Home / Resources / Clinical Gems / Practical Diabetes Care, 3rd Ed., Excerpt #9: Pharmacological Treatment of Hyperglycemia Part 2 of 2

Practical Diabetes Care, 3rd Ed., Excerpt #9: Pharmacological Treatment of Hyperglycemia Part 2 of 2

Mar 29, 2015

David Levy, MD, FRCP

Drugs acting on the incretin system (entero-insular axis) (Fig. 6.2)
The incretin effect


Bayliss and Starling inferred the existence of gut-derived substances affecting carbohydrate metabolism more than a century ago, but the incretin effect was not described until 1969: the observation that for a given achieved level of plasma glucose, oral glucose produced a higher insulin concentration than intravenous glucose (Fig. 6.3). In the same year, the term ‘entero-insular axis’ was coined — the broader concept of all stimuli coming from the small intestine and influencing release of islet hormones. GIP (now named glucose-dependent insulinotropic hormone) was the first incretin to be sequenced in 1970, and the main therapeutic target in diabetes, GLP-1, in 1983….

altThe incretin system is critical in stimulating postprandial insulin secretion in response to nutrients entering the gut; about 50% of secretion is thought to be mediated by incretins and therefore is at least as important as glucose itself. In type 2 diabetes, the incretin effect is reduced, through a combination of GLP-1 deficiency and defective GLP-1 signalling, and possibly reduced islet responsiveness to GIP (Fig. 6.3). However, there is currently nothing to suggest that abnormalities of incretins themselves contribute to type 2 diabetes [14]. GIP itself does not appear to be deficient, does not lower glucose levels and may even augment postprandial glucose excursions. However, its therapeutic potential continues to be studied, particularly in obesity, for example, using GIP receptor antagonists.

GLP-1 is clearly deficient in type 2 diabetes. It is secreted from L cells situated mostly in the ileum and colon. In non-diabetic individuals, plasma GLP-1 levels increase within 10–15 min of starting to eat, and remain elevated for several hours, due to different populations of L cells being sequentially stimulated [15]. Native GLP-1, which has a circulating half-life of only 1–2 min, cannot be used clinically, though continuous intravenous infusions have been used to beneficial effect in pilot studies in myocardial infarction and heart failure, where GLP-1 may increase myocardial glucose uptake and be anti-apoptotic for cardiomyocytes.

The dominant effect of GLP-1 is therefore on postprandial glucose levels, and two additional mechanisms are important here: (i) decreasing postprandial glucagon secretion (an important factor in type 2 diabetes) and (ii) slowing gastric emptying. In addition, it decreases food intake and weight through a direct hypothalamic effect. Established and more speculative aspects of GLP-1 action are shown in Fig. 6.2.

Two strategies have been used pharmacologically to prolong the therapeutic effects of GLP-1:

  • development of GLP-1 analogues/receptor agonists (incretin mimetics) that directly act at the beta cell;
  • development of drugs (incretin enhancers) that boost endogenous GLP-1 by inhibiting the enzyme DPP-4, which rapidly degrades GLP-1 in plasma.

GLP-1 analogues, like insulin analogues, are GLP-1 molecules modified in various ways, but primarily to resist degradation by DPP-4. Since these are all peptide molecules, they require subcutaneous injection. DPP-4 inhibitors, not themselves peptides, are orally active.

