The Right Starting Point: Diagnosis First
Treatment in diabetes cannot be separated from type. The drugs that are essential for Type 1 are unnecessary for many people with Type 2. The drug class that works best for a patient with kidney disease may be contraindicated in someone with recurrent urinary tract infections. And as discussed in Post 2, a patient with LADA who is treated as Type 2 will be mismanaged for years with medications that can't compensate for ongoing autoimmune beta-cell destruction.
So the starting point is always: what type of diabetes does this person actually have? From there, treatment decisions layer on top of individual factors — cardiovascular risk, kidney function, weight, other medications, cost, and patient preference. What follows is a walk through the major drug classes and how they fit each type.
Type 1 Diabetes: Insulin Is Not Optional
In Type 1 diabetes, the immune system has destroyed the beta cells that produce insulin. There is nothing to enhance, sensitize, or stimulate — the insulin-producing machinery is gone. The only treatment that addresses the root problem is insulin replacement. This is not a preference or a last resort; it is a biological necessity. Without it, glucose cannot enter cells, ketones accumulate, and diabetic ketoacidosis (DKA) — a life-threatening emergency — follows within days.
Modern insulin therapy for Type 1 aims to mimic the normal pattern of insulin secretion as closely as possible: a steady background level throughout the day and night (basal insulin), with additional doses timed to meals (bolus insulin). This basal-bolus approach, combined with careful carbohydrate counting and glucose monitoring, allows people with Type 1 to achieve tight glucose control when the circumstances allow.
Adjunct therapy in Type 1
Insulin is the foundation, but several non-insulin agents can be added as adjuncts in selected patients with Type 1. SGLT2 inhibitors have shown modest A1c reductions and weight benefits in Type 1, though they come with an elevated risk of diabetic ketoacidosis — including at near-normal blood glucose levels (euglycemic DKA) — that requires careful patient selection and education. Pramlintide, a synthetic version of the hormone amylin (which is co-secreted with insulin and suppresses post-meal glucose spikes), is approved for Type 1 and reduces A1c and post-meal glucose. GLP-1 receptor agonists are not currently FDA-approved for Type 1 but are sometimes used off-label in overweight patients. The evidence base for all adjuncts in Type 1 is smaller than in Type 2, and their use requires specialist input.
For patients with Type 1, the single biggest advance in management over the last decade has not been a new drug — it's been continuous glucose monitoring (CGM) combined with automated insulin delivery systems (sometimes called closed-loop or artificial pancreas systems). These technologies dramatically reduce the cognitive burden of managing insulin doses and have improved time-in-range in clinical trials beyond what most patients can achieve with manual dosing. We cover these in Post 7. On the medication side, the adjuncts have a role in selected patients, but I approach them cautiously in Type 1 — the euglycemic DKA risk with SGLT2 inhibitors in particular is underappreciated and requires explicit patient education.
Type 2 Diabetes: A Ladder With Better Rungs Than It Used to Have
Type 2 treatment has been transformed over the last fifteen years. For decades, the standard approach was metformin first, then sulfonylureas, then eventually insulin. That sequence made sense when those were the only tools available. The evidence has shifted decisively since then. Two drug classes — GLP-1 receptor agonists and SGLT2 inhibitors — have demonstrated benefits that go far beyond glucose control: reductions in cardiovascular death, hospitalizations for heart failure, and kidney disease progression. When these medications are accessible, there is rarely a reason to lead with anything else.
The central problem with a glucose-only framework is that lowering an A1c number is not, by itself, the goal. The goal is living longer with fewer complications — better kidneys, a healthier heart, preserved quality of life. GLP-1 agonists and SGLT2 inhibitors deliver on those endpoints in ways that metformin, DPP-4 inhibitors, and sulfonylureas simply do not.
In my practice, if a patient with Type 2 diabetes can afford a GLP-1 receptor agonist or SGLT2 inhibitor, that is where I start — not metformin. The cardiac, renal, and mortality benefits are too important to defer. Metformin has its place when cost or insurance requirements make it the only realistic option, but I do not treat it as the obligatory first step it once was. I use metformin when I have to; I use GLP-1s and SGLT2 inhibitors when I can. I recommend against DPP-4 inhibitors entirely: they cost as much as SGLT2 inhibitors, produce the weakest glucose-lowering effect of any class, and offer no cardiovascular, renal, or mortality benefit. There is no situation where a DPP-4 inhibitor is the best available choice for a patient who can access alternatives.
