The Effects of Thiazolidinedione Therapy on NT-proBNP Levels in Patients with Type 2 Diabetes

This Article


Article Information:

Group: 2008
Subgroup: Volume 6, Issue 2, Spring
Date: April 2008
Type: Original Article
Start Page: 70
End Page: 77


  • HD Tildesley
  • St. Paul’s Hospital,bUniversity of British Columbia,and,Center for Health Evaluationand Outcome Sciences, Vancouver BC, Canada
  • CM Aydin
  • St. Paul’s Hospital, Vancouver BC, Canada
  • E Billington
  • St. Paul’s Hospital, Vancouver BC, Canada
  • AP Ignaszewski
  • St. Paul’s Hospital,bUniversity of British Columbia,and,Center for Health Evaluationand Outcome Sciences, Vancouver BC, Canada
  • JA Strelzow
  • St. Paul’s Hospital, Vancouver BC, Canada
  • SJ Wise
  • St. Paul’s Hospital, Vancouver BC, Canada
  • E Yu
  • St. Paul’s Hospital,bUniversity of British Columbia,and,Center for Health Evaluationand Outcome Sciences, Vancouver BC, Canada
  • GP Bondy
  • St. Paul’s Hospital,bUniversity of British Columbia,and,Center for Health Evaluationand Outcome Sciences, Vancouver BC, Canada


      Affiliation: St. Paul’s Hospital,bUniversity of British Columbia,and,Center for Health Evaluationand Outcome Sciences
      City, Province: Vancouver BC,
      Country: Canada


We sought to determine whether thiazolidinedione (TZD) therapy affects levels of serum N-terminal pro-brain natriuretic peptide (NT-proBNP) in patients with type 2 diabetes. Materials and Methods: This study population consisted of 76 patients with type 2 diabetes and no history of heart failure. Subjects had NT-proBNP levels determined prior to initiating TZD therapy, and after 3 months of treatment. We compared within-person changes in NT-proBNP over the 3 month duration. We determined if the magnitude of change in NT-proBNP over the treatment period was correlated with baseline parameters or nature/dose of TZD medication. Results: The subjects were 42% female and 58% male, and the mean age and duration was 59.8±11.8 years old and 11.4±8.3 years respectively. The baseline mean A1C and BMI was 8.7±1.1% and 30.9±8.7 kg/m2 respectively. We found that NT-proBNP levels did not vary significantly between baseline (mean±SD: 143.8±203.9 pg/mL) and 3 month follow-up (150.6±186.2 pg/mL). Conversely, A1C levels declined significantly (p<0.0001) and BMI increased significantly (p< 0.05). Conclusion: Adding TZD therapy to patients with type 2 diabetes and no history of heart failure does not have a significant effect on NT-proBNP levels.

Keywords: Thiazolidinedione;NT-proBNP;Type 2 diabetes;Congestive heart failure

Manuscript Body:


