David Levy, MD, FRCP
General Recommendations on Macronutrients
- Recommended daily allowance is about 130 g.
- Low glycemic index foods (e.g. barley, oats, beans, lentils, rye bread) can reduce HbA1c by about 0.4% (4 mmol/mol) compared with high glycemic index foods. More importantly, high whole-grain, cereal fiber, bran and germ intake was associated with lower cardiovascular and all-cause mortality in women with diabetes in the Nurses Health Study.
- Fiber intake: similar to that in non-diabetic people, i.e. 14 g per 1000 kcal as high-fiber foods. A daily fiber intake of about 50 g (difficult to maintain) improves glycemia in both type 1 and 2 diabetes and bran intake was independently associated with lower mortality in the Nurses Health Study. However, the soluble fiber guar gum supplements in vogue in the mid-1990s have fallen out of use because of their weak and inconsistent glycemic effects, though they more consistently improve lipids and seem to improve hepatic extraction of insulin.
- Sucrose need not be excluded from the diet of diabetic patients: it yields similar blood glucose levels as starch.
- Fructose is increasing rapidly in the diet, mostly as the ubiquitous high-fructose corn syrup used in soft drinks and many prepared foods, rather than in fruit, which contains relatively small amounts. Fructose does not stimulate insulin secretion, and raises very low-density lipoprotein (VLDL) and urate levels, both components of the metabolic syndrome; there is concern that it is a significant promoter of population obesity .
The usual recommendation is 15–20% of total energy intake. High protein diets (∼30%) seem to be safe and more effective in achieving sustained weight loss than low-fat, high-carbohydrate diets, though the inevitable increase in fat should be as monounsaturated fatty acids rather than saturates.
Fats and fatty acids
- The ADA and the American Heart Association recommend less than 7% of total calorie intake as saturated fat, and limiting cholesterol intake to less than 200 mg/day. However, moderate intake of prawns and eggs, widely believed to be high in cholesterol, does not affect lipid profiles in non-diabetic subjects .
- Minimize trans fatty acid intake; their use is banned in some countries.
- Take two or more portions per week of oily fish (w-3 fatty acids, DHA and EPA).
- Increase cis-monounsaturated fatty acid intake as, for example, olive and rapeseed (canola) oils.
Many people with diabetes take vitamin and mineral supplements, but there is almost no evidence for their benefit. Several RCTs have included additional arms of micronutrient supplementation, but neither HOPE (vitamin E 400 IU/day) nor HOPE2 (folic acid, vitamins B6 and B12) found any benefit on cardiovascular or renal outcomes, and folic acid 1 mg plus vitamin B12 1 mg daily for 7 years did not reduce vascular outcomes in patients recovering after myocardial infarction in the SEARCH study, despite substantial reductions in homocysteine levels, a putative cardiovascular risk factor much in vogue a few years ago . Folic acid and vitamin B6 and B12 supplements are harmful in diabetic nephropathy (see Chapter 8). Vitamin D deficiency has often been linked epidemiologically with insulin resistance, and vitamin D insufficiency and deficiency is common, especially in obesity and in dark-skinned people. However, there is no conclusive evidence that it clinically improves insulin resistance or glycemic control. Nevertheless, because of its critical role in the skeleton (both type 1 and type 2 patients are at increased fracture risk), where detected or clinically likely, active vitamin D supplementation is important. The current recommendation is a daily intake of 800–1000 IU (20–25 µg). There is no consensus on optimum blood 25-hydroxyvitamin D levels or the daily intake of vitamin D3 needed to achieve them. However, levels of 37.5 nmol/L (15 ng/mL) or more are widely recommended, and some suggest levels in excess of 75 nmol/L (30 ng/mL) or even 90–100 nmol/L (35–40 ng/mL). Supplementation with 1000 IU (25 µg) daily will raise serum vitamin D levels by about 25 nmol/L (10 mg/mL), and will be required in the high proportion of UK patients with inadequate sun exposure.
Chromium, especially as chromium picolinate, has long been regarded as an insulin-sensitizing factor. Clinical studies have shown inconsistent glycemic benefits, but there may be minor improvements in dyslipidemia. High-cocoa content chocolate (> 70%, e.g. 50 g/day) has been shown to have short-term vasodilator and hypotensive effects; definitive trial results are very eagerly awaited, though British milk chocolate (cocoa content ∼20%, sugar ∼55%) is unlikely to figure prominently in current or future dietary recommendations.
The Diabetes Prevention Program, studying people with IGT, set an activity goal of 150 min/week, now a widely adopted target. These values lie within the evidence-based range of 150–250 min/week (energy expenditure of 1200–2000 kcal/week) to prevent weight gain in most non-diabetic adults. For weight loss, current recommendations are as follows:
- < 150 min/week leads to minimal weight loss;
- > 150 min/week leads to modest weight loss (about 2–3 kg);
- > 225 to 420 min/week leads to 5–7 kg weight loss.