GLP-1 receptor agonists and GLP-1 analogues
(BNF, section
GLP-1 receptor agonist: exenatide (synthetic exendin-4)

This remarkable substance was discovered in a trawl for bioactive agents in reptiles. Exendin-4 occurs in the saliva and venom of the rare binge-eating gila lizard (Heloderma suspectum) and beaded lizard (H. horridum) of the southwestern USA, though its unclear role in reptile physiology is not the same as in mammals. In its modified synthetic form, exenatide, it was introduced therapeutically in the USA in 2005. Exenatide is detectable in plasma up to 10 hours after injection, and therefore requires twice-daily administration (Box 6.2). HbA1c reductions of about 1%, sustained up to 3 years, are reported, but in practice individual glycemic changes are highly variable. Weight loss in trials is progressive and meaningful (about 2 kg at 6 months increasing to about 5 kg at 3 years), but like glycemic improvement is highly variable in practice, and there is a weak relationship between weight loss and change in HbA1c. Cases of good glycemic results with virtually no weight change, and vice versa, are frequently seen (approximate rates of response: weight loss and glycemic improvement, 72%; weight loss and glycemic deterioration, 9%; glycemic improvement and weight loss, 9%; glycemic improvement and weight gain, 14%). In RCTs, HbA1c reaches a nadir at about 12 weeks, while weight loss continues for much longer. Consequently, in individual patients tolerating exenatide treatment, a clinical trial of about 4 months should be offered before discontinuation. However, this decision can be difficult; some patients report consistently decreased satiety, while objectively showing minor changes in both weight and glycemic control. Minor improvements in blood pressure, lipids (especially HDL- cholesterol) and liver function tests have been seen, probably mostly due to weight loss.

Adverse Effects

  • Nausea, usually described as ‘mild to moderate’ occurs in most patients (~60%) for a short time after each injection, over the first 8 weeks of treatment, declining thereafter. Weight loss with exenatide is independent of gastrointestinal side-effects. Vomiting is described in 17% of patients; rarely it is severe and recurrent, and very rarely it can destabilize diabetes control sufficiently to require admission. Advise patients to stop exenatide immediately if there is any vomiting. In trials, about 5% of patients discontinue treatment because of gastrointestinal side-effects. Avoid in patients with known gastroparesis, and use with great care in patients with advanced peripheral neuropathy (e.g. foot ulceration), where clinically inapparent gastroparesis may be uncovered.
  • Hypoglycemia. Increased in patients taking metformin/sulphonylurea or sulphonylurea; reduce sulfonylurea dose.
  • Acute pancreatitis. About 40–50 cases have been reported so far. Although a rare complication, and a causal link has not been confirmed, patients with a past history of pancreatitis, or risk factors for it (e.g. severe hypertriglyceridemia), should not have exenatide treatment.

Exenatide vs. insulin in triple therapy with metformin and sulphonylurea
In trials comparing exenatide and insulin (usually basal glargine) added to metformin plus sulphonylurea, changes in glycemia are similar, though some studies have been criticized for not optimizing insulin doses. Patients in very poor control on metformin plus sulphonylurea (HbA1c about 10%, 86 mmol/mol) have a better glycemic response to even low-dose biphasic insulin (12 units twice daily) than to twice-daily exenatide 10 µg over 6 months. A contentious study found that replacing insulin treatment with exenatide prevented significant deterioration in glycemic control, but this was also criticized for its ‘futile’ attempt to replace an established treatment with a new one [16]. NICE (2009) strongly supports insulin treatment in patients failing on dual therapy, rather than adding a third drug, ‘unless there is strong justification not to’. However, the benefits of GLP-1 treatment over insulin (weight loss, reduced clinic time because of less intensive education and simple dosage titration, and lower risk of hypoglycemia) mean that outside the RCT setting, in the overweight or obese patient failing on dual oral therapy, a trial of GLP-1 treatment with careful clinical supervision should often be considered, with insulin a pre-discussed alternative if unsuccessful. Local practice is likely to vary on this difficult question; expertise of the team, close monitoring, and intensive support and education are probably as important as the specific pharmacological approach adopted.

Combined insulin and exenatide treatment
This unlicensed combination is in widespread use where additional oral agents have failed, and glycemic control and weight are deteriorating despite large daily doses of insulin. Prospective studies are needed, but in a retrospective 2-year study in patients taking a mean total daily insulin dose of 100 units, adding exenatide reduced Hb1c by 0.5%. Prandial insulin doses fell by more than 50%, although there was no change in basal insulin doses. Nadir weight (6-kg loss) occurred at 18 months, climbing again in the last 6 months [17]. Another option in this situation, addition of pioglitazone to insulin, gives similar glycemic results but weight increases.

