32 Diabetes mellitus

Heart failure (HF) and diabetes mellitus (DM) are both increasing in prevalence worldwide, and HF is a serious and increasingly common comorbid factor in the patient with DM. The patient with HF and DM may present particular problems in relation to a number of management areas. As a consequence, it is increasingly important that practitioners dealing with patients with HF have some knowledge of the interplay between DM and HF and the potential pitfalls encountered when treating the diabetic patient with HF, and vice versa.

The initial observations from the Framingham Heart Study population demonstrated that HF was twice as common in men with DM and five times as common in women with DM aged 45–74 as compared to their age-matched controls, and the risk of HF was independent of age, the presence of hypertension, obesity, coronary artery disease (CAD), or dyslipidaemia.¹ Furthermore, in patients with DM aged 65 or less, the prevalence of HF was even greater—four times higher in men with DM and eight times higher in women with DM. The increased risk of developing HF conferred by DM in women was confirmed in the HERS population where DM was the strongest predictor of HF, more so than the presence of CAD.²

The presence of DM itself is also an independent predictor of developing HF following myocardial infarction,³ and if developed predicts a poorer outcome than in nondiabetic patients—greater even than the presence of CAD.^4,5 Furthermore, patients hospitalized with HF have a poorer outcome if they have DM, with a blood glucose level of more than 10 mmol/L being associated with a poor outcome.^6,7

Population-based studies have estimated that 0.3–0.5% of the population have both DM and HF,^8,9 whereas the prevalence of HF in patients with DM is between 12–22%, becoming more prevalent with age (Table 32.1).^8,10

The prevalence of DM in patients with left ventricular systolic dysfunction (LVSD) varies significantly dependent upon the population studied. The background prevalence of DM is 4–7% in northern hemisphere populations,^8,11 but varies from 6% to 44% in patients with varying degrees of severity of HF (Table 32.2). Studies of treatment of HF due to LVSD have consistently demonstrated that 20–30% of patients with HF have DM,^12,–16 although a similar prevalence was observed in patients with HF and preserved left ventricular function.¹⁷

By contrast, hospital-based studies have demonstrated consistently higher prevalence of DM in patients with HF ranging from 34% to 44%.^18,19 Although it appears that intervention studies underestimate the true prevalence of DM in patients with HF, it is still abundantly clear that the prevalence of DM in patients with HF is significantly higher than the background prevalence of DM in the non-HF-affected population.

The Framingham study identified DM as an important independent risk factor for developing HF.¹ This has been confirmed by a number of population-based studies in the US and Europe, which all confirm with some consistency an age-adjusted odds ratio for developing HF of around 2 compared with nondiabetic subjects.^1,8,10 The annual incidence of HF in patients with DM has been assessed in five studies.^20,–24 The UKPDS reported HF incidence rates of 2.3–11.9 per 1000 patient-years over a 10-year follow-up period. The DIABHYCAR study determined the annual incidence rate of HF requiring hospitalization in subjects with DM and estimated it to be 1% (10 per 1000 patient-years). A large diabetic population (48 000 subjects with a mean age of 58 years) was studied and demonstrated an incidence rate of 4.5–9.2 per 1000 patient-years. This study only recorded those who were hospitalized with the principal diagnosis of HF, thus excluding less severe cases and underestimating the true incidence of HF. The MICRO-HOPE substudy placebo group of patients with DM had an incidence of HF of 13.3% over 4.5 years.

A retrospective cohort study of over 16 000 patients with and without type 2 DM followed for 6 years confirmed that patients with DM were much more likely to develop HF than those without (incidence rate 30.9 vs. 12.4 cases per 1000 person-years, giving a rate ratio of 2.5), and in particular the rate of developing HF was greatest in younger age groups (age 45–54 with DM vs. no DM odds ratio 8.6; age 75–84, odds ratio 1.2). In addition to the effect of age, the authors also concluded that poor glycaemic control and obesity were also important factors affecting the development of HF in patients with type 2 DM.