Thiazolidinediones (TZDs) are a unique class of oral anti-hyperglycemic agents used in the treatment of type 2 diabetes. The efficacy of the two second-generation TZDs, pioglitazone and rosiglitazone, has been well established.1-4 However, side effects of TZD therapy include weight gain, fluid retention (6% to 7% of patients) and edema (2% to 5% of patients).2-7 These are particularly conce-rning for patients with a history of cardiac dysfunction, as fluid retention and edema can serve to exacerbate heart failure.6,7Given the highly significant epidemiolo-gical relationship between diabetes and heart failure,8,9 several studies have explored the association between TZD therapy and cardio-vascular endpoints.10-12 Despite Delea et al’s suggestion that TZD therapy might increase the risk of heart failure,10 a review and cons-ensus statement released by the American Diabetes Association in 2004 concluded that the cardiovascular risk associated with TZD therapy is, in fact, quite low.13 As well, in 2004, results of a year-long randomized controlled trial suggested that there is no difference in cardiovascular outcomes for patients taking pioglitazone and patients on a non-TZD therapy.12More recently, the PROactive Study dem-onstrated an increased incidence of serious heart failure with pioglitazone (5.7% absolute risk) versus placebo (4.1% absolute risk) in patients with type 2 diabetes and preexisting cardiovascular disease. However, increased mortality or morbidity in treatment versus placebo patients with serious heart failure was not found.14Nissen and Wolski’s meta analysis of 42 clinical trials determined that rosiglitazone was associated with approximately 43% greater risk for myocardial infarction (MI) and approximately 64% greater risk for cardiovascular death than placebo or other anti-diabetic therapies.15 However this data should be considered carefully: the MI and cardiovascular death absolute risk differences between treatment and control groups are 0% and 0.1% respectively. As well, the method-ology that was used required the exclusion of trials with zero events in the treatment and placebo groups and alternative meta-analysis approaches that use continuity corrections show statistically nonsignificant low odds ratios;16 thus it can be concluded that it is uncertain whether risk for MI or cardiovasc-ular death is increased or decreased for patie-nts undergoing rosiglitazone therapy.Interestingly, it has been suggested that TZDs actually have macrovascular benefits for patients both with and without pre-exis-ting cardiovascular disease. Specifically, these medications have been found to reduce blood pressure, improve endothelial function -possibly by exerting protective effects on the vessel wall, decrease the prevalence of infla-mmatory markers, and reduce the prevalence of non-fatal myocardial infarction and stroke.2,12,17-19 In summary, it is thought that the cardiovascular and glycemic benefits of TZD therapy often outweigh the risks assoc-iated with fluid retention and edema, although these side effects are important to consider.13,20,21Brain Natriuretic Peptide (BNP) has recently been recognized as an efficient screening method for heart dysfunction that has predictive and diagnostic value22-25. This hormone is secreted primarily from the left ventricle in response to increased volume or pressure overload in the heart24. BNP is produced as a prohormone and is subsequen-tly processed into two parts: BNP, the active moiety, and N-terminal pro-Brain Natriuretic Peptide (NT-proBNP), the inactive moiety. 25,26 Although both hormones provide similar diagnostic information, comparisons suggest that NT-proBNP is the more sensitive diag-nostic marker.25-27The present study aims to address the current uncertainty in literature that exists with respect to the effect of TZD therapy on heart function in type 2 diabetes patients. As NT-proBNP is a reliable indicator of heart failure, this study assesses the within-person changes of this hormone that occur when patients with type 2 diabetes and no known heart failure, initiate TZD therapy.

Materials and Methods


The present study included 76 consecutive consenting outpatients with type 2 diabetes who were prescribed either rosiglitazone (67% of patients) or pioglitazone (33% of patients) while visiting St. Paul’s Hospital Diabetes Education and Treatment Center in Vancouver, British Columbia, Canada. Recr-uited patients had been diagnosed with type 2 diabetes according to the current American Diabetes Association criteria for diagnosis28 and were not currently on a TZD, but could be using any combination of anti-diabetic therapies. Patients were prescribed TZD dosages depending on the patient’s therape-utic requirement. When considering rosiglita-zone, 53% of patients received 4 mg, 9% of patient received 8 mg and 5% of patient received 2 mg. When considering pioglita-zone, 20% of patients received 30 mg and 13% of patients received 15 mg. Patients with heart failure were excluded. Informed consent was obtained prior to enrolment in the study, and the St. Paul’s Hospital Ethics Board approved the study protocol, which conforms to the ethical guidelines of the 1975 Declaration of Helsinki.
Prior to initiating TZD therapy (baseline), patients completed a non-fasting serum NT-proBNP assay. Three months later (follow-up), participants completed a second assay for NT-proBNP. A medical chart review was also performed at both baseline and follow-up in order to determine demographic information (age, gender, duration of diabetes) and to assess changes in diabetic parameters (A1C, BMI, serum creatinine).


Blood was drawn by venopuncture and separated for serum samples (Heparin/ EDTA) in glass tubes. Analysis was perfor-med within 24 hours of sample collection. Samples were kept at room temperature for temporary storage. The Roche Elecsys 1010 bench top analyzer for heterogeneous immu-noassay analyzed the samples (Roche Diagn-ostics Inc., Germany).22 The system is an Elect-rochemiluminescense (ECL) machine that uses 2 polyclonal antibodies for detecting NT-proBNP.22 The between run coefficient of variation for this assay platform ranges from 4.4-5.3%, as stated by the manufacturer.22



Data was analyzed using a statistical software program (R, version 1.7.0). A two-tailed matched-pairs t-test was used to compare clinical parameters (NT-proBNP levels, A1C, BMI, and creatinine) at baseline and follow-up. Two-sample t-tests were performed to determine whether changes in NT-proBNP differed between females and males, and between patients who took pioglitazone and those who took rosiglitazone.

