Secondary Logo

Journal Logo

The modulating effect of differences in cardiovascular risk factors on major cardiovascular outcomes in glucose-lowering trials

‘good to know’ but not the whole story

Gasevic, Danijela; Zoungas, Sophia

doi: 10.1097/HJH.0000000000002181

Diabetes and Vascular Medicine Unit, The School of Public Health and Preventive Medicine, Monash University, Melbourne, Victoria, Australia

Correspondence to Sophia Zoungas, The School of Public Health and Preventive Medicine, Monash University, Victoria, Australia. E-mail:

The International Diabetes Federation estimated that there were 425 million people between the ages of 20 and 79 years who had diabetes in 2017 (8.8% of world population in this age group); and these numbers are projected to rise to 629 million by 2045 [1]. Cardiovascular disease (CVD) is a major cause of death and disability among people with type 2 diabetes (T2D) [2]. A recent review of the prevalence of CVD in T2D estimated that CVD affected one-third (32.2%) of all persons and accounted for about half (50.3%) of all deaths [3]. Cardiovascular complications contribute substantially to the total healthcare expenditure on diabetes [4]; estimated to increase from USD 727 billion in 2017 to USD 776 billion by 2045 (20–79 years) [1].

Reducing CVD burden in T2D has become a major clinical imperative. This imperative has been underscored by regulatory agencies (Food and Drug Administration in 2008 [5] and European Medicines Agency in 2012 [6]) presenting guidance for addressing and demonstrating cardiovascular safety of new glucose-lowering therapies. As a consequence, multiple large-scale cardiovascular outcome trials have recently been conducted [7].

In this issue of Journal of Hypertension, Thomopoulos et al.[8] report a comprehensive overview of glucose-lowering randomized controlled trials (RCTs) and present the relative effects of glucose lowering on major cardiovascular outcomes. Namely, they perform a meta-analysis of 25 trials investigating glucose lowering with 174 235 participants followed up for a mean of 3.5 years. Pooled studies include more versus less intensive glucose-lowering trials, placebo-controlled intentional glucose-lowering trials and placebo-controlled nonintentional glucose-lowering trials [in which 0.2% or greater glycated haemoglobin (HbA1c) between-group differences were observed]. The objectives of the overview were three-fold: first, to quantify the relative effects of glucose lowering on major clinical outcomes including therapy discontinuation; second, to relate the proportionality of outcome reductions with effects on HbA1c; and third, to explore whether ongoing blood pressure (BP) and LDL-cholesterol (LDL-C) differences in RCTs modulate the effects of glucose lowering. Eight outcomes were considered including coronary heart disease (CHD) events (coronary death and nonfatal myocardial infarction), fatal and nonfatal stroke, hospitalized heart failure, composite of CHD and stroke, composite of CHD, stroke and heart failure, cardiovascular death, all-cause death and treatment-related discontinuation. Taking all trials together, the observed SBP/DBP and LDL-C differences at the end of follow-up were −1.4/−0.4 mmHg and 0.38 mg/dl, respectively.

The authors report significant but modest risk reductions from assignment to glucose lowering across all clinical outcomes apart from treatment-related discontinuation which as anticipated increased. Treatment effects were related to HbA1c reductions for the outcomes of CHD, stroke and treatment-related discontinuations. Adjustment for the modest BP differences did not produce any substantive changes to the magnitude of the effects; however, the risk reductions for stroke and all-cause death were no longer statistically significant. No differences in BP-adjusted risk reductions were observed across subgroups defined by baseline or achieved SBP or DBP, baseline or achieved HbA1c, baseline LDL-C, baseline 10-year cardiovascular risk, diabetes duration or study year. Significantly, the LDL-C differences were deemed too small to consider in adjusted analyses.

We commend the authors for performing a comprehensive overview and meta-analysis of trials of glucose-lowering while exploring the interesting question of the modulating role of BP or LDL-C differences. Understanding the effects of glucose-lowering strategies on other cardiovascular risk factors is of clear interest given the need to manage overall cardiovascular risk. However, the authors’ conclusions should be considered in light of some limitations: first, the impact of any measurement error of the mediators, that is BP or LDL-C; second, the validity of pooling of disparate studies which may have diminished any real BP or LDL-C differences from within therapeutic classes; and third, the appropriateness of adjusting for the individual biological effects intrinsic to a drug per se.

First, BP is subject to measurement error; and large intraindividual variations [9] in BP measurement are related to time of day or week, season, environment, changes in person's physical and mental health, and behaviours around BP measurement (e.g. vigorous activity, heavy meal, smoking before measurement). To add to complexity, BP is subject to regression to the mean. In the context of BP measurement, regression to the mean is a statistical phenomenon that occurs when repeated BP measurements are made on the same person, whereby extreme BP values are followed by values that are closer to the population mean [10]. Regression to the mean is expected to occur in both treatment arms of an RCT; however, key features of RCT design (randomization, double-blinding) tend to reduce (but not exclude) the effect of regression to the mean on primary outcomes [11]. This raises questions about the appropriateness of exploring the modulating effect of a variable that is subject to measurement error and intraindividual variation, and validity of mediation inference [12]. Namely, it has been reported that the effect of measurement error of the mediator can substantially bias direct-effect estimates [13]. Therefore, the observed effects across clinical outcomes after adjustment for BP difference may be an underestimation of the direct effect as a result of the measurement error of the mediator.