Thus, the incidence of HF in DM is significantly greater than the nondiabetic population, and whilst the size of the diabetic population increases, the potential burden of HF comorbidity is increasing too, perhaps at a rate greater than expected based on the rise in diabetes cases.²⁵

In one population-based study in elderly Italians, the odds ratio for developing DM in patients with HF was 1.6 versus patients without HF, with the absolute incidence of DM in the HF group 28% over 3 years.²⁶ Within the context of clinical trials, the incidence of DM in patients with HF is 5.9–7.4% over 3 years,^27,28 while the BIPS demonstrated an incidence of DM in patients without HF of 13% over a mean of 7.7 years versus 15% in patients with NYHA class II and 20% in patients with NYHA class III HF.²⁹ Clearly, DM is an independent risk factor for developing HF, but it is apparent that HF is an independent risk factor for developing DM. However, the mechanism behind this increased risk is unclear.

The worldwide prevalence of DM is alarmingly high. Current estimates suggest that over 200 million people have diabetes worldwide, with the total number projected to be 300 million by 2025.³³ In developed countries over 25% of people aged 65 or older will have diabetes.

Patients with diabetes have an excess burden of vascular complications, both macrovascular (i.e. coronary artery disease, cerebrovascular disease, peripheral arterial disease and HF) and microvascular (i.e. diabetic retinopathy, nephropathy and neuropathy), and three out of four deaths in patients with DM are attributed to cardiovascular causes.²⁰ In addition to hyperglycaemia, the diabetic patient has an increased prevalence of other key cardiovascular risk factors, namely hypertension, obesity, dyslipidaemia, and chronic kidney disease (CKD). It is likely that the interaction of hyperglycaemia with these cardiovascular risk factors leads to the development of structural and functional changes in the vasculature and myocardium which contribute to the excess prevalence of HF in patients with DM.

The rise in the prevalence of obesity together with the reclassification of the diagnostic criteria for DM has led to an increased awareness of diabetes. This has led to the implementation of screening of individuals at risk for developing DM, e.g. those with obesity or positive family history of DM. Similarly, improved patient access to diagnostic testing, the application of evidence-based cardiovascular prevention strategies for patients with DM, improved detection and treatment of renal disease, and increasing use of revascularization strategies for coronary and peripheral vascular disease have all contributed to the increase in size of the patient population with DM, and in particular those surviving with HF or the substrates to develop HF.

In 1997, the diagnostic criteria and aetiological classification of DM were updated.³⁴ The critical changes made were firstly, the lowering of the level of fasting blood glucose required to diagnose DM to 7 mmol/L or greater; to propose the routine use of fasting blood glucose rather than the standard 75-g oral glucose tolerance test; and the introduction of a new category of abnormal glucose regulation—impaired fasting glucose (Table 32.3).

Most recently, the use of a single measure of HbA_1c (the glycosylated fraction of haemoglobin typically used to monitor glycaemic control) to diagnose DM (random HbA_1c ≥ 6.5%) has been advocated, although this is still a matter of considerable debate.^35,36

Type 1 DM accounts for 5–10% of cases of DM and is characterized by an absolute deficiency of insulin, most commonly in the context of associated cellular autoimmunity directed towards the pancreatic beta cells. Between 80% and 90% of those affected have evidence of autoimmunity against pancreatic beta cells, e.g. the presence of anti-islet cell antibodies, or anti-GAD₆₅ antibodies. As a consequence, individuals with type 1 DM require insulin therapy lifelong from diagnosis and are prone to developing ketoacidosis. Such individuals are also at high risk of developing the classic manifestations of DM such as diabetic retinopathy and nephropathy in young adulthood, and have a reduced life expectancy principally due to cardiovascular deaths.

Type 2 DM is characterized by the presence of peripheral tissue insulin resistance and elevated insulin concentrations and is typically found in individuals who are centrally obese. As time progresses, the continuum from normal glucose tolerance to the development of DM in susceptible individuals with insulin resistance is associated with a gradual rise in fasting and meal-stimulated insulin concentrations associated with rising levels of fasting and postprandial blood glucose. The precise mechanism of insulin resistance is still unclear, although peripheral tissue resistance (principally adipose tissue and skeletal muscle) to the transmembrane and intracellular effects of insulin associated with fasting and postprandial hyperinsulinaemia, the presence of centripetal obesity and abnormal regulation of hepatic glucose production and fatty acid metabolism are contributory features. At some point, the progressive rise in insulin concentration will plateau for a variable period of time, and subsequently begin to fall leading to an eventual state of relative or absolute insulin deficiency. This fall in insulin production is related to the presence of impaired beta cell function, which is virtually always present when type 2 DM is diagnosed. The rate of decline in beta cell function in cases of type 2 DM varies between individuals but clearly contributes to the progressive hyperglycaemia seen in type

2 DM, the progressive failure of glycaemic response to agents such as sulphonylureas, and the increasing requirement for exogenous insulin to maintain levels of glycaemic control as the duration of type 2 DM continues.