GLP-1 analogue: liraglutide
Liraglutide is a GLP-1 analogue with high homology to human GLP-1. It gained a European licence in 2009, and was approved in the USA in 2010, but with a caution relating to a possible increased risk of medullary thyroid cancer in non-human studies. The molecule is linked to a fatty acid–albumin complex, similar to the long-acting insulin analogue detemir, adding to its DPP-4 resistance and conferring 24-hour duration of action (Box 6.3).

There were multiple Phase III trials involving liraglutide (LEAD studies). In summary it is more effective than maximum-dose glimepiride (8 mg daily) in monotherapy, more effective than rosiglitazone 4 mg daily when added to glimepiride 2–4 mg daily, and improved HbA1c by 1% (11 mmol/mol) when added to metformin 2 g daily and rosiglitazone 4 mg daily.

A 6-month head-to-head study comparing exenatide 10 µg twice daily with maximum-dose liraglutide 1.8 mg daily found slightly better glycemic control with liraglutide (treatment difference 0.3% HbA1c), but similar weight loss (~ 3 kg). Nausea with liraglutide occurred for a shorter time, 2–4% of patients reporting nausea by 16 weeks compared with 14% at onset of treatment. There was no difference in major hypoglycemic events. The two agents are therefore similar in clinical practice, though the clinical impression is that liraglutide is somewhat better tolerated. Cases of acute pancreatitis have been linked to liraglutide use.

Other GLP-1 receptor agonists and analogues
Several agents are in various stages of development.

  • Albiglutide, a GLP-1 dimer fused to human albumin, with a very long half-life suitable for weekly administration
  • Lixisenatide, a modified exendin-4 molecule (daily)
  • Taspoglutide (weekly; rare hypersensitivity reactions have been reported)
  • Semaglutide (weekly)
  • Nasally administered form of exenatide.

DPP-4 inhibitors (gliptins) (BNF, section
These are once- or twice-daily orally active drugs which are weight neutral, with overall similar or slightly less powerful glycemic effect compared with other medications for type 2 diabetes. Like the injectable GLP-1 analogues, they are glucose-sensitive and do not cause hypoglycemia when used as monotherapy or with agents that themselves do not cause hypoglycemia. Although it is assumed they act by increasing endogenous GLP-1 levels by inhibiting its breakdown, GLP-1 levels rise only modestly with DPP-4 treatment, and other mechanisms may be relevant, for example improved GLP-1 signalling and insulin processing, and effects on other neuropeptides.

RCTs have involved patients in good-to-fair glycemic control (typically HbA1c 7.5–8.0%, 59–64 mmol/mol) and show a fall in HbA1c of 0.5–1.0% (5–9 mmol/mol). However, like all agents the glycemic effect is proportional to baseline HbA1c and a mean decrease of 2% has been demonstrated in people with poor initial control (HbA1c ~ 9%, 75 mmol/mol). They are generally well tolerated, with no specific side-effects, other than occasional mild nausea. Pancreatitis has been reported with sitagliptin, but there is no confirmed association.

Three agents, sitagliptin, vildagliptin and saxagliptin, are currently licensed in the UK, with several more in late-stage development. Sitagliptin has the broadest licensed indications.

  • Monotherapy in the rare situation where metformin is not tolerated or contraindicated.
  • Dual therapy with metformin, or a sulphonylurea, or pioglitazone.
  • Triple oral therapy: sulfonylurea plus metformin even in moderate renal impairment (eGFR > 50 mL/min); pioglitazone plus metformin.
  • Added to insulin with or without metformin: HbA1c reduction 0.6% from baseline 8.6% (70 mmol/mol), even in long-duration diabetes.

