protocol was approved by the Cleveland Clinic Foundation’s Institutional Review Board. Using a computerized search of our hospital’s HF registry compiled from an electronic medical records system (EpiCare, Epic Systems Corp., Madison, Wisconsin), we identified consecutive outpatients with a documented clinical diagnosis of chronic, stable systolic HF (New York Heart Association [NYHA] functional class I to III, left ventricular ejection fraction ≤45%) and a clinical diagnosis of type II diabetes mellitus (according to the latest American Diabetes Association guidelines ) treated in our clinic between January 1999 and June 2001. All patients in the cohort had a standard evaluation for HF that included echocardiography, stress testing, and/or coronary angiography, as needed. From the clinical records, we retrospectively identified all patients from the cohort who had received troglitazone (Rezulin; Parke-Davis, Morris-Plains, New Jersey), pioglitazone (Actos; Takeda Pharmaceuticals, Lincolnshire, Illinois), or rosiglitazone (Avandia; GlaxoSmithKline, Philadelphia, Pennsylvania) at any point during their care (here defined as “TZD users,” comprising the study group). Retrospective chart reviews were conducted to validate the search results, to document the approximate date of TZD initiation (baseline), and to track subsequent changes in weight and other evidence of fluid retention over a course of 12 months. Data extracted from the chart, including standard information on the patients’ weight, list of medications, laboratory test results, symptoms, and physical findings, were collected in a standardized data collection form in an unblinded fashion by a single reviewer. The baseline severity of HF was defined within six months before TZD initiation.
Fluid retention was arbitrarily defined as maximal involuntary weight gain of over 10 pounds from baseline at any point in time within 12 months after TZD initiation, accompanied by documented worsening signs of volume overload as determined by the practitioner (e.g., worsening peripheral edema, jugular venous distention, ascites, or pulmonary edema requiring treatment). Using this clinical definition, we selected all patients with type II diabetes mellitus from the HF registry who had clinical manifestations of fluid retention. We then compared TZD users who experienced fluid retention at any point during the 12-month follow-up period with an age- and gender-matched control group of diabetic, non-TZD users who experienced HF-related fluid retention within a 12-month follow-up period (Fig. 1). Because of the lack of a standard definition of TZD-related fluid retention, univariate analyses using the chi-square test (for dichotomous variables) and unpaired two-tailed ttest (for continuous variables) to compare demographic and echocardiographic parameters between the upper tertile (weight gainers) and lower tertile (non–weight gainers) were performed. All data are expressed as the mean value ± SD. A p value <0.05 is considered statistically significant.
From a registry of 925 diabetic patients with established chronic, stable systolic HF, a total of 115 consecutive TZD users were identified (Fig. 1). Forty-eight patients had been taking troglitazone, and 32 and 35 patients had been taking pioglitazone and rosiglitazone, respectively (including those who had previously used troglitazone). Four patients had no follow-up data or incomplete clinical information. The remaining 111 TZD users were followed up for the next 12 months after TZD initiation (47 using troglitazone, 31 using pioglitazone, and 33 using rosiglitazone). Overall, the average maximal involuntary weight gain within the first 12 months of TZD therapy was 5.9 ± 8.3 pounds. Table 1illustrates the baseline characteristics of all TZD users as well as patients experiencing fluid retention with (study group) or without (control group) TZD use. Only differences in baseline insulin use reached statistical significance between TZD users with and without fluid retention.
Thirty-five patients (31%) discontinued TZD therapy within 12 months of follow-up. Ten patients had no documented reason for TZD discontinuation, nor were they able to identify any drug-related reasons for discontinuation (such as an allergic reaction, hepatotoxicity, HF, or fluid retention). In 20 of the remaining 25 patients who discontinued TZD therapy, the primary reason for discontinuation was persistent edema, despite increased diuretic dosages (including all 19 patients who fulfilled the criteria for TZD-related fluid retention and 1 patient who did not).