Based on the key aetiological observations regarding the development of type 2 DM, interventions to control blood glucose levels in patients with type 2 DM are directed at reducing insulin resistance, directly stimulating remaining beta cell insulin release, and modifying meal-related insulin release. Insulin therapy in type 2 DM is usually initiated once one or a combination of these approaches has been tried and deemed to have failed by the treating clinician.

The two most common risk factors for the development of HF are the presence of CAD and arterial hypertension. Both of these are more prevalent in patients with DM than nondiabetic subjects and clearly this impacts upon the increased prevalence and incidence of HF in diabetic patients.

In addition to these two major risk factors, a number of features associated with DM have been identified as independent risk factors for developing HF. Poor glycaemic control, increasing BMI, increasing age, the use of insulin,²⁴ the presence of any measurable renal insult from microalbuminuria to endstage renal failure (ESRF),^10,24 and duration of DM are all independent risk factors for the development of HF in patients with DM. For instance, a 1% reduction in HbA_1c in UKPDS reduced the risk of HF by 16%,²⁰ and a 2.5-unit increase in BMI increases the risk of HF by 12%.²⁴

Although most registry-based studies and intervention trials for the treatment of chronic HF suggest that the principal cause of LVSD in patients with DM is ischaemic in origin, a sizeable proportion appear to arise from the entity sometimes referred to as ‘diabetic cardiomyopathy’.³⁷ The existence of a specific diabetic cardiomyopathy has long been debated and attempts made to characterize specific pathological and diagnostic features. The initial description was based on a small number of individuals noted to have a clinical diagnosis of HF but with no evidence of prior CAD or hypertension.³⁸ Using this simple clinical approach suggested that diabetic cardiomyopathy was a rare entity. However, further investigation has revealed a range of abnormalities in the diabetic heart which suggests not only that diabetic cardiomyopathy is a real entity, but also that it is extremely prevalent in patients with DM. These abnormalities are summarized in Table 32.4.^37,39

The initial feature of diabetic cardiomyopathy appears to be the development of features of diastolic dysfunction, and this early feature correlates closely with HbA_1c in diabetic patients.⁴⁰ Although it is suggested that diastolic dysfunction may account for approximately one-half of all HF cases,³¹ studies of patients with both type 1 and type 2 DM have suggested that diastolic dysfunction is much more common than previously reported in subjects who are free of clinically detectable heart disease.^41,–43

Once present, diastolic dysfunction has a prognosis similar to systolic dysfunction and the combination of either left ventricular hypertrophy, CAD, or both has a profound deleterious effect on the diabetic heart.^3,–7 Thus, in addition to optimizing glycaemic control, rigorous attention to blood pressure and cardiovascular risk modification is essential in the patient with DM.

Fasting and postprandial hyperglycaemia are the key diagnostic features of DM and this is clearly responsible for the development of the microvascular complications of DM, and is an important factor in the genesis of the excess burden of cardiovascular complications seen in patients with DM; however, the management of the patient with DM should not be simply glucocentric.

The management of cardiovascular risk, with the purpose of reducing end-organ damage, specifically HF, is complex and requires meticulous attention to the control of blood pressure, aggressive modification of dyslipidaemia, appropriate use of antiplatelet therapy, and constant attention to lifestyle modification (Table 32.5).