Vildagliptin and saxagliptin are licensed for use only in dual therapy with metformin, a sulphonylurea or a glitazone. Vildagliptin and sitagliptin can be used in moderate renal impairment (GFR > 50 mL/min). Vildagliptin has been rarely associated with liver dysfunction, and baseline alanine aminotransferase should be less than three times the upper limit of normal; 3-monthly liver function tests are suggested during the first year of treatment, annually thereafter.

The dosages of these agents are as follows.

  • Sitagliptin: 100 mg daily as monotherapy; in fixed-dose combination with metformin (1000 mg), 50 mg twice daily.
  • Vildagliptin: 50 mg b.d. (fixed-dose combinations with metformin 850 mg and 1 g are available).
  • Saxagliptin: 5 mg daily.

The practical use of these agents within license, apart from sitagliptin, is currently limited, but the low risk of hypoglycemia (except when combined with insulin) makes them attractive alternative second-line agents either when a sulphonylurea has caused hypoglycemia or in others at high risk of hypoglycemia, for example the elderly. Although well tolerated, they are expensive and need frequent review for efficacy.

Pramlintide is synthetic amylin, a beta-cell hormone co-secreted with insulin. It is therefore relatively lacking in type 2 diabetes, and absent in type 1. Since it is derived from the pancreas and not the gut, it is not a true incretin, and has no significant direct pancreatic effects. For such an abundant hormone, little is known of amylin’s true physiological functions, but pramlintide has been licensed in the USA for use in type 1 and insulin-treated type 2 diabetes. Like the GLP-1 analogues, it slows gastric emptying, suppresses postprandial glucagon and glucose levels and increases satiety, but has no effects on peripheral insulin action. In intensively treated type 1 patients, pramlintide 15–60 µg s.c. with meals reduces weight by about 2 kg and HbA1c by about 0.5%. Hypoglycemia, sometimes severe, can be avoided by reducing prandial insulin doses by 30–50%. In type 2 diabetes, it is as effective in reducing HbA1c as titrated prandial rapid-acting insulin, and is weight neutral. Nausea is a frequent side-effect. Pramlintide is unlikely to be introduced in Europe.

Combination non-insulin treatment
Many bodies have issued guidelines and consensus documents on combination treatment in type 2 diabetes, but there is surprisingly little common ground among them, other than the initial steps of lifestyle intervention and metformin. The ADA/EASD consensus (2009) makes a valuable distinction between ‘well-validated core therapies’ and ‘less well-validated therapies’ (but unlike the NICE guidelines does not include many licensed combinations in this latter group). The number of combinations might therefore be considered conservative, DPP-4 inhibitors are not included, and exenatide appears in only one combination.

Well-validated core therapies (all pharmacological treatments are combined with intensive lifestyle intervention)

  • Metformin
  • Metformin plus basal insulin
  • Metformin plus sulphonylurea
  • Metformin plus intensive insulin.

Less well-validated therapies

  • Metformin plus pioglitazone
  • Metformin plus exenatide (insufficient evidence for safety)
  • Metformin plus sulphonylurea plus pioglitazone.

Insulin treatment in type 2 diabetes

Insulin has always been a mainstay of treatment for type 2 diabetes. Indeed, until the 1950s, it was the only treatment available and, until the mid-1990s, the only combination treatment available in the USA was insulin with a sulphonylurea. In these simple historical facts lie the problems of objectively assessing insulin treatment in type 2 diabetes, which naturally has not been subjected to the same comparative large-scale RCTs as more recently introduced agents (Box 6.4). The first comparison of insulin treatment against other therapies (other than the fraught University Group Diabetes Program in the early 1960s) was UKPDS, but this aspect of the trial was somewhat overlooked in comparison with the novel intensive versus conventional treatment strategy and its effect on complications. However, UKPDS indicated that insulin treatment:

  • was not more effective in glycemic control than intensive treatment with sulfonylureas or metformin;
  • did not confer microvascular or macrovascular advantages over sulphonylureas (there was anxiety at the time that hyperinsulinemia with insulin treatment might accelerate atheroma);
  • carried a consistently higher risk of any and major hypoglycemia than intensive sulphonylurea treatment (see www.dtu.ox.ac/ukpds for more information).

