UKPDS 34 Extension Data and What Happened After the Trial Ended

By
Published Updated Last reviewed
Clinical medical image for trials ukpds 34: UKPDS 34 Extension Data and What Happened After the Trial Ended

At a glance

| Parameter | Detail | |---|---| | Trial | UKPDS 34 (original) + UKPDS 80 (post-trial monitoring) | | N | 1,704 overweight patients in the original metformin analysis; 3 to 277 in the broader UKPDS 80 cohort | | Intervention | Metformin (up to 2 to 550 mg/day) as initial pharmacotherapy | | Comparator | Conventional therapy (diet alone) | | Duration | Median 10.7 years (UKPDS 34) + 8.5 years post-trial monitoring (UKPDS 80) | | Primary endpoint | Any diabetes-related endpoint (sudden death, hyperglycemia, hypoglycemia, renal failure, amputation, vitreous hemorrhage, MI, angina, heart failure, stroke) | | Key result | 32% risk reduction in diabetes-related endpoints with metformin during trial; persistent 21% reduction in any diabetes-related endpoint and 33% reduction in MI at post-trial follow-up |

Why the Original Trial Could Not Answer the Durability Question

UKPDS 34 was published in 1998, reporting that metformin reduced diabetes-related endpoints by 32% (p = 0.002), all-cause mortality by 36% (p = 0.011), and myocardial infarction by 39% (p = 0.01) compared with conventional (diet-only) therapy in overweight patients with newly diagnosed type 2 diabetes. These results immediately set metformin apart from sulfonylureas and insulin, which had shown microvascular but not macrovascular benefits in the parallel UKPDS 33.

But UKPDS 34 left several questions unanswered. Over the median 10.7 years of follow-up, HbA1c in all groups drifted upward as beta-cell function declined. The metformin group maintained a mean HbA1c roughly 0.6% lower than conventional therapy, but this gap was narrowing. Clinicians needed to know whether the early benefits would persist, vanish, or reverse once glycemic differences equalized. The trial itself could not resolve this because it was still actively running its intervention phase.

The Post-Trial Monitoring Program: UKPDS 80

In 2007, the UKPDS investigators published the post-trial monitoring results (UKPDS 80) in the New England Journal of Medicine. After the original trial ended on September 30, 1997, surviving patients entered a monitoring-only phase. No attempt was made to maintain previous treatment allocations. Within one year, HbA1c differences between the metformin and conventional groups had converged completely. By five years of post-trial follow-up, both groups had nearly identical HbA1c values hovering around 8.0%.

The Legacy Effect Framework

The concept that emerged from UKPDS 80, now widely called the "legacy effect" or "metabolic memory," can be summarized in a simple framework:

Phase 1 (Active trial, years 0 to 10.7): Early intensive therapy with metformin produces HbA1c separation from the control group and accumulates a cardiovascular benefit that is proportional to the glucose difference multiplied by time exposed.

Phase 2 (Post-trial convergence, years 10.7 to 12): HbA1c equalizes between groups within 12 months. No ongoing treatment difference exists. Conventional reasoning would predict that outcomes should also converge.

Phase 3 (Extended follow-up, years 12 to 19+): Despite identical glycemic control, the original metformin group continues to show significantly fewer MIs and lower mortality. The early benefit has been "locked in."

This three-phase pattern differs from what would be expected if metformin's benefit were purely a function of concurrent glucose control. Something about early exposure, whether direct vascular protection, reduced glucotoxicity during a critical window, or slowed atherosclerotic progression, appears to produce effects that outlast the drug's pharmacologic presence.

Post-Trial Results in Detail

The metformin subgroup in UKPDS 80 included 342 patients originally assigned to metformin and 411 assigned to conventional therapy. At the end of the extended follow-up (median total of 17.7 years from randomization), the following risk reductions were observed:

| Outcome | Risk Reduction (metformin vs. conventional) | p-value | |---|---|---| | Any diabetes-related endpoint | 21% | 0.01 | | Myocardial infarction | 33% | 0.005 | | All-cause mortality | 27% | 0.002 | | Diabetes-related death | 30% | 0.01 |

Compare these to the original UKPDS 34 results at trial end: 32% for any diabetes-related endpoint, 39% for MI, 36% for all-cause mortality. The effect sizes shrank modestly but remained statistically significant years after between-group glucose differences had vanished. The MI result, in particular, remained strong across multiple sensitivity analyses.

Regression to Mean and Other Statistical Concerns

Critics raised several legitimate objections to the legacy effect interpretation.

Regression to the mean. Because randomization occurred decades before the post-trial analysis, differential dropout and competing mortality could have biased the surviving cohort. Patients who survived long enough to be counted in UKPDS 80 may have been inherently healthier. The investigators addressed this by performing intention-to-treat analyses that included all originally randomized patients, with deaths counted as events rather than censored.

Small subgroup size. The metformin arm of UKPDS was only 342 patients. While this was adequate for the original primary endpoint, subgroup analyses (such as metformin plus sulfonylurea combinations) were underpowered and produced some paradoxical results. The original UKPDS 34 publication itself flagged the unexpected 96% increased risk of diabetes-related death when metformin was added to sulfonylurea therapy (p = 0.039), a finding that has never been convincingly replicated and is generally attributed to the small sample size and multiple comparisons.

Confounding by subsequent therapy. After the trial ended, treatment decisions were left to individual clinicians. The investigators tracked medications but could not control for them. It is possible that the metformin group's physicians made different prescribing choices based on the known trial allocation, though there is no evidence this occurred systematically.

Safety Signals That Emerged Over Two Decades

Metformin's safety profile remained remarkably stable across the extended follow-up period. The main signals worth noting:

Lactic acidosis. The theoretical risk of metformin-associated lactic acidosis was