‘Lifestyle’ physical activity (e.g. walking during commuting), often promoted as beneficial, but difficult to define or quantify, may contribute to prevention of weight gain, but even when done for long periods is unlikely to result in significant weight loss. However, exercise and calorie restriction are invariably linked in ‘lifestyle’ approaches to weight reduction. Combining moderate calorie restriction (500–700 kcal/day) with physical activity results in greater weight loss than diet alone; interestingly, this does not seem to be the case with more restricted calorie intakes. Even with calorie restriction, resistance training does not promote weight loss, though it may increase lean mass and loss of body fat and improve blood pressure and lipid profiles .
Drug Treatment of Type 2 Diabetes
The many classes of drugs now available for the treatment of type 2 diabetes operate through largely independent mechanisms. Most have been introduced in the past decade (Fig. 5.3). There are several hundred theoretical combinations of these classes, but even allowing for functional duplication and evident incompatibilities, only a handful have been subjected to RCTs, and no studies so far have systematically studied combinations beyond triple therapy. There is a strong analogy with antihypertensive agents, three or more of which are needed for control of blood pressure in many patients, most combinations unlicensed because untrialled in formal RCTs. This difficulty is reflected in current guidelines (e.g. NICE 2009), which includes about 11 licensed combinations, quite sufficient options for establishing glycemic control in the majority of patients but probably inadequate for the 10–20% of patients in persistently poor control, many of whom will have, or be at risk of developing, vascular complications.
- Strongly related to baseline glycemia: the higher the initial fasting glucose/HbA1c, the larger the initial fall. With a maximum dose of an additional agent (including insulin) expect:
- a. A fall in HbA1c of about 1.5% (17 mmol/mol) from a starting value of about 8.5% (69 mmol/mol); the corresponding fall in FPG is about 3 mmol/L (54 mg/dL).
- b. A fall in HbA1c of about 2% (21 mmol/mol) from an initial value of about 10%.
- c. Exception is acarbose, where maximum doses result in lower falls in HbA1c (e.g. up to 0.5%, 5 mmol/mol).
- Nadir glucose levels are usually reached at 16–24 weeks with active dose titration.
- Dose–response relationships: moderately strong for metformin, relatively weak for other agents, especially sulphonylureas.
- Sulphonylureas give a larger initial fall in glucose levels compared with metformin or the glitazones, but thereafter the long-term rise in glucose is faster than with metformin or the glitazones.
- Individual responses to the GLP-1 analogues, DPP-4 inhibitors and the glitazones are highly variable, and there is a null-response rate that is evident in clinical practice.
- For all agents except metformin, invest in slow and careful dose titration with 2 to 3-monthly HbA1c monitoring
- Above all, explain the variability of response in advance with patients; assuming good adherence, a weak response or non-response is likely to be related to the agent and not the patient.
Implications for glycemic management of type 2 diabetes in studies reporting in 2008 and 2009 Summary of UKPDS glycemic study
The glycemic arm of UKPDS (1998) concluded that intensive glycemic control from diagnosis with insulin or sulphonylureas (mean HbA1c 7.0% vs. 7.9%, 53 vs. 63 mmol/mol, over 10 years):
- significantly reduced the risk of serious microvascular end points (e.g. renal failure, vitreous hemorrhage, laser treatment);
- had a borderline significant effect on incidence of acute myocardial infarction; and
- had no effect on diabetes-related deaths.
Outcomes were similar with sulphonylureas (chlorpropamide or glibenclamide) and insulin, which had to be maintained for 6 years for improvements in retinal complications and 9 years for improvements in microalbuminuria. Intensive treatment with metformin in a small group of obese subjects:
- had no beneficial effects on microvascular end points;
- reduced myocardial infarction (this finding, one of the major headlines from UKPDS, may not be as clear-cut as widely believed) ; and
- reduced diabetes-related deaths.
Metformin thereby became first-line management, along with lifestyle intervention; intensive glycemic control established at diagnosis improved microvascular and macrovascular outcomes and glycemic targets were set at 7.0% (53 mmol/mol) or less by most international bodies. Epidemiological evidence consistently linking increasing HbA1c levels, even within the non-diabetic HbA1c reference range, with macrovascular events supported the progressive lowering of HbA1c targets to 6.5% (48 mmol/mol) or less despite a signal of increased cardiovascular event rates in intensively controlled long-standing diabetes (HbA1c 7.2% vs. 9.2%, 55 vs. 77 mmol/mol) in the Veterans Administration feasibility study (1995). The full Veterans Affairs Diabetes Trial (VADT) study , and two others, ACCORD  and ADVANCE , were performed specifically in the light of the equivocal UKPDS outcomes to address the question of whether intensive glycemic control improved cardiovascular outcomes. They were all treat-to-target studies in patients at high cardiovascular risk (either with a previous macrovascular event or several risk factors for it) designed to maintain stable glycemic differences between intensive and conventional control arms over several years. ACCORD patients were widely treated with rosiglitazone, ADVANCE with modified-release gliclazide. In VADT, patients with BMI above 27 were started with rosiglitazone and metformin, those with BMI below 27 with rosiglitazone and glimepiride, though additional agents were permitted in order to achieve target HbA1c levels. The final study in this group was the natural history follow-up of the UKPDS cohort, 10 years after randomization had ceased.