Crucially, treating hypertension in diabetics reduces both microvascular and macrovascular complications. In particular, tight blood pressure control in UKPDS reduced new cases of HF by 44%.⁴⁴ However, this benefit did not persist after relaxation of blood pressure control, suggesting that aggressive blood pressure control should be instituted when first detected and maintained lifelong.^45,46

Treating hyperglycaemia to reduce the vascular complications of DM seems intuitively straightforward. However, the first study that attempted to address this question suggested that pharmacological measures to lower blood glucose levels in patients with type 2 DM with the sulphonylurea drug tolbutamide actually resulted in an excess of cardiovascular deaths.⁴⁷ This study has been the centre of much controversy and more recently a number of well-conducted, informative studies have assessed the benefit, if any, of blood glucose lowering via pharmacological intervention in patients with type 1 and type 2 DM.

Principal amongst these studies has been the UKPDS.⁴⁸ This UK-based study assessed the value of ‘tight’ glycaemic control requiring escalating doses of oral hypoglycaemic agents, insulin or both (i.e. treatment based on achieving near normal fasting or postprandial blood glucose levels), versus ‘conventional’ glycaemic control (i.e. treatment based on symptom control). Overall, the application of the tight control approach resulted in an absolute reduction in HbA_1c of approximately 1% compared to the conventional treatment group,⁴⁸ which translated into a number of clear benefits in terms of vascular risk modification. Although there was no clear significant benefit of tight glycaemic control on the overall risk of myocardial infarction, treatment with metformin in obese patients with type 2 DM did result in a significant reduction in the risk of myocardial infarction. Overall, the risk of developing HF increased by 8% for every 1% absolute rise in HbA_1c in keeping with the intuitive notion of good glycaemic control reducing the risk of HF.²⁰

Although prolonged follow-up of the UKPDS demonstrated that HbA_1c measures in the tight control group and the conventional control group gravitated towards each other after study completion, a legacy effect was demonstrable in the tight control group beyond the period of strict tight control.⁴⁹ This translated to 10 years of clear benefits in terms of a reduction in myocardial infarction of 15% (p = 0.01) in the sulphonylurea-treated group, and a reduction in myocardial infarction of 33% (p 〈 0.005) and death from any cause of 27% (p = 0.002) in those overweight patients treated with metformin.⁴⁹

The cornerstone of the management of all patients with DM is lifestyle modification aimed at maximizing daily activity, stopping tobacco exposure, attaining as near an ideal weight and BMI as practical, and encouraging healthy eating habits to minimize salt, saturated fat, refined carbohydrate, and excess calorie intake.⁵⁰ However, in the case of type 1 DM and the vast majority of cases of type 2 DM, adjunctive therapy with either oral agents, insulin, or both is necessary, not just to relieve the symptoms of hyperglycaemia, but to achieve as near normal blood glucose control as possible with the hope of minimizing DM-related complications including HF.

Until the mid 1990s, the only oral agents available for treating hyperglycaemia in DM were old drugs, or their mildly altered derivatives. Since the late 1990s, however, two new classes of agents have become widely available for treating hyperglycaemia in DM, with other novel agents in development (Table 32.6).

Metformin is a biguanide drug which has been available for treating type 2 DM for over 40 years. Its precise mode of action is unclear, but it does reduce hepatic gluconeogenesis and peripheral insulin resistance leading to a reduction in both fasting and postprandial hyperglycaemia. Metformin is widely used for the treatment of type 2 DM worldwide, with most national guidelines suggesting it as the first-line oral agent for type 2 DM once lifestyle modification has been implemented and is no longer sufficient to maintain the set glycaemic target.⁵¹ Metformin is frequently used in combination with other oral agents, insulin, and other injectable agents in type 2 DM, and is used in patients with type 1 DM when there is clinical evidence of insulin resistance.

Metformin’s popularity stems from its positive association with reducing cardiovascular events in obese patients with type 2 DM in the UKPDS.⁵² Within the UKPDS, metformin treatment was associated with a reduction myocardial infarction which was sustained at 10 years of follow-up.⁴⁹

The major limitation to treatment with metformin is its propensity to cause gastrointestinal upset, particularly nausea and diarrhoea, which necessitates discontinuation of the drug in up to 20% of cases. Much less common is the potential for megaloblastic anaemia due to interference with vitamin B₁₂ absorption. Metformin is renally excreted, and can accumulate in the presence of renal impairment. Current guidance in the UK suggests that withdrawal of metformin be considered when serum creatinine reaches 150 µmol/L or eGFR falls below 30 mL/min.⁵³