Since then, countless large-scale RCTs have compared different insulin regimens for short- and medium-term glycemic control. Many have explored insulin as an early intervention in type 2 diabetes, concluded that it is effective, safe and associated with no detrimental effects on quality-of-life measures, and in many instances seems to have improved it, and have increasingly recommended it in clinical practice. Most RCT protocols require frequent and intensive patient contact and education with algorithmically driven titration regimens and treat-to-target objectives; the result can be long-term glycemic stability, in contrast with, for example, the near-continual upward glycemic drift seen in all treatment arms of the UKPDS, which probably more closely resembles what happens in real life.

Mandatory insulin in type 2 diabetes
While much of the discussion and all the trials of insulin treatment in type 2 diabetes relate to overweight or obese subjects failing on routine combined oral hypoglycemic agents, the relatively small proportion, but large absolute number, of patients with late-onset type 1 diabetes must be recognized (see Chapter 1). While insulin deficiency is rarely complete, and ketonuria therefore uncommon, the persistent presence of urinary ketones should be a warning that insulin treatment may be needed; occasionally patients will be ketonuric during an intercurrent illness that uncovers insulin deficiency. Remember to be alert to the need at some stage, sometimes urgent, for insulin treatment in:

  • normal-weight or thin patients, and those with progressive weight loss and persistently poor control on non-insulin agents;
  • GAD antibody-positive people;
  • those with a personal or first-degree family history of other autoimmune disorders;
  • subjects who respond poorly to oral hypoglycemic agents, especially sulphonylureas;
  • the uncommon patient with proximal femoral neuropathy (neuropathic cachexia; see Chapter 10).

Insulin regimens in these patients should be the same as those in type 1 diabetes with the same intensity and targets. Discontinue sulphonylureas; metformin may still be of value in normal-weight patients. The remaining discussion relates to the much more contentious question of insulin treatment in overweight or obese patients.

Glycemic effects and limitations of insulin treatment
‘The potential for glucose lowering with insulin is unlimited’ [18]. Statements like this are common, and continue to be made even after the recent trials implicating hypoglycemia as a serious adverse prognostic factor, and suggesting that HbA1c levels below 7.0% do not meaningfully improve the vascular prognosis in many patients. Nevertheless, it is still widely believed that insulin is a more potent and more effective treatment than other agents.

However, there is little contemporary evidence to support this notion, when insulin (usually with metformin) is compared with combination treatment. In a 6-month trial comparing twice-daily biphasic insulin and metformin with triple oral therapy (glitazone, sulphonylurea and metformin), both reduced HbA1c eventually by 2.1% (9.7% to 7.6%, 83 to 60 mmol/L), though the initial fall was quicker with insulin. This phenomenon is of no long-term importance, but the very rapid fall in glucose levels often seen in the early stages of insulin treatment may be one factor contributing to an impression that insulin treatment is more effective (compare sulphonylurea treatment in newly diagnosed symptomatic patients).

Only in very few studies, whatever the design, are HbA1c levels below 7.0% (53 mmol/mol) consistently achieved, never mind near- normoglycemia.

  • In the ATLANTUS study (2005), HbA1c fell from 8.9 to 7.8% (74 to 62 mmol/mol) over 6 months with basal glargine at a mean dose of 43 units [19].
  • More intensive treatment in the 3-year 4-T Study (2009) resulted in a fall in HbA1c of about 1.5%, from a baseline of 8.5% to about 7.0% (69 to 53 mmol/mol), with well-maintained stable control (see below).
  • Basal glargine at a lower mean dose (24 units) and liraglutide 1.8 mg daily both reduced HbA1c by about 1.1–1.3% (11–15 mmol/mol) to 7.0– 7.2% (53–55 mol/mol; slightly lower with liraglutide); hypoglycemia rates were similar, but there was a 3.4-kg weight difference between the two groups [20].