In the study cohort, 19 patients (17.1%) had documented fluid retention after TZD initiation. Fluid retention was seen with the use of all three TZD drugs, across all dosages (troglitazone, 8 patients; pioglitazone, 5 patients; rosiglitazone, 5 patients). Nine patients (47%) were taking concomitant insulin therapy. Of the 19 TZD users who experienced fluid retention, 18 (95%) had evidence of worsening peripheral edema by physical examination (Fig. 2). Although measurements of jugular venous pressure were not systematically quantitated or recorded, six patients had documented worsening jugular venous distention. Two (11%) of 19 edematous TZD users also had documented physical signs of pulmonary edema (i.e., worsening rales by auscultation or radiographic evidence of pulmonary congestion) at the time of drug discontinuation. Five of the 19 patients were hospitalized for the management of fluid retention. No patients experienced hepatotoxicity related to TZD use. All patients with fluid retention demonstrated resolution of edema after TZD discontinuation.
A comparison of baseline demographic and clinical parameters of HF severity and glycemic control between the TZD users in the upper and lower tertiles of weight gain is illustrated in Table 2. Except for a trend toward more female patients and insulin users in the weight-gain group, there were no obvious predictors of weight gain or fluid retention after TZD initiation. In particular, baseline left ventricular size and function and baseline NYHA functional class were similar between patients who developed TZD-related fluid retention and patients who did not.
This report demonstrates the tolerability of long-term TZD therapy in a well-characterized population of diabetic patients with established chronic systolic HF. Consistent with previous reports in patients without underlying HF, TZD-related fluid retention typically manifests as peripheral edema that develops within the first few months of drug initiation (15). Our findings, though strictly observational, suggest that the development of TZD-related fluid retention does not correlate with underlying HF severity. Therefore, one cannot reliably predict from baseline assessment of cardiac function or clinical status which patients with chronic HF will develop TZD-related fluid retention. The predominantly peripheral pattern of fluid retention observed in our study further implies that the development of TZD-related fluid retention should not necessarily be equated with worsening HF, unless there is clinical evidence of elevated intracardiac filling pressures.
The exact mechanism of TZD-related fluid retention in patients with established chronic HF is currently unknown. It is possible that transient or persistent neurohormonal activation may cause direct myocardial depression and fluid retention. However, available animal data and single-dosing studies have demonstrated that TZD may actually have favorable effects on the failing and hypertrophied heart (3,16). Recent placebo-controlled, echocardiographic studies also found no significant changes in myocardial structure and function with long-term rosiglitazone (17)or pioglitazone therapy (Takeda Pharmaceuticals North America Inc., unpublished data, 2002) in diabetic patients without underlying HF. Our finding that underlying HF severity did not predict development of TZD-related fluid retention adds credence to the hypothesis that TZDs do not have direct myocardial depressant effects. Other potential mechanisms, such as plasma volume expansion, reduction of sodium renal excretion, suppression of endothelin, and augmentation of vascular permeability, should be explored further in patients with and without underlying HF (18–20).
According to the manufacturers’ information, troglitazone, rosiglitazone, and pioglitazone may cause fluid retention in 2% to 5% of patients receiving monotherapy and 5% to 15% receiving concomitant insulin therapy (21–23). The 17% one-year incidence of TZD-related fluid retention in our HF cohort is probably an overestimation due to a selection bias in identifying patients from a specialized HF clinic. Also, a large proportion (63%) of patients in our cohort who developed TZD-related fluid retention were taking concomitant insulin therapy. The interaction between TZD and insulin in promoting fluid retention has been previously observed. It is possible that TZD potentiates the effects of insulin on sodium and water retention, favoring edema formation. However, it is also noteworthy that a large proportion of TZD users (68%) continued to use TZD after 12 months of therapy, without significant fluid retention, supporting the notion that these drugs can be used safely and effectively in patients with stable chronic HF under careful monitoring.
There is little doubt that TZDs, like non-steroidal anti-inflammatory drugs (NSAIDs), should be avoided in highly symptomatic patients with HF who are already having difficulty maintaining a balanced volume status. It is conceivable that additional fluid retention caused by TZDs (regardless of the mechanism) may alter the already precarious volume status in patients with underlying cardiac dysfunction, thereby justifying the “black box” warning on the package inserts that “these drugs can worsen heart failure.” Recent TZD initiation should still be considered in the differential diagnosis of a diabetic patient with cardiac dysfunction who presents with signs and symptoms of volume overload. However, patients with diabetes mellitus and chronic HF who are stable should not necessarily be deprived of these useful agents as long as they are closely monitored for weight gain, fluid retention, and other signs and symptoms of decompensated HF.