Of greater concern is the association between metformin and lactic acidosis. This association, although well documented, is extremely rare and spontaneous isolated cases of metformin-induced lacticacidosis are extremely uncommon (less than 1 case per 100 000 treated patients).^54,55 In clinical practice, it is wise to withdraw metformin temporarily in patients undergoing diagnostic procedures with the use of contrast, and those patients with HF who experience intercurrent illness which might be associated with tissue hypoxia, e.g. acute decompensated HF, myocardial infarction, or pneumonia. Similarly, patients with HF treated with metformin should have regular monitoring of renal function to ensure that metformin can be withdrawn in the event of a rapid decline in renal function, or in the context of a slow decline to a level where concern about accumulation outweighs potential benefits on glycaemic control and cardiovascular event reduction.

Initial experience with sulphonylureas was tempered by the UGDP experience⁴⁷ and some observational and retrospective studies suggest that these drugs are associated with increased cardiovascular mortality.^56,57 Subsequent investigations using newer agents with different properties have established the class as a popular choice for treating hyperglycaemia in patients who are no longer satisfactorily controlled on metformin alone.

However, neither the UKPDS nor ADVANCE studies demonstrated any adverse cardiovascular mortality effects associated with the use of sulphonylureas.^49,58 Their rapid onset of action and low cost is particularly attractive when treating symptomatic patients, but on the downside, use of sulphonylureas to achieve tight glycaemic control is associated with weight gain and significant potential for symptomatic hypoglycaemia.^58,59 Furthermore, sulphonylureas appear to exhibit a rather more rapid decline in efficacy over time than either metformin or thiazolidinediones,⁶⁰ making their use as first-line agents less attractive.

Thiazolidinediones (TZDs, glitazones) are modulators of the peroxisome proliferator-activated receptor γ (PPAR-γ), and increase the sensitivity of skeletal muscle, adipose tissue and liver to insulin.⁶¹ As single agents, thiazolidinediones have a similar magnitude and duration of effect on glycaemic control to metformin,⁶⁰ with lower risks of hypoglycaemia than sulphonylureas. However, they have two adverse effects that have limited their use in the treatment of patients with DM, especially those with or at risk of developing HF—weight gain and fluid retention.

Weight gain with thiazolidinediones is similar or greater than that seen with sulphonylureas, although it is often associated with some modest benefits on lipid profiles.⁶⁰ Fluid retention is mediated via the kidney, and peripheral oedema and HF are both increased in users of thiazolidinediones. Although it appears that thiazolidinediones do not directly affect left ventricular function,⁶² there is no question that they do cause HF, even in the presence of normal left ventricular function.^63,64

Perhaps the greatest controversy surrounding thiazolidinediones is the suggestion that they may increase the risk of myocardial infarction or death. Initial evidence suggested that pioglitazone may have a modest beneficial effect on cardiovascular outcomes in patients with type 2 DM.⁶⁵ However, a controversial meta-analysis of data pertaining to rosiglitazone suggested that it was associated with a 43% increased risk of myocardial infarction.⁶⁶ This result was followed by a raft of publications which have confirmed that both thiazolidinediones increase rates of peripheral oedema, HF, and hospitalization due to HF.^54,63

A retrospective cohort study of adverse cardiovascular events associated with thiazolidinedione use demonstrated no difference in the incidence of myocardial infarction in those treated with either pioglitazone or rosiglitazone, although the risk of death or HF was higher in those treated with rosiglitazone. This translated into a numbers needed to harm (NNH) of 120 for HF and 293 for death, for users of rosiglitazone rather than pioglitazone.⁶³ However, it should be borne in mind that the absolute event rates quoted (risk of myocardial infarction plus HF plus death, pioglitazone 5.3% vs rosiglitazone 6.9% over 6 years) equate to a 10-year risk of cardiovascular disease that barely reaches 10%.