Of the recent major outcome studies, only BARI 2D can guide long-term practice, though only in the broadest sense, and patients were in fairly good baseline control (mean HbA1c 7.7%, 61 mmol/mol). After 3 years, the insulin provision group (insulin and sulphonylureas) had significantly higher HbA1c (mean 7.5%, 59 mmol/mol) than those treated with the insulin-sensitizing regimen of metformin and rosiglitazone (7.0%, 53 mmol/mol) and had more severe hypoglycemia (9% vs. 6%). From this admittedly non-random selection of studies, it is difficult to conclude that insulin treatment, however intensive, gives consistently better glycemic control than judiciously chosen regimens using other agents (Box 6.5).

Early versus late treatment with insulin
Insulin improves glycemia at any stage of type 2 diabetes. The rationale behind its early use is that it preserves beta-cell function better than other agents, but there are no long-term data to support this. It is simpler to achieve good glycemia in the early stages of diabetes with insulin, but this goes for other treatments; in the LANCET study, insulin and metformin were equally successful in reducing HbA1c from 6.9 to 6.1% (52 to 43 mmol/mol) [21]. The rationale for using it later is that it is the logical agent in patients with depleted beta-cell function, though if this were the only mechanism operating it should be simple to achieve the kind of glycemia we target in type 1 diabetes. Some studies show improved quality of life with insulin therapy (usually together with metformin) compared with multiple non-insulin agents, but outside the clinical trial setting, individualization of treatment, often involving changes of therapy where necessary, is the best option. Insulin does not improve microvascular or macrovascular outcomes compared with other regimens of approximately similar glycemic effectiveness, whether used early (e.g. UKPDS) or late (e.g. VADT, BARI 2D). Cardiovascular risk factors, including inflammatory markers (e.g. high-sensitivity CRP), do not significantly improve with insulin (LANCET study), in contrast to some studies with metformin treatment (or, as in the Diabetes Prevention Program, intensive lifestyle intervention).

Practical insulin regimens
While there is still controversy surrounding the indications for insulin treatment in patients with type 2 diabetes, earlier studies progressively supported the use of basal insulin with continuing oral hypoglycemics as the most effective initial insulin regimen. The most recent study is the long-term follow-up of the 4-T, in typical overweight patients (e.g. BMI 30, weight 85 kg) failing on dual sulphonylurea and metformin treatment (mean HbA1c 8.5%, 69 mmol/mol) after 10 years of diagnosed diabetes [22]. The trial randomized patients to twice-daily biphasic insulin, basal bedtime insulin or prandial rapid-acting insulin three times daily with meals, targeting HbA1c 6.5% (48 mmol/mol) or less (unfortunately a full MDI regimen was not included at the start, but many starting with basal insulin progressed to MDI; see below). Metformin was continued, but the sulphonylurea discontinued if HbA1c was consistently 8% or more within the first year, or above 6.5% after the first year, and replaced with:

  • lunchtime prandial insulin 4–6 units in patients taking twice-daily biphasic insulin;
  • mealtime insulin (X3) 4–6 units in those taking basal insulin;
  • basal insulin 10 units was added in those taking prandial insulin. Basal insulin treatment carried the lowest risk of hypoglycemia and the least weight gain. This study confirms others (e.g. the APOLLO study [23]) that basal insulin with oral hypoglycemic agents is an effective and simple starting insulin regimen (Fig. 6.4; Boxes 6.6 and 6.7), though in a shorter 6-month study (INITIATE, 2007) biphasic aspart 30/70 was more effective than basal glargine.