In our study, the management of TZD-related fluid retention consisted of discontinuation of the drug and an increase in diuretic medications. This strategy was highly effective in reversing TZD-related fluid retention. Dose reductions of TZD, instead of complete drug withdrawal, have been reported in the literature (15), but the effectiveness of this strategy needs prospective validation.
Our study is limited by a lack of a rigorous definition of “fluid retention” and the incomplete nature of retrospective data collection. Information on other potential causes of fluid retention (such as dietary and medication non-compliance, including recent NSAID use, and thyroid and renal abnormalities at the time of clinical presentation of fluid retention) was not available in all patients. Therefore, to avoid false inferences, detailed multivariate analyses have not been performed. Furthermore, reliable assessments of jugular venous distention were not available from clinic charts to provide consistent quantitative estimates of right atrial pressures. Biomarkers for elevated filling pressures (e.g., plasma B-type natriuretic peptide levels) were also not available. Despite these limitations, this is the first large clinical series to provide a glimpse of the real-world experience of TZD use in patients with chronic HF. Just as beta-adrenergic blocking therapy in HF can produce profound long-term beneficial effects, despite potential initial decompensation, the TZDs may prove to be a valuable adjuvant or even primary therapy in diabetics with chronic HF (or even in insulin-resistant patients with HF) as long as there is careful follow-up of these patients. Further studies will be important in helping to define the indications for TZD use in these patient populations. We firmly believe that TZD should be studied rather than avoided in patients with HF.
In our cohort, a large majority of patients with chronic HF tolerated TZD well, without significant fluid retention. Although fluid retention after treatment with TZD in diabetic patients with chronic systolic HF occurs, the mechanism is undefined, and the edema is usually reversible after drug withdrawal. A direct association between the risk of fluid retention and the baseline degree of severity of HF has not been established. Our experience, albeit strictly observational, suggests that the pattern of edema formation in most cases is not typical of that seen during direct myocardial depression (i.e., there is more peripheral edema than pulmonary edema). However, direct myocardial depression by TZD has not been excluded, and transient or persistent neurohormonal activation could be causally related to myocardial depression and fluid retention in some patients. Further studies are clearly needed to define the mechanisms of fluid retention associated with TZD use and to determine the safety of cautious use of these novel agents in selected patients with chronic, stable HF before espousing their more widespread use.
- heart failure
- non-steroidal anti-inflammatory drug
- New York Heart Association
- Received December 26, 2002.
- Accepted January 9, 2003.
- American College of Cardiology Foundation
- Yamamoto K.,
- Ohki R.,
- Lee R.T.,
- Ikeda U.,
- Shimada K.
- Sakai S.,
- Miyauchi T.,
- Irukayama-Tomobe Y.,
- Ogata T.,
- Goto K.,
- Yamaguchi I.
- Yue T.,
- Chen J.,
- Bao W.,
- et al.
- Jamali A.H.,
- Witteles R.M.,
- Tang W.H.W.,
- Chu J.W.,
- Reaven G.M.,
- Fowler M.B.
- Delea T.,
- Hagiwara M.,
- Edelsberg J.,
- Oster G.
- King A.B.,
- Armstrong D.
- Benbow A.,
- Stewart M.,
- Yeoman G.
- Thomas M.L.,
- Lloyd S.J.
- St. John Sutton M.,
- Rendell M.,
- Dandona P.,
- et al.
- Donnelly R.,
- Gray S.,
- Idris I.
- ↵Parke-Davis Pharmaceuticals. Rezulin (troglitazone) package insert. Morris Plains, NY: 1997
- Takeda Pharmaceuticals. Actos (pioglitazone) package insert. Lincolnshire, IL: 2000
- GlaxoSmithKline Pharmaceuticals. Avandia (rosiglitazone) package insert. Philadelphia, PA: 2001