Skip to content

Type 2 diabetes — glycaemic management and legacy trials

TL;DR — UKPDS established that early intensive glycaemic control reduces microvascular complications and later revealed a durable “legacy effect”; 24-year monitoring still found lower mortality and myocardial infarction after early intensive assignment (UKPDS 1998, PMID 9742976; Adler 2024, PMID 38772405). ACCORD, ADVANCE and VADT showed that rapidly targeting near-normal HbA1c in older, longer-duration, high-risk populations does not reproduce the same cardiovascular benefit and can cause harm (ACCORD 2008, PMID 18539917; ADVANCE 2008, PMID 18539916; VADT 2009, PMID 19092145). GRADE found greater glycaemic durability with glargine and liraglutide than sitagliptin or glimepiride when added to metformin, but treatment choice now also turns on heart, kidney, weight, hypoglycaemia and cost outcomes (GRADE 2022, PMID 36129996).

Landmark trial comparison

Trial Disease context Glycaemic strategy Main lesson
UKPDS 33 Newly diagnosed T2D Sulfonylurea/insulin intensive vs conventional Microvascular benefit (PMID 9742976)
UKPDS 34 Overweight, newly diagnosed Metformin vs conventional Diabetes-related outcomes and later legacy benefit (PMID 9742977)
ACCORD Older/high CV risk, longer duration HbA1c <6.0% target Intensive arm stopped for excess mortality (PMID 18539917)
ADVANCE Established T2D/high risk Intensive gliclazide-based strategy Nephropathy benefit; no major macrovascular reduction (PMID 18539916)
VADT Long-duration, largely male veterans Intensive vs standard No significant primary CV benefit during trial (PMID 19092145)
GRADE Metformin-treated, <10 years duration Four second-line drugs Glargine/liraglutide more durable for HbA1c target (PMID 36129996)

UKPDS legacy

At trial end, intensive therapy reduced microvascular endpoints but cardiovascular differences were limited (UKPDS 1998, PMID 9742976). After ten years of post-trial follow-up, between-group HbA1c differences had disappeared while MI and mortality benefits emerged (Holman 2008, PMID 18784090). At 24 years, early intensive control retained lower all-cause mortality and MI, supporting durable biological or care-pathway memory (Adler 2024, PMID 38772405).

Follow-up Intensive-control legacy Metformin legacy
10-year post-trial MI −15%; all-cause death −13% Persistent outcome benefit (Holman 2008, PMID 18784090)
24-year monitoring All-cause death −10%; MI −17% Benefits did not wane (Adler 2024, PMID 38772405)

Why ACCORD differed

ACCORD participants were not newly diagnosed and had high cardiovascular risk; its intensive arm pursued rapid near-normalisation using multiple drugs and experienced more severe hypoglycaemia and weight gain. Excess mortality led to early termination (ACCORD 2008, PMID 18539917). The result argues against one universal HbA1c target, not against all glycaemic control.

Individualised targets

Factor Supports lower target Supports less stringent target
Age/life expectancy Younger/long horizon Limited life expectancy
Disease duration Early Long duration/low reserve
Complications Preventable early microvascular risk Advanced comorbidity/treatment hazard
Hypoglycaemia Low-risk regimen/awareness Severe events/unawareness
Treatment burden Simple, acceptable Complex, costly or distressing
Patient priorities Values prevention/low burden Values symptom avoidance/deprescribing

ADA/EASD frames targets and drug selection around person-level benefit, comorbidity and preference rather than a single sequence (Davies 2022, PMID 36148880).

GRADE

GRADE compared insulin glargine, glimepiride, liraglutide and sitagliptin added to metformin. Glargine and liraglutide maintained HbA1c <7% longer than the other options; severe hypoglycaemia was uncommon but more frequent with glimepiride, and liraglutide produced weight loss (GRADE 2022, PMID 36129996).