GLP-1 receptor agonists: the most consequential advance in Type 2 treatment
GLP-1 receptor agonists work by mimicking glucagon-like peptide-1, a hormone released from the gut after meals that stimulates insulin secretion, suppresses glucagon, slows gastric emptying, and reduces appetite. The result is lower post-meal glucose, reduced overall food intake, and in most patients, meaningful weight loss. The class includes both older agents — exenatide, liraglutide, dulaglutide — and the newer, more potent drugs that have largely replaced them: semaglutide (Ozempic, Wegovy) and tirzepatide (Mounjaro, Zepbound), the latter of which also acts on GIP receptors and represents a dual agonist with superior efficacy. Oral semaglutide (Rybelsus) is also available, though it is meaningfully less effective than injectable formulations and generally not my preference when the injectable is accessible. Retatrutide — a triple agonist acting on GLP-1, GIP, and glucagon receptors — is in late-stage trials and expected to be available later this year, with early data showing even greater weight and glucose effects than tirzepatide.
We've covered the full mechanism and pharmacology of GLP-1 receptor agonists in our dedicated GLP-1 Receptor Agonist Series, which goes into considerably more depth on how these drugs work, their side effect profile, and the evidence across indications. What follows are the diabetes-specific results most relevant to understanding why this drug class belongs at the center of Type 2 treatment.
In my practice, I now exclusively prescribe semaglutide or tirzepatide — I no longer use the older GLP-1 agents (exenatide, liraglutide, dulaglutide). The reasons are straightforward: costs across the class are roughly comparable, yet semaglutide and tirzepatide are substantially more effective on A1c reduction, weight loss, and cardiovascular outcomes. There is simply no clinical reason to use a less effective drug at the same price. When patients ask about oral GLP-1 therapy, I explain that oral semaglutide is a legitimate option for those who genuinely cannot tolerate injections, but the injectable formulations achieve meaningfully better results and that tradeoff is worth having an honest conversation about.
One underappreciated benefit of GLP-1 receptor agonists in Type 2 diabetes is their effect on beta-cell function — the very cells that progressive Type 2 destroys over time. A network meta-analysis of 360 randomized trials including over 157,000 patients found that GLP-1 agonists significantly improve markers of beta-cell function (HOMA-β improvement of +20.3 units) and reduce insulin resistance (HOMA-IR reduction of −0.67) — improvements superior to those seen with DPP-4 inhibitors. Mechanistic studies show that GLP-1 agonists protect beta cells from the programmed cell death (apoptosis) that drives their progressive loss, and may stimulate beta-cell survival pathways. Direct evidence for beta-cell mass expansion in living humans remains elusive — we can't yet measure it reliably — but the functional improvements are consistent and clinically meaningful. This is one reason I think of GLP-1 agonists as potentially disease-modifying in Type 2, not just symptom-managing.
GLP-1 agonists are also a genuinely useful all-rounder from a kidney standpoint: unlike SGLT2 inhibitors, they can be used at any degree of kidney function — there is no eGFR threshold below which they become problematic for use, making them valuable for patients with advanced kidney disease who may not tolerate other agents.
I also frequently combine GLP-1 agonists or SGLT2 inhibitors with insulin in patients who require insulin therapy. Insulin's main drawback is weight gain; adding a GLP-1 agonist alongside insulin substantially mitigates this, and in many patients allows insulin doses to be reduced over time as metabolic control improves.
A1c reduction: GLP-1 receptor agonists reduce A1c by approximately 1.0–2.0 percentage points across trials. Tirzepatide (dual GIP/GLP-1) achieves reductions of up to 2.0–2.3 percentage points in the SURPASS program — among the largest reductions seen with any non-insulin agent, and superior to semaglutide 1mg in head-to-head comparison.
Cardiovascular outcomes: GLP-1 agonists reduce cardiovascular mortality (OR 0.87), all-cause mortality (OR 0.88), and stroke (OR 0.87) with high certainty in network meta-analysis. The landmark trial-level data — LEADER (liraglutide), SUSTAIN-6 (semaglutide), REWIND (dulaglutide) — were conducted with older or lower-dose agents; LEADER reduced the combined endpoint of cardiovascular death, heart attack, and stroke from 14.9% to 13.0% over ~3.8 years (ARR 1.9%, NNT ~52), and SUSTAIN-6 reduced it from 8.9% to 6.6% (ARR 2.3%, NNT ~43). The SELECT trial (semaglutide 2.4mg — the current standard dose) showed a 20% reduction in major cardiovascular events in patients without diabetes, suggesting the benefits extend with higher doses. In a 1.5 million patient head-to-head analysis (LEGEND-T2DM), GLP-1 agonists and SGLT2 inhibitors showed comparable cardiovascular effectiveness — both substantially superior to DPP-4 inhibitors (11–17% lower risk) and sulfonylureas (24–28% lower risk).