A Spearman rank correlation coefficient

was calculated to assess association between change in NT-proBNP levels and the following: age, duration of diabetes, TZD dosage, baseline NT-proBNP, baseline A1C and baseline serum creatinine. A simple regression model was fitted in order to determine whether changes in NT-proBNP were related to concurrent diabetes medicat-ions such as metformin, sulfonylureas, or insulin. In all cases, statistical significance was established at p<0.05.


Of the 76 patients who consented to participate in this study, 50 were included in data analysis. Twelve patients were excluded due to noncompliance in obtaining blood, while 9 discontinued TZD therapy during follow-up for various reasons such as noncompliance (n=2), stroke (n=1), high blood pressure (n=1), edema (n=2), dizziness (n=1), hypoglycemia (n=1), and arthralgia (n=1). Two patients were switched to insulin therapy during the follow-up period. Of the remaining 53 patients, 3 were excluded from analysis because their baseline NT-proBNP levels exceeded 1000 pg/mL.19 Demographic data for the study population is presented in Table 1. The subjects (42% female, 58% male) ranged in age from 30 to 81 years.

Table 1. Demographic characteristics for study population

Baseline Characteristic  
Female (n)
Male (n) 
Duration of Diabetes 
11.4 ± 8.3*
Blood pressure (mmHg) 
30.9 ± 8.7
Anti-Diabetic Medications
38 (76%)
6 (12%)
 Metformin and Insulin
2 (4%)
1 (2%)
1 (2%)
1 (2%)
 Metformin, Sulfonylurea+Acarbose

* mean±SD

A comparison of NT-proBNP levels, A1C, BMI and creatinine at baseline and follow-up are shown in Table 2 for all subjects. Although mean NT-proBNP and creatinine both increased slightly throughout this timeframe, neither of these trends was significant. On the other hand, A1C values did improve markedly, decreasing by 1.0±1.2% (p<0.0001). A significant mean increase in BMI of 0.6±1.5 kg/m2 was also observed (p<0.05).


Table 2. Within-person changes in clinical characteristics

Characteristic Before Treatment After Treatment Net Change P Value
NT-proBNP (pg/mL)
150.6±186.2 6.8 ± 91.0 0.6
HbA1c (%)
8.7±1.1 7.7±1.5 -1.0 ± 1.2 < 0.0001
BMI (kg/m2)  
30.9±8.7 31.5±8.9 < 0.05 < 0.05
 Creatinine (μmol/L)  
2 ± 12

Data are means ± SD unless otherwise indicated; HbA1c : Hemoglobin A1c; BMI: Body mass index.

NT-proBNP levels of the female subgroup increased nonsignificantly from 145.9±202.6 pg/mL to 172.8±225.7 pg/mL, whereas mean NT-proBNP levels of the male subgroup decreased nonsignificantly from 142.2±208.4 pg/mL to 134.5±153.7 pg/mL during the therapy period. However, changes in NT-proBNP were not significantly differ-ent between males and females.

We assessed correlation between change in NT-proBNP and the following demogra-phic and clinical parameters: age, duration of diabetes, baseline NT-proBNP, baseline A1C, baseline creatinine and dosage of TZD. Results are presented in Table 3. Results indicated no association between change in NT-proBNP and age, duration of diabetes, baseline creatinine, baseline HbA1c or dosage of TZD. A negative correlation between change in NT-proBNP and baseline NT-proBNP existed (p<0.05).

Table 3. Relationship between change in NTproBNP and clinical characteristics



P Value




Duration of Diabetes 



Baseline NT-proBNP 



Baseline HbA1c 



Baseline Creatinine 



Dosage of TZD



Throughout the study period, 67% of patients took rosiglitazone and 33% took pioglitazone. NT-proBNP levels of patients prescribed pioglitazone decreased from 105.3±175.4 pg/mL to 95.9±85.9 pg/mL, while NT-proBNP levels of patients who initiated rosiglitazone increased from 159.8±213.6 pg/mL to 176.5±219.4 pg/mL. Changes in NT-proBNP were not significantly different for those prescribed pioglitazone versus those prescribed rosiglitazone.
We assessed whether existing diabetes medications-those that the patient was taking prior to TZD prescription and continued to take throughout the treatment period - were associated with changes in NT-proBNP. When recruited, patients were on various combination or monotherapy regimes. For each type of existing medication, no significant association between taking the medication and change in NT-proBNP was evident.


TZD medications are shown to have a number of cardiac benefits.2,13,18-21 The present study aimed to assess the relationship between TZD therapy and levels of NT-pr