GRADE did not include SGLT2 inhibitors and was not powered like a dedicated cardiovascular/kidney outcome trial. Its central contribution is comparative glycaemic durability, not a universal ranking of organ protection.

Drug-class matrix

Class HbA1c Weight Hypoglycaemia alone Organ-outcome signal
Metformin Moderate Neutral/small loss Low Long UKPDS legacy in selected population
Sulfonylurea High initially Gain High No independent organ protection established
DPP-4 inhibitor Modest Neutral Low CV safety, neutral outcomes
Basal insulin Highest ceiling Gain Meaningful Required when deficiency/decompensation dominates
GLP-1RA High Loss Low MACE; molecule/population dependent
SGLT2 inhibitor Moderate; less at low eGFR Small loss Low HF and CKD protection
TZD High/durable Gain/oedema Low HF/fracture trade-offs

HbA1c is incomplete

HbA1c reflects average glycaemia but not variability, hypoglycaemia, treatment burden or organ protection. Anaemia, haemoglobin variants, altered red-cell lifespan, pregnancy and kidney disease can bias it. Discordance with glucose data should prompt investigation, not automatic intensification.

Glycaemic-pattern interpretation

Pattern Likely question Useful next information
High fasting, acceptable post-meal Overnight hepatic output/basal coverage Bedtime and waking glucose
Acceptable fasting, high HbA1c Postprandial excursions or discordant HbA1c Meal-linked glucose/CGM
Low HbA1c with symptoms Hypoglycaemia or red-cell bias CGM/SMBG, blood count, kidney function
Rapid unexplained deterioration Illness, medication, insulin deficiency, classification Symptoms, ketones, C-peptide/antibodies as indicated
High variability Timing, meals, insulin mismatch Structured glucose record

Intensification and deintensification

Situation Evidence-aligned direction
Catabolic symptoms/ketosis Start insulin promptly
CKD/HF despite target HbA1c Add organ-protective therapy independent of glucose
Recurrent hypoglycaemia Reduce insulin/secretagogue burden
Frailty/limited life expectancy Relax target and simplify
Major weight priority Select weight-effective therapy
Cost-related nonadherence Redesign regimen around affordable access

Deintensification is active risk management, not abandonment. The relevant outcome is net benefit: fewer severe lows, falls and dosing errors without symptomatic hyperglycaemia.

Insulin transition principles

Step Purpose
Identify fasting versus meal problem Avoid unnecessary regimen complexity
Start/titrate basal when indicated Correct fasting exposure
Reassess sulfonylurea Reduce duplicate hypoglycaemia risk
Preserve SGLT2/GLP-1 organ benefit where safe Avoid glucose-only substitution
Teach hypoglycaemia and sick-day response Prevent acute harm
Revisit dose after weight/renal change Prevent overtreatment

Evidence hierarchy by decision

Decision Most relevant evidence
Does lowering early HbA1c prevent microvascular disease? UKPDS
Is near-normal rapid lowering safe in high-risk long-duration disease? ACCORD/ADVANCE/VADT
Which second agent maintains HbA1c? GRADE
Which agent prevents HF/CKD/MACE? Dedicated outcome trials
Which target fits one person? Trial evidence + comorbidity, burden and preference

Common interpretive errors

  • Treating the ACCORD intensive strategy as equivalent to every low HbA1c achieved safely.
  • Treating UKPDS metformin findings as proof that metformin is always required before organ protection.
  • Ranking drugs by HbA1c while ignoring heart, kidney, weight and hypoglycaemia.
  • Intensifying an artefactually high HbA1c without checking glucose discordance.
  • Continuing insulin or sulfonylurea unchanged after major weight loss or CKD progression.

The safest target is therefore an outcome strategy, not a number in isolation: improve glycaemic exposure while minimising severe lows, weight/treatment burden and missed organ-protective opportunities.

Network outcome evidence and the current ADA pharmacologic standard supply the organ-risk and implementation layers that GRADE’s glycaemic comparison did not test (Palmer 2021, PMID 33441402; ADA 2026, PMID 41358900).