Thiazolidinediones are available for use in the treatment of type 2 DM worldwide, either in combination with oral agents, or insulin in selected cases. They are not recommended for use in patients with known HF, and should be used with caution in patients at risk of developing HF.⁶⁴

Insulin is the longest-serving therapeutic option for the treatment of DM. Multiple formulations exist and several means of insulin administration are used in an attempt to mimic physiological insulin profiles and restore normal blood glucose profiles in patients with DM. The use of insulin in the treatment of patients with type 2 DM is increasingly common, principally due to the increasing use of glycaemic targets promoted by national and international expert committees.⁶⁷

Patients with HF exhibit increased resistance to insulin, mediated via a variety of different mechanisms. As a consequence, insulin doses required to achieve adequate glycaemic control in type 2 patients with HF are substantially higher than in patients with type 1 DM or type 2 DM without HF.

UKPDS did not show that insulin treatment increased the incidence of HF or mortality.⁴⁸ However, some studies have found that the use of insulin in patients with DM is an independent predictor for the development of HF, as well as being associated with increased mortality.^24,68,–70 Diabetics without HF commenced on insulin may also have a higher rate of hospitalization due to HF than those commenced on sulphonylureas.⁷⁰ The CHARM study also demonstrated a greater risk of mortality than for non-insulin-treated patients.⁷¹ A further retrospective analysis of patients with advanced HF suggested that insulin treatment was an independent predictor of mortality.⁶⁹ Therefore, at present there are no prospective evaluations of insulin therapy in patients with DM with or without HF to advise on the precise role of insulin treatment.

A number of other drugs are available for patient use to lower blood glucose in patients with DM. At present there are no data to determine their efficacy or otherwise in the management of the patient with DM and HF.

Data on treatment efficacy for HF patients with DM are largely derived from subgroup analyses from the major HF treatment trials (Table 32.7).

ACE inhibitors are widely used to treat hypertension, microalbuminuria, and proteinuria in patients with DM. Their benefits in the treatment of HF are well established, although in patients with DM the benefits are less apparent (Table 32.7).⁹⁰

β-Blockers are effective treatments for HF. In patients with HF and DM the benefits of β-blocker therapy are similar to those seen in nondiabetic subjects in reducing mortality and hospitalization due to HF (Table 32.7).⁹⁰

A major concern regarding the application of β-blocker therapy is the potential for this class of drugs to alter insulin sensitivity and alter hypoglycaemia awareness. Hypoglycaemia results in a pronounced activation of the sympathetic nervous system, and release of cortisol and glucagon. These responses aim to increase hepatic glucose output, and increase blood supply to the brain and other glucose sensitive tissues. The UKPDS assessed the rates of hypoglycaemia in patients with hypertension and DM and did not discern any difference in the rate of hypoglycaemia in patients treated with atenolol or captopril,⁴⁴ whereas one study of elderly diabetic patients suggested that insulin-treated patients were more likely to experience severe hypoglycaemia than those treated with sulphonylureas.⁹¹

In general, there are few or no data to suggest that β-blockers used to treat HF in patients with DM specifically affect hypoglycaemic awareness or recovery from symptomatic hypoglycaemia, or adversely alter lipid metabolism to the detriment of patients. Furthermore, the benefits of treatment with β-blockers in terms of reduction in critical endpoints far outweighs any potential effects on other measures.

There are no specific data in relation to the effect of digoxin in patients with HF and DM. However, DM did not affect the benefit of hydralazine plus isosorbide dinitrate in A-HeFT.⁸⁹ Thiazide diuretics have the potential to increase fasting blood glucose levels although their effect on reducing blood pressure, preventing HF and reducing strokes far outweigh any minor effect on glycaemic control.⁹⁰

Table 32.8 lists a number of key practice points that are worth paying attention to in day-to-day practice when dealing with the patient with HF and DM.

DM and HF are increasingly common conditions, and may therefore frequently coexist. Diabetes imparts a greater risk of morbidity and mortality on the patient with HF and complicates the management of HF, and vice versa. Optimizing cardiovascular risk factors is essential for limiting adverse outcomes for both conditions, and patients with diabetes mellitus are, at present, still less likely to receive optimal care whether as a result of fear of application of some evidence-based strategies, or the inability to tolerate a number of therapeutic agents. Glycaemic control is an important factor in the management of the patient with DM and HF, and optimized patient-specific approaches are most likely to minimize adverse events while maximizing the opportunity for reducing long-term complications.