Kidney outcomes: The FLOW trial (semaglutide 1mg) — the first dedicated kidney outcomes trial for a GLP-1 agent — showed a 24% reduction in the composite kidney endpoint (absolute event rate 17.3% vs. 21.7% placebo; ARR 4.4%, NNT ~23) over ~3.4 years in patients with Type 2 and chronic kidney disease. Semaglutide now carries FDA approval specifically for reducing kidney disease progression. Importantly, GLP-1 agonists can be used at any level of kidney function — no eGFR cutoff applies.
Beta-cell preservation: GLP-1 agonists improve beta-cell function markers significantly versus placebo and DPP-4 inhibitors in meta-analysis (HOMA-β +20.3, fasting C-peptide +0.16 ng/mL). Mechanistic data supports anti-apoptotic effects on beta cells. Human evidence for actual beta-cell mass expansion is not yet established, but functional preservation is consistent across studies.
Weight: Average weight loss of 5–15% of body weight in clinical trials, with tirzepatide at the higher end (~5 kg average for semaglutide in diabetes trials). When combined with insulin, GLP-1 agonists substantially reduce insulin-associated weight gain and often allow dose reduction.
Insulin sensitivity: Independent of weight loss, GLP-1 agonists improve hepatic insulin sensitivity and reduce the insulin resistance that drives Type 2 progression.
Cancer signal: Observational data and mechanistic evidence suggest potential reductions in certain obesity-related cancers (colorectal, endometrial, pancreatic signals have been reported). These findings are preliminary and not yet sufficient to drive prescribing decisions, but they are being actively studied in prospective trials.
Hypoglycemia risk: Low — GLP-1 agonists stimulate insulin secretion in a glucose-dependent manner, meaning they reduce their effect as blood sugar normalizes. This makes hypoglycemia uncommon when used without insulin or sulfonylureas.
SGLT2 inhibitors: organ protection independent of insulin
SGLT2 inhibitors — which include empagliflozin (Jardiance), dapagliflozin (Farxiga), and canagliflozin (Invokana) — work by blocking a transporter in the kidney that normally reabsorbs glucose from urine back into the bloodstream. By blocking this reabsorption, they cause excess glucose to be excreted in urine, lowering blood sugar. The mechanism is completely independent of insulin, which is why SGLT2 inhibitors work regardless of how much insulin the pancreas is still producing — and why they are useful even late in the disease course.
Their glucose-lowering effect is modest — A1c reductions of approximately 0.5–1.0 percentage points. Their cardiovascular and kidney outcomes data is not modest at all. I think of SGLT2 inhibitors as having two particular superpowers: heart failure (where their benefits are among the strongest of any drug ever studied in that condition) and kidney protection (where they cut the rate of kidney function decline approximately in half).
Cardiovascular outcomes: SGLT2 inhibitors reduce cardiovascular mortality (OR 0.82) and all-cause mortality (OR 0.84) with moderate certainty, and reduce hospitalization for heart failure (OR 0.65) with high certainty. EMPA-REG OUTCOME (empagliflozin): cardiovascular death fell from 5.9% to 3.7% over ~3.1 years (ARR 2.2%, NNT ~46). Heart failure hospitalization fell from 2.7% to 1.6% — a 35% absolute reduction.
Heart failure: SGLT2 inhibitors are now approved for heart failure with both reduced and preserved ejection fraction, independent of diabetes status. This is one of their primary indications — the benefits here are among the strongest seen for any drug in this condition, and they now form part of standard heart failure therapy regardless of whether diabetes is present.
