22.8 — Diabetes Drugs

For most of the last century, treating type 2 diabetes meant one thing: lowering blood glucose. The assumption was that a lower number meant fewer complications, and that any drug that lowered it was doing its job.

Then a large trial in 2008 found a drug that lowered glucose effectively and increased heart failure. Regulators responded by requiring every new diabetes drug to prove it did not increase cardiovascular risk.

And the trials came back showing something nobody had asked for: several of the new drugs reduced heart attacks, strokes, heart failure admissions, kidney failure and death.

Which is why two classes of diabetes drug are now used in people who do not have diabetes at all.

Insulin

The oldest and the one that is never optional in type 1 diabetes (Chapter 18.7), because there the pancreas produces essentially none.

Insulin cannot be taken by mouth — it is a protein, and the digestive system would break it down like any other protein. Which is why it is injected, and why an oral insulin has been a goal for a century.

The types, organised by how fast and how long:

Rapid-acting — aspart, lispro, glulisine. Onset around 10 to 20 minutes, peak 1 to 3 hours, lasting 3 to 5 hours. Taken with meals.

Short-acting (regular human insulin) — onset 30 minutes, so it needs taking well before eating.

Intermediate-acting — NPH, lasting 12 to 18 hours with a definite peak.

Long-acting — glargine, detemir, degludec. Flat, essentially peakless, lasting 24 hours or more. These provide the background insulin the body normally secretes continuously.

Mixed preparations — a fixed combination, fewer injections, less flexibility.

The basal-bolus approach mimics normal physiology: a long-acting insulin once daily for the background, plus rapid-acting insulin with each meal, dosed according to the carbohydrate in that meal and the current glucose reading.

Insulin pumps deliver rapid-acting insulin continuously with boosts at meals, and closed-loop systems now link a pump to a continuous glucose monitor with an algorithm adjusting delivery automatically — an artificial pancreas in practical terms, and one of the genuine advances of the last decade.

Hypoglycaemia is the main hazard, and the practical points are in Chapter 23.11.

Other issues: weight gain, because insulin is a storage hormone; injection site lipohypertrophy, which causes erratic absorption and is prevented by rotating sites (Chapter 22.4); and the fact that insulin requirements change with illness, exercise, alcohol and stress.

And a specific safety point: insulin doses are written in units, and "U" written by hand has been misread as a zero, turning 4 units into 40. Units is always written out in full, which is one of those small conventions that exists because of deaths.

Metformin

The first-line drug for type 2 diabetes almost everywhere, and it has been in use since the 1950s.

It reduces glucose production by the liver, and improves the sensitivity of tissues to insulin.

What makes it the anchor drug:

It does not cause hypoglycaemia on its own, because it does not force insulin release — it reduces glucose output, which stops when glucose is already low.

It is weight-neutral or causes modest weight loss.

It is very cheap.

And it has a long track record with evidence of cardiovascular benefit.

Side effects: nausea, diarrhoea and abdominal discomfort, which affect a substantial minority. Starting at a low dose, increasing slowly, taking it with food, and switching to the modified-release form solve this for most people.

B12 deficiency with long-term use — worth checking, particularly in anyone developing neuropathy (Chapter 20.6).

Lactic acidosis — very rare, and the reason for the main precautions. Metformin is stopped in severe kidney impairment, in acute illness with dehydration, and temporarily around scans using contrast dye and around surgery.

Which is a sick day rule worth knowing: stop metformin during vomiting, diarrhoea or severe illness, and restart when eating and drinking normally (Chapter 21.3).

SGLT2 inhibitors

Dapagliflozin, empagliflozin, canagliflozin — the "-flozins".

And the mechanism is unusual enough to be worth explaining properly.

The kidney filters around 180 grams of glucose a day and normally reabsorbs essentially all of it (Chapter 10.2), using a transporter called SGLT2.

These drugs block that transporter, so glucose is lost in the urine — around 50 to 80 grams a day.

Which is treating diabetes by deliberately causing glycosuria, something that was for a century regarded purely as a sign of disease.

The consequences are more than glucose:

Weight loss, because those grams are calories leaving the body.

Blood pressure reduction, because glucose in the urine takes water with it.

And the cardiovascular and kidney benefits, which were the surprise. They reduce heart failure hospitalisation substantially, slow the progression of chronic kidney disease, and reduce cardiovascular death — in people with and without diabetes.

They are now standard treatment in heart failure and in chronic kidney disease independent of glucose (Chapters 22.7, 21.3), which is a rare thing for a drug to achieve.

Side effects: genital thrush, which is common and follows directly from sugar in the urine; urinary infections; and dehydration, particularly with diuretics.

And one specific hazard worth knowing: euglycaemic diabetic ketoacidosis. Ketoacidosis occurring with a glucose level that is normal or only mildly raised, which means the usual warning sign is absent. Risk rises during fasting, surgery, severe illness and very low carbohydrate diets. These drugs are stopped during acute illness and before surgery, which is another sick day rule.