Beyond basal insulin: complex regimens
Even when basal insulin treatment has been maximized to achieve good fasting levels, HbA1c remains very poor (e.g. >9%, 75 mmol/mol) in a substantial proportion. Most patients (80% in 4-T) required additional insulin, effectively converting them to MDI, 60% of the total insulin dose being prandial. In general, the more intensive the insulin regimen, the greater the weight gain and risk of significant hypoglycemia.

Over 6 months, HbA1c fell by about 2% (22 mmol/mol) from 8.9% (74 mmol/mol) using either three times daily prandial insulin added to the basal glargine or conversion to three times daily prandial 50/50 biphasic insulin; in practice, dose titration might not be rigorous enough to achieve this, and an already obese American study population gained a further 4 kg [25]. Several possible strategies are therefore valuable after establishing an optimum basal insulin regimen with oral hypoglycemic agents (Fig. 6.4). No standard regimen can be stipulated, but consider the following, and individualize the approach.

  • Sequentially add prandial insulin to one meal at a time, starting with the most carbohydrate-rich meal, and thereafter to the other main meals. Some patients may not eat breakfast, and long-acting analogue bedtime insulin may maintain satisfactory control until lunch, thereby limiting the number of prandial insulin injections.
  • Move to twice-daily prandial biphasic insulin (e.g. Humulin M3, NovoMix 30, Insuman Comb 25); in some short-term studies, this is more effective than basal insulin.
  • Move to three times daily prandial biphasic insulin (e.g. Humalog Mix 50), changing the dinner-time dose to a lower mix (e.g. Humalog Mix 25) if more intermediate-acting insulin is needed to achieve good fasting levels. This regimen is as effective as a full basal-bolus regimen.
  • Add twice-daily exenatide (unlicensed and no RCTs reported). However, the durability of these regimens is not assured, and results of long-term outcome studies are badly needed, for example the 2-year DURABLE study.

Beyond complex regimens: the patient poorly controlled on multiple daily insulin doses plus oral hypoglycemic agents
Regrettably there are no RCTs to help guide practice in this common, difficult and distressing clinical problem, the frequency of which could appear to the sceptical as undermining the concept of the limitless impact of insulin. The syndrome is characterized by:

  • glycemic control that has not improved with insulin, or improved transiently but returned to pre-insulin treatment levels or worse;
  • high total daily insulin doses, often greater than 1 unit/kg body weight or more than 100 units;
  • increasing weight, with associated worsening insulin resistance features (rising blood pressure, deteriorating lipid profile).

There are no simple explanations or easy solutions, but consider the following (Fig. 6.4).

  1. Taking into account the results of the recent glycemia trials, the duration of diabetes and complications, would focusing on more achievable, non-glycemic targets be a better use of time and resources?
  2. Modifying oral agents, for example:
        a. Add metformin up to 1 g b.d., though most patients will already be taking it. The results are variable and not as striking in practice as in clinical trials.
        b. Give a trial of modified-release metformin in patients previously intolerant of immediate-release metformin.
        c. Some patients may respond to reinstating a discontinued sulphonylurea.
        d. Add sitagliptin 100 mg daily (licensed) or vildagliptin 50 mg b.d. (unlicensed), which may modestly lower HbA1c (0.5–0.7%, 6–8 mmol/mol) over 6 months when given to patients poorly controlled on high doses of insulin.
      e. Add pioglitazone 15 mg daily, increasing very gradually up to 45 mg if beneficial and tolerated.
  3. Add exenatide or liraglutide (unlicensed).
  4. Bariatric surgery in the obese or very obese patient on multiple ineffective medications with or without late complications.