Second, glucose-lowering drugs have previously been reported to impact BP differentially [14]. For example, when compared with placebo, sodium glucose cotransporter 2 inhibitors produce average SBP and DBP reductions of 4.68 (−6.69, −2.68) and 1.72 mmHg (−2.77, −0.66), respectively, whereas dipeptidyl peptidase-4 inhibitors and glucagon-like peptide 1 receptor agonists produce no or little effect [14]. As the overview was examining the modulating effects of BP differences, it would have been more prudent for it to have pooled and considered trials of each drug class separately.

Third, there is a question of how clinically relevant it is to disentangle the total effect of a treatment (antidiabetic medication) on cardiovascular outcomes while considering potential indirect effects (SBP/DBP difference). As stated above, BP-lowering is intrinsic to some, but not all glucose-lowering drugs. Similarly, intrinsic to some of the glucose-lowering drugs are effects on other CVD factors (e.g. an increase in HDL cholesterol, decrease in triglyceride concentration and decrease in body weight) [14]. Specific effects of individual CVD risk factors may be of less clinical relevance to absolute CVD risk as they each make a minor contribution and particularly in the absence of other risk factors [15]. Guidelines, therefore, recommend a multifactorial approach with simultaneous targeting of multiple CVD risk factors in individuals with T2D [7,16]. Therefore, considering the synergistic effects of CVD risk factors, the emphasis should be on multifactorial approaches rather than on isolated risk factors.

Overall, Thomopoulos et al.[8] explore the modulating effect of differences in CVD risk factors, specifically BP differences, after assignment to glucose lowering. The results are of interest but do not replace the need for further trials of new drugs that go beyond glucose lowering while addressing issues of study design (placebo controlled versus active comparator), treatment duration and real-world generalizability.

Back to Top | Article Outline


Conflicts of interest

There are no conflicts of interest.

Back to Top | Article Outline


1. International Diabetes Federation. IDF diabetes atlas. 8th edn.2017; Brussels, Belgium: International Diabetes Federation, IDF_DA_8e-EN-final%20(1).pdf. [Accessed 20 May 2019].
2. International Diabetes Federation. Diabetes and cardiovascular disease. 2016; Brussels, Belgium: International Diabetes Federation,
3. Einarson TR, Acs A, Ludwig C, Panton UH. Prevalence of cardiovascular disease in type 2 diabetes: a systematic literature review of scientific evidence from across the world in 2007–2017. Cardiovasc Diabetol 2018; 17:83.
4. Einarson TR, Acs A, Ludwig C, Panton UH. Economic burden of cardiovascular disease in type 2 diabetes: a systematic review. Value Health 2018; 21:881–890.
5. US Department of Health and Human Services; Food and Drug Administration; Center for Drug Evaluation and Research. Guidance for industry. Diabetes mellitus - evaluating cardiovascular risk in new antidiabetic therapies to treat type 2 diabetes. December 2008; Available at https://www.fda.-gov/media/71297/download. [Accessed 21 May 2019]
6. European Medicines Agency. Guideline on clinical investigation of medical products in the treatment or prevention of diabetes mellitus. May 2012; Available at [Accessed 21 May 2019]
7. American Diabetes Association. Cardiovascular disease and risk management: standards of medical care in diabetes – 2019. Diabetes Care 2019; 42 (Suppl 1):S103–S123.
8. Thomopoulos C, Bazoukis G, Ilias I, Tsioufis C, Makris T. Effects of glucose-lowering on outcome incidence in diabetes mellitus and the modulating role of blood pressure and other clinical variables: overview, meta-analysis of randomized trials. J Hypertens 2019; 37:1939–1949.
9. Warren RE, Marshall T, Padfield PL, Chrubasik S. Variability of office, 24-h ambulatory, and self-monitored blood pressure measurements. Br J Gen Pract 2010; 60:675–680.
10. Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. Int J Epidemiol 2005; 34:215–220.
11. Pocock SJ, Bakris G, Bhatt DL, Brar S, Fahy M, Gersh BJ. Regression to the mean in SYMPLICITY HTN-3: implications for design and reporting of future trials. J Am Coll Cardiol 2016; 68:2016–2025.
12. Corraini P, Olsen M, Pedersen L, Dekkers OM, Vandenbroucke JP. Effect modification, interaction and mediation: an overview of theoretical insights for clinical investigators. Clin Epidemiol 2017; 9:331–338.
13. Le Cessie S, Debeij J, Rosendaal FR, Cannegieter SC, Vandenbrouckea JP. Quantification of bias in direct effects estimates due to different types of measurement error in the mediator. Epidemiology 2012; 23:551–560.
14. Lo C, Toyama T, Wang Y, Lin J, Hirakawa Y, Jun M, et al. Insulin and glucose-lowering agents for treating people with diabetes and chronic kidney disease. Cochrane Database Syst Rev 2018; 9:CD011798.
15. Jackson R, Lawes CMM, Bennett DA, Milne RJ, Rodgers A. Treatment with drugs to lower blood pressure and blood cholesterol on an individual's absolute cardiovascular risk. Lancet 2005; 365:434–441.
16. IDF Clinical Guidelines Task Force. Global guideline for type 2 diabetes. Brussels: International Diabetes Federation; 2012.
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.