GLP-1 receptor agonists

Semaglutide, liraglutide, dulaglutide, and the dual agonist tirzepatide.

GLP-1 is a hormone released by the gut when you eat (Chapter 12.5). It does four things: stimulates insulin release, suppresses glucagon, slows stomach emptying, and reduces appetite through the brain.

These drugs are longer-lasting versions of it.

And the insulin release is glucose-dependent — it only happens when glucose is raised — which is why they do not cause hypoglycaemia on their own.

They produce substantial weight loss, which is why they moved from diabetes into obesity treatment. Trials of semaglutide and tirzepatide have shown mean weight reductions in the range of 15 to 20 percent, which is well beyond anything previous drug treatment achieved.

And cardiovascular benefit is established for several of them.

Side effects: nausea, vomiting and constipation, which are the dose-limiting problem and are managed by slow escalation. Gallstones with rapid weight loss. Pancreatitis rarely. And they are avoided with a personal or family history of medullary thyroid cancer.

Given by weekly injection, and an oral semaglutide exists.

And a practical point people frequently get wrong: weight regain occurs when the drug is stopped, because the underlying appetite regulation returns to where it was. These are treatments for a chronic condition rather than a course of therapy, which changes how the decision to start should be framed.

The older classes

Sulfonylureas — gliclazide, glimepiride.

They force the pancreatic beta cells to release insulin, regardless of the glucose level.

Which is why they cause hypoglycaemia and weight gain — the two features that have moved them down the list.

They are cheap and effective, and they remain widely used where cost is the binding constraint. And they can be genuinely useful when a rapid glucose reduction is needed.

Caution in older people and in kidney impairment, where hypoglycaemia can be prolonged and severe.

DPP-4 inhibitors — sitagliptin, linagliptin.

They block the enzyme that breaks down GLP-1, so the body's own GLP-1 lasts longer. A gentler version of the same idea as the GLP-1 agonists.

Well tolerated, weight-neutral, no hypoglycaemia — and only modest glucose lowering and no cardiovascular benefit. Which is why they have been displaced somewhat.

Pioglitazone — improving insulin sensitivity. Effective, and causes weight gain, fluid retention and increased fracture risk, and is avoided in heart failure.

Acarbose — slowing carbohydrate digestion. Effective and produces considerable flatulence, which limits its use.

How treatment is chosen now

And the logic has changed from "how much does it lower glucose" to "what else does this person need".

Metformin first, unless contraindicated.

Then the choice depends on the person rather than only the glucose:

Established cardiovascular disease — an SGLT2 inhibitor or a GLP-1 agonist with proven benefit.

Heart failure — an SGLT2 inhibitor.

Chronic kidney disease — an SGLT2 inhibitor.

Obesity as a major issue — a GLP-1 agonist.

Cost the dominant constraint — a sulfonylurea.

Which means two people with the same HbA1c may correctly receive completely different drugs.

And targets are individualised too. A tight HbA1c target makes sense in a young person with decades ahead. In a frail older person, the risk of hypoglycaemia outweighs the benefit of a lower number, and a relaxed target is the right answer rather than a compromise.

The things that are not drugs

And they deserve to be stated in a pharmacology chapter, because they outperform much of it.

Type 2 diabetes can go into remission.

Substantial weight loss — through a structured low-calorie programme, or through bariatric surgery — produces remission in a meaningful proportion of people, particularly within the first few years of diagnosis. The DiRECT trial achieved remission in around 46 percent at one year with a total diet replacement programme.

Which reframes type 2 diabetes from a permanently progressive condition to one that is sometimes reversible, and that is a genuinely recent change in understanding.

Exercise improves insulin sensitivity independently of weight, and the effect on glucose uptake by muscle begins immediately and lasts for a day or more after a session.

And structured education programmes produce measurable improvements in control and in confidence, particularly in type 1 diabetes.

Monitoring

HbA1c — reflecting average glucose over roughly the previous 8 to 12 weeks, because glucose attaches to haemoglobin over the red cell's lifespan.

And it can mislead in conditions that change red cell survival — anaemia, haemoglobinopathies, recent transfusion — where it reads falsely low or high.

Continuous glucose monitoring — a sensor worn on the skin reading glucose in the fluid between cells every few minutes.

It has changed type 1 diabetes management more than any drug in decades, because it shows the pattern rather than isolated points, alarms before a low, and lets people see the effect of a meal or a walk directly.

Time in range — the proportion of the day spent within target — is increasingly used alongside HbA1c, because two people with the same average can have very different amounts of time spent dangerously high or low.

What the next page fixes

Chapter 22.9 covers gut drugs — acid suppression, laxatives, anti-sickness drugs, and the ones most commonly taken without much thought.