Bariatric surgery
Bariatric surgery (Greek baros meaning ‘weight’) is the term used for all surgical procedures to reduce weight. Restrictive procedures (e.g. laparoscopic adjustable gastric banding) markedly reduce stomach volume and decrease food intake. The most commonly used and most effective routine bariatric procedure, the Roux-en-Y gastric bypass (Fig. 6.5), reduces stomach volume and causes some degree of malabsorption. Roux-en-Y bypass causes greater weight loss than restrictive procedures (around 60% of excess weight, 10–15 BMI units, or 30–50 kg). Following any procedure there is some rebound weight gain over the years, but this is minor in relation to the overall weight loss. Diabetes resolves (i.e. normoglycemia with no requirement for diabetes medication) in about 80% after Roux-en-Y bypass, dyslipidemia improves in nearly all patients (though hypercholesterolemia itself may not improve), and hypertension in about three-quarters. Non-alcoholic fatty liver disease and sleep apnoea often dramatically improve. The diabetes is more likely to resolve in patients on tablets, as opposed to insulin, and those with a shorter duration of diabetes. Diabetes resolves within days of bypass procedures, well before weight loss occurs, but takes months to occur after restrictive surgery. GLP-1 levels substantially increase, and remain elevated for years after bypass, but other hormones influencing satiety and appetite (e.g. PYY) are probably involved. Alternatively, bypass of the proximal gut may exclude as yet unknown ‘anti-incretin’ or ‘decretin’ hormones.

Roux-en-Y bypass is usually performed laparoscopically, is safe (operative mortality about 0.5%), and NICE suggests considering it in diabetic patients with a BMI of 35–40. Careful postoperative management is important (Box 6.8). Ten years after surgery, the Swedish Obesity Study reported a 30% reduction in mortality, with substantial reductions in myocardial infarction and cancers [26], but as yet there are no systematic studies on microvascular complications. It would be surprising if there were no effects, even if the impact of near-normoglycemia might not be as dramatic as on resolution of hypertension and its associated features (e.g. sleep apnoea, microalbuminuria). Surgical techniques are continually being refined; for example, laparoscopic sleeve (subtotal) gastrectomy, which leaves a small tubular gastric remnant, is a relatively new restrictive procedure (with removal of the appetite-regulating ghrelin produced in the gastric fundus being a possible endocrine mechanism for weight loss). Although its place in the bariatric repertoire is not yet clear, it can be used as a single procedure in the moderately obese or as a bridging procedure in the high-risk severely obese patient, converting it later, after initial weight loss, to a Roux-en-Y bypass or a biliopancreatic diversion with duodenal switch.

New developments
While many new drugs are in early development, probably only one novel group of agents will be introduced over the next few years, the type 2 sodium glucose cotransporter (SGLT2) inhibitors. Inhibition of glucose reabsorption in the proximal renal tubules by the prototype clinical compound dapagliflozin is insulin independent and causes mild glycosuria and diuresis without hypoglycemia. In short-term studies, HbA1c reduction is moderate, up to 0.9% (10 mmol/mol), dose-dependent (2.5, 5 and 10 mg daily), and associated with weight loss up to 4 kg. Dapagliflozin is effective in poorly controlled patients taking combination insulin and oral hypoglycemic treatments. Similar results are seen in monotherapy, and patients in very poor control (HbA1c >9%, 75 mmol/mol) already with glycosuria had HbA1c reductions of nearly 2% (22 mmol/mol).

Genital infections are more common with high-dose treatment (20 mg vs. 10 mg). Several other SGLT2 inhibitors, for example sergliflozin, are in late-stage development. High-ratio biphasic insulins (e.g. containing up to 70% rapid-acting analogue) may be introduced and have a place in intensified insulin regimens, though long-term benefits may be limited. Insulin degludec, a very long-acting soluble insulin basal analogue, may be of value. Further down the line are G protein-coupled receptor 119 (GPR119) agonists, which seem to have a dual action on intestinal L cells to promote GLP-1 secretion and on the beta-cell. Experimentally, combination with DPP-4 inhibitor treatment increases their therapeutic effect.

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)