Late intensive glycemic control: conclusions from ACCORD, ADVANCE and VADT
Broad characteristics of the study populations (see Table 5.2 for more details)
- Mean age: 60–66 years.
- Mean known diabetes duration at study entry: 8–11.5 years.
- Mean HbA1c at baseline: 7.2–9.4% (55–79 mmol/mol).
- Difference in mean HbA1c at end of trial: 0.7–1.5% (8–17 mmol/mol)
- Mean duration of follow-up: 3.5–5.6 years.
From year 1 onwards in ACCORD, all-cause mortality was significantly increased in the intensive treatment group (mean achieved HbA1c 6.4%, 46 mmol/mol); the glycemic arm of the trial was terminated prematurely, while the lipid and blood pressure randomizations in the same study continued (see Chapters 11 and 12). There has been much post-hoc analysis in an attempt to discover the reasons for, or characteristics of those with, the increased mortality in the intensively treated group. Severe symptomatic hypoglycemia (blood glucose < 2.8 mmol/L, 50 mg/dL) was an early suspect, as was the rate of glucose lowering, weight gain, and the treatments themselves. Later analyses exonerated these factors; severe hypoglycemia was associated with increased mortality in both treatment groups. To a certain extent counter-intuitively, those intensively treated patients with high baseline HbA1c (more than 8.5%) had increased mortality. It is not yet clear whether the patients in this group actually achieved target HbA1c on inevitably multiple medications, or whether they remained poorly controlled despite intensive treatment.
Regardless, it is difficult to translate into a practical approach, other than to urge caution when embarking on intensive treatment in poorly controlled patients. There is a hint, so far unexplained, that aspirin use and self-reported history of neuropathy may contribute weakly to mortality, but they may just be indicators of patients already at high cardiovascular risk .
There was no excess of cardiovascular deaths in the intensive arms of ADVANCE and VADT, though cardiovascular events were significantly lower in the ACCORD intensive group without previous cardiovascular disease and HbA1c below 8% (64 mmol/mol) at baseline, i.e. ‘earlier’ diabetes. Intensive glycemic control in VADT reduced events in those patients with less coronary calcification as compared with those who had heavier calcification, again suggesting that patients with less advanced diabetes might benefit from intensive glycemic control, but this is not a practical method to stratify risk and allocate treatment intensity. The situation is further complicated by the continuous gradation of risk across the whole range of coronary calcification scores .
Severe hypoglycemia as a risk factor of death continues to be a concern. In ADVANCE, such episodes were associated with a nearly three-fold increased risk of cardiovascular events and deaths, and with almost a doubling of the risk of microvascular events. However, serious non-vascular outcomes were also increased, suggesting that hypoglycemia may be marker of severe comorbidity, rather than a direct causal factor. Nonetheless, the message is clear .
While the focus of these trials has been on the primary outcome of cardiovascular events, they also investigated microvascular outcomes. Broadly, advanced microvascular end points were not reduced with intensive glycemic intervention, including the following.
- Retinopathy: clinically significant macular oedema, progression to proliferative retinopathy and ophthalmological procedures, e.g. vitrectomy.
- Renal: dialysis, transplantation and advanced renal impairment.
However, progression of albuminuria was slowed by intensive treatment in ADVANCE, and ACCORD Eye (see Chapter 9) found some slowing of progression of retinopathy; however, the retinopathy findings are inconsistent between these trials, and may relate to differential effects of intensive glycemic control on proliferative retinopathy versus maculopathy. There was a non-significant benefit of intensive treatment on autonomic neuropathy in VADT, and some measures of peripheral nerve function were improved in ACCORD. These mixed results are fascinating, but they do not help therapeutic decisions in the individual case, where the balance will have to be taken on possible benefits on microvascular complications, likely beneficial in younger patients, contrasting with the essentially negative findings on macrovascular outcomes, important in older people without microvascular complications .
UKPDS: long-term follow-up
The 10-year UKPDS  provided more optimistic data. Despite convergence of mean HbA1c levels at 8.5% (69 mmol/mol) immediately after the trial, falling in both groups to 7.7% (61 mmol/mol) 5 years later, there was a ‘legacy’ effect of good glycemic control, with continuing benefits for the previous intensive control groups, the effects on myocardial infarction and all-cause death being more marked in the intensively treated metformin group compared with the insulin/sulphonylurea group. However, in this long-term follow-up, microvascular complications were not reduced with metformin treatment. Squaring the results of the UKPDS and the three more recent studies has been the subject of extensive discussion, broadly summarized in Box 5.1.
For more information and to purchase this book, just follow this link:
David Levy, MD, FRCP, Consultant Physician, Gillian Hanson Centre, Whipps Cross University Hospital; Honorary Senior Lecturer
Queen Mary University of London London, UK
This edition first published 2011, © 2011 by David Levy. 1st edition 1998 (Greenwich Medical Media/Cambridge University Press) 2nd edition 2006 (Altman Publications)