Cross-domain evidence crosswalk

These adjacent studies constrain interpretation of this page and make explicit where its conclusions depend on prevention, organ-outcome, remission, burden or implementation evidence.

Verified evidence anchor Connection
(Zinman 2015, PMID 26378978) Zinman B, et al. Empagliflozin Cardiovascular Outcomes. N Engl J Med. 2015
(Marso 2016, PMID 27295427) Marso SP, et al. Liraglutide Cardiovascular Outcomes. N Engl J Med. 2016
(Marso 2016, PMID 27633186) Marso SP, et al. Semaglutide Cardiovascular Outcomes. N Engl J Med. 2016
(Perkovic 2019, PMID 30990260) Perkovic V, et al. Canagliflozin Renal Outcomes. N Engl J Med. 2019
(Perkovic 2024, PMID 38785209) Perkovic V, et al. Semaglutide on CKD. N Engl J Med. 2024
(Frías 2021, PMID 34170647) Frías JP, et al. Tirzepatide versus Semaglutide. N Engl J Med. 2021
(Gerstein 2019, PMID 31189511) Gerstein HC, et al. Dulaglutide and cardiovascular outcomes in type 2 diabetes. Lancet. 2019;394:121-130
(Wiviott 2019, PMID 30415602) Wiviott SD, et al. Dapagliflozin and Cardiovascular Outcomes in Type 2 Diabetes. N Engl J Med. 2019
(Heerspink 2020, PMID 32970396) Heerspink HJL, et al. Dapagliflozin in Patients with Chronic Kidney Disease. N Engl J Med. 2020
(EMPA-KIDNEY 2023, PMID 36331190) EMPA-KIDNEY Collaborative Group. Empagliflozin in Patients with Chronic Kidney Disease. N Engl J Med. 2023
(Lean 2018, PMID 29221645) Lean MEJ, et al. Primary care-led weight management for remission. Lancet. 2018
(Lean 2024, PMID 38423026) Lean MEJ, et al. Five-year follow-up of DiRECT. Lancet Diabetes Endocrinol. 2024
(Schauer 2017, PMID 28199805) Schauer PR, et al. Bariatric Surgery versus Intensive Medical Therapy for Diabetes: 5-Year Outcomes. N Engl J Med. 2017
(Agarwal 2025, PMID 40813129) Agarwal A, et al. Living guideline for T2D therapeutics. BMJ. 2025

What intensive glycaemia changes—and what it does not

An individual-participant meta-analysis of ACCORD, ADVANCE, UKPDS and VADT (27,049 participants; median five years) found that a 0.90 percentage-point mean HbA1c separation reduced kidney events by 20% (HR 0.80, 95% CI 0.72–0.88) and eye events by 13% (0.87, 0.76–1.00), but not nerve events (0.98, 0.87–1.09) (Zoungas 2017, PMID 28365411). “Microvascular benefit” is therefore not uniform across organs.

ACCORD Eye provides endpoint detail: four-year retinopathy progression was 7.3% under intensive glycaemia versus 10.4% under standard treatment (adjusted OR 0.67, 95% CI 0.51–0.87); fenofibrate plus simvastatin also reduced progression (6.5% vs 10.2%; OR 0.60, 0.42–0.87), whereas systolic BP <120 versus <140 mmHg did not (Chew 2010, PMID 20587587). Four years after trial closeout, prior intensive glycaemia retained a retinopathy signal (5.8% vs 12.7%; OR 0.42, 0.28–0.63), but the fenofibrate difference did not persist (ACCORDION Eye 2016, PMID 27289122).