Kidney outcomes: This is where SGLT2 inhibitors truly stand apart. A 2026 meta-analysis of 10 randomized trials including 70,361 participants found that SGLT2 inhibitors reduce the annual rate of eGFR decline by approximately 51% — or about 1.26 mL/min/1.73m² per year in absolute terms. The risk of CKD progression (kidney failure, ≥50% eGFR decline, or kidney death) was reduced by 38% (HR 0.62), with consistent effects regardless of baseline eGFR or albuminuria level. CREDENCE (canagliflozin): 30% reduction in primary kidney endpoint (NNT ~6 over 2.6 years). DAPA-CKD (dapagliflozin): 39% reduction in the primary composite, 36% reduction in end-stage kidney disease specifically, including a 34% reduction in the need for long-term dialysis. The ADA and KDIGO now recommend SGLT2 inhibitors in most patients with Type 2 diabetes and chronic kidney disease with eGFR ≥20, independent of A1c or glucose-lowering need — the kidney protection is the indication, not the blood sugar effect.
Weight and blood pressure: Modest weight loss of 2–3 kg on average from caloric loss through glycosuria. Systolic blood pressure reductions of 3–4 mmHg. Both effects are independent of glucose lowering.
Hypoglycemia risk: Low — 40–51% lower than sulfonylureas in real-world data.
The chart below illustrates the eGFR trajectory with and without SGLT2 inhibitor therapy, based on meta-analytic data. The initial small dip after starting treatment is a well-characterized hemodynamic effect — reversible and not harmful — followed by a substantially slower rate of decline compared to placebo.
Illustrative based on Neuen et al., JAMA 2026 (N=70,361). Individual trial results vary.
High blood sugar at initiation: In patients with very elevated glucose (roughly 300 mg/dL or above), I generally prefer to bring levels down with other medications first before starting an SGLT2 inhibitor. When blood sugar is very high, the kidneys are already excreting large amounts of glucose — adding an SGLT2 inhibitor on top of that significantly increases the osmotic diuresis (water loss driven by glucose in the urine), raising the risk of dehydration and genital infections. Once glucose is better controlled, SGLT2 inhibitors can be added safely.
Genital infections: Mycotic infections (yeast infections) are the most common side effect, particularly in women, affecting roughly 10% of patients. Urinary tract infections are modestly increased. Rare but serious: Fournier's gangrene (necrotizing fasciitis of the genitalia) has been reported — seek immediate care for any genital pain, swelling, or redness.
Perioperative use: SGLT2 inhibitors should be held before any surgical procedure or prolonged fasting due to euglycemic DKA risk. For glucose-lowering purposes they are not recommended when eGFR falls below ~20 mL/min/1.73m², though cardiorenal benefits extend lower for some agents. Canagliflozin carries an additional signal for lower limb amputation risk and fractures not seen clearly with other agents in the class.
Metformin: useful when needed, not the universal starting point
Metformin works primarily by reducing the liver's glucose output between meals — the hepatic glucose production that runs unchecked when insulin signaling fails. It also modestly improves peripheral insulin sensitivity. It does not stimulate insulin secretion, so it does not cause hypoglycemia on its own. It is inexpensive, generally well-tolerated, and has decades of safety data. GI side effects (nausea, diarrhea) are common when starting — I prefer the extended-release formulation (metformin XR), which costs the same as immediate-release but significantly reduces GI side effects for most patients. Taking it with the largest meal of the day also helps. Long-term use is associated with mild vitamin B12 depletion, worth monitoring periodically.
Metformin's A1c reduction is approximately 1.0–2.0 percentage points — comparable to sulfonylureas on paper. Its cardiovascular and mortality profile is neutral, not beneficial. It does not reduce heart failure hospitalization, kidney disease progression, or all-cause mortality in the way GLP-1 agonists and SGLT2 inhibitors do. It remains a reasonable medication when cost or insurance requirements make it the practical choice, but it should not be treated as the obligatory first step when better options are available.
The other drug classes: an honest assessment
Sulfonylureas (glipizide, glimepiride, glyburide) directly stimulate insulin secretion from beta cells regardless of blood glucose level — which is why they cause hypoglycemia. They lower A1c by approximately 1.0–2.0 percentage points and are genuinely inexpensive, making them one of the most accessible options for cost-constrained patients. I use them regularly in practice when GLP-1 agonists and SGLT2 inhibitors are inaccessible, and they can be effective. Their limitations are real: hypoglycemia risk (particularly with glyburide, which is the least forgiving), weight gain of 2–5 kg, and accelerated beta-cell exhaustion over time. Their cardiovascular profile is the weakest of any commonly used class — real-world data consistently shows higher MACE rates with sulfonylureas than with newer agents. When cost allows a choice, they are not my first preference; when it doesn't, they are a legitimate and effective tool. Among the class, glimepiride is the preferred agent — it appears less cardiovascularly harmful than glyburide or glipizide, likely due to differences in how it interacts with cardiac muscle.