Competing legacy results

Evidence Legacy finding Interpretation
UKPDS Early intensive control produced durable MI/mortality benefit after HbA1c convergence Supports early treatment in newly diagnosed disease
VADT 15-year Primary CV HR 0.91 (95% CI 0.78–1.06), death HR 1.02 (0.88–1.18); benefit only while HbA1c separation persisted No durable legacy or mortality benefit in older, long-duration disease (Reaven 2019, PMID 31167051)
Diabetes & Aging cohort First-year HbA1c 6.5%–<7.0% vs <6.5% associated with microvascular HR 1.20 (1.06–1.37); 7%–<8% with mortality HR 1.29 (1.10–1.51) Observational support vulnerable to residual confounding (Laiteerapong 2019, PMID 30104301)
Steno-2 Multifactorial treatment CV HR 0.47 (0.24–0.73), later death HR 0.54 (0.32–0.89) Risk-factor package cannot identify the glucose-specific contribution (Gaede 2003, PMID 12556541; Gaede 2008, PMID 18256393)

At 21.2 years, the Steno-2 intensive group lived a median 7.9 years longer and remained free of a first cardiovascular event 8.1 years longer, but the n=160 microalbuminuric cohort and evolving post-trial care limit generalisation (Gæde 2016, PMID 27531506). The practical conclusion is early avoidance of prolonged marked hyperglycaemia alongside simultaneous organ-risk treatment, with less aggressive targets when hypoglycaemia, frailty or treatment burden dominate.

CGM beyond intensive insulin therapy

In MOBILE, 175 adults using basal insulin without prandial insulin were randomised to CGM or capillary monitoring. At eight months, adjusted HbA1c difference was −0.4 percentage points (95% CI −0.8 to −0.1), time 70–180 mg/dL was 59% versus 43% (difference 15%, 8%–23%), and time >250 mg/dL was 11% versus 27% (difference −16%, −21% to −11%) (Martens 2021, PMID 34077499). A smaller 72-person non-insulin study reported time-in-range increasing from 46% to 72% and HbA1c falling 1.3 points at six months, but lacked a no-CGM usual-care arm and cannot isolate sensor from education effects (Martens 2025, PMID 39757879).

CGM controversy Evidence for Evidence against overreach
Extend CGM beyond intensive insulin Randomised basal-insulin benefit with improved time-in-range Absolute benefit and cost depend on baseline HbA1c and engagement
Use CGM to guide diet without medication Improvement in non-insulin users during education Small study, within-group primary endpoint, no sensor-free comparator
Replace HbA1c with time-in-range Captures excursions and hypoglycaemia Long-term complication thresholds remain anchored mainly to HbA1c trials

Open questions

  • Do early GLP-1RA/SGLT2 benefits persist after discontinuation as a legacy effect? UKPDS establishes the concept only for older glycaemic strategies (Adler 2024, PMID 38772405).
  • Which time-in-range or glucose-variability metrics add outcome value in non-insulin-treated T2D?
  • How should targets incorporate treatment burden and diabetes distress alongside event risk?
  • Can pragmatic head-to-head trials compare modern combinations on hard outcomes and cost?

References

  1. UKPDS Group. Intensive blood-glucose control with sulphonylureas or insulin (UKPDS 33). Lancet. 1998. PMID 9742976
  2. UKPDS Group. Effect of intensive blood-glucose control with metformin (UKPDS 34). Lancet. 1998. PMID 9742977
  3. Holman RR, et al. 10-year follow-up of intensive glucose control. N Engl J Med. 2008;359:1577-1589. PMID 18784090
  4. Adler AI, et al. UKPDS 91: post-trial monitoring extended to 24 years. Lancet. 2024;404:145-155. PMID 38772405
  5. ACCORD Study Group. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med. 2008. PMID 18539917
  6. ADVANCE Collaborative Group. Intensive blood glucose control and vascular outcomes. N Engl J Med. 2008. PMID 18539916
  7. Duckworth W, et al. Glucose control and vascular complications in veterans. N Engl J Med. 2009. PMID 19092145
  8. Nathan DM, et al. Glycemia Reduction in Type 2 Diabetes — Glycemic Outcomes. N Engl J Med. 2022;387:1063-1074. PMID 36129996
  9. Davies MJ, et al. Management of Hyperglycemia in Type 2 Diabetes, 2022. Diabetes Care. 2022. PMID 36148880
  10. Palmer SC, et al. SGLT2 inhibitors and GLP-1 receptor agonists: network meta-analysis. BMJ. 2021. PMID 33441402
  11. American Diabetes Association Professional Practice Committee. Pharmacologic Approaches: Standards of Care-2026. Diabetes Care. 2026. PMID 41358900
  12. Zinman B, et al. Empagliflozin Cardiovascular Outcomes. N Engl J Med. 2015. PMID 26378978
  13. Marso SP, et al. Liraglutide Cardiovascular Outcomes. N Engl J Med. 2016. PMID 27295427
  14. Marso SP, et al. Semaglutide Cardiovascular Outcomes. N Engl J Med. 2016. PMID 27633186
  15. Perkovic V, et al. Canagliflozin Renal Outcomes. N Engl J Med. 2019. PMID 30990260
  16. Perkovic V, et al. Semaglutide on CKD. N Engl J Med. 2024. PMID 38785209
  17. Frías JP, et al. Tirzepatide versus Semaglutide. N Engl J Med. 2021. PMID 34170647
  18. Gerstein HC, et al. Dulaglutide and cardiovascular outcomes in type 2 diabetes. Lancet. 2019;394:121-130. PMID 31189511
  19. Wiviott SD, et al. Dapagliflozin and Cardiovascular Outcomes in Type 2 Diabetes. N Engl J Med. 2019. PMID 30415602
  20. Heerspink HJL, et al. Dapagliflozin in Patients with Chronic Kidney Disease. N Engl J Med. 2020. PMID 32970396
  21. EMPA-KIDNEY Collaborative Group. Empagliflozin in Patients with Chronic Kidney Disease. N Engl J Med. 2023;388:117-127. PMID 36331190
  22. Lean MEJ, et al. Primary care-led weight management for remission. Lancet. 2018. PMID 29221645
  23. Lean MEJ, et al. Five-year follow-up of DiRECT. Lancet Diabetes Endocrinol. 2024. PMID 38423026
  24. Schauer PR, et al. Bariatric Surgery versus Intensive Medical Therapy for Diabetes: 5-Year Outcomes. N Engl J Med. 2017. PMID 28199805
  25. Agarwal A, et al. Living guideline for T2D therapeutics. BMJ. 2025. PMID 40813129
  26. Zoungas S, et al. Intensive glucose control and microvascular outcomes. Lancet Diabetes Endocrinol. 2017;5:431-437. PMID 28365411
  27. Chew EY, et al. Medical therapies and retinopathy progression. N Engl J Med. 2010;363:233-244. PMID 20587587
  28. ACCORDION Eye Study Group. Persistent Effects of Intensive Glycemic Control on Retinopathy. Diabetes Care. 2016;39:1089-1100. PMID 27289122
  29. Reaven PD, et al. Intensive Glucose Control—15-Year Follow-up. N Engl J Med. 2019;380:2215-2224. PMID 31167051
  30. Laiteerapong N, et al. The Legacy Effect in Type 2 Diabetes. Diabetes Care. 2019;42:416-426. PMID 30104301
  31. Gaede P, et al. Multifactorial intervention and cardiovascular disease. N Engl J Med. 2003;348:383-393. PMID 12556541
  32. Gaede P, et al. Multifactorial intervention and mortality. N Engl J Med. 2008;358:580-591. PMID 18256393
  33. Gæde P, et al. Years of life gained: 21-year Steno-2 follow-up. Diabetologia. 2016;59:2298-2307. PMID 27531506
  34. Martens T, et al. CGM in Type 2 Diabetes Treated With Basal Insulin. JAMA. 2021;325:2262-2272. PMID 34077499
  35. Martens TW, et al. CGM to Guide Food Choices in People Not Taking Insulin. Diabetes Technol Ther. 2025;27:261-270. PMID 39757879