I do not recommend DPP-4 inhibitors (sitagliptin/Januvia, saxagliptin/Onglyza, linagliptin/Tradjenta). They work by blocking the enzyme that breaks down naturally produced GLP-1, producing a modest boost in insulin secretion after meals. They are weight-neutral and have a low hypoglycemia risk. But their efficacy data is the weakest of any major diabetes drug class — A1c reductions of approximately 0.5–1.0 percentage points, no reduction in cardiovascular mortality, myocardial infarction, stroke, or all-cause mortality in high-certainty evidence, and no meaningful kidney benefit. Saxagliptin was associated with increased heart failure hospitalization in one major trial. They cost as much as SGLT2 inhibitors while delivering none of the cardiovascular, renal, or mortality benefits. There is no clinical scenario where a DPP-4 inhibitor is the best available choice for a patient who can access a GLP-1 agonist or SGLT2 inhibitor — and if cost is the concern, sulfonylureas are far cheaper. DPP-4 inhibitors occupy an unfortunate middle ground: too expensive to justify over sulfonylureas on cost grounds, too ineffective to justify over GLP-1s and SGLT2 inhibitors on outcomes grounds.
Thiazolidinediones (pioglitazone) improve insulin sensitivity in muscle and fat tissue and reduce A1c by approximately 1.0–1.5 percentage points. They cause fluid retention, weight gain of 2–4 kg, increased fracture risk in women, and possible bladder cancer risk with long-term use. I use them rarely — primarily in patients who want to avoid injectable medications entirely and are willing to accept the risk profile, and where cost is a driver, as generic pioglitazone is inexpensive. They are not a first or second choice in my practice.
Alpha-glucosidase inhibitors (acarbose) slow carbohydrate absorption in the gut, modestly blunting post-meal glucose spikes. A1c reductions are small, GI side effects (bloating, flatulence) are common, and outcomes data is limited. I rarely if ever use them — there is nothing in their profile that makes them competitive with the available alternatives.
Two Myths Worth Naming Here
LADA: Why Standard Type 2 Treatment Falls Short
LADA is mechanistically autoimmune diabetes — the same beta-cell destruction as Type 1, progressing on a slower timeline. That distinction has direct treatment implications. Medications that work by stimulating insulin secretion from beta cells (sulfonylureas, DPP-4 inhibitors, GLP-1 agonists) are pushing harder on a system that is actively being destroyed. The effect diminishes as beta cells are lost, and the clinical picture looks like treatment failure when it is actually disease progression in the wrong framework.
There is growing evidence that early insulin use in LADA — rather than waiting for failure of oral medications — may preserve residual beta-cell function longer by reducing the metabolic stress on remaining cells. This is an area of active research, and the optimal approach is not yet fully settled. What is clear is that standard Type 2 escalation protocols do not serve LADA patients well, and that the diagnosis should prompt a different treatment conversation — ideally with an endocrinologist familiar with the nuance.
A 2019 meta-analysis found that sulfonylurea use in LADA patients was associated with faster progression to insulin dependence compared to insulin therapy initiated earlier, suggesting that forcing secretion from already-stressed beta cells accelerates their loss. Insulin therapy — particularly at lower doses preserving some beta-cell activity — is generally preferred for LADA over oral secretagogues. SGLT2 inhibitors and lifestyle modification may have a role, but the evidence base for LADA specifically is limited compared to Type 2.
A Note on Cost and Access
The medications with the strongest evidence — GLP-1 receptor agonists and SGLT2 inhibitors — are also among the most expensive. A month's supply of semaglutide (Ozempic) can exceed $900 without insurance coverage. SGLT2 inhibitors are more accessible but still carry significant cost for uninsured or underinsured patients. This is not a minor footnote: cost directly shapes what is possible for a given patient.
A few practical points worth knowing. Many manufacturers offer patient assistance programs for qualifying patients. Generic versions of older drug classes (metformin, glipizide, pioglitazone) remain effective and inexpensive options when cost is the dominant constraint. Compounded GLP-1 medications have been available during supply shortages and can reduce cost substantially — I use compounded semaglutide and tirzepatide through a vetted pharmacy for appropriate patients, with the same clinical monitoring I'd apply to brand-name agents. The goal is always to get patients on the most effective therapy their situation allows, not to default to the cheapest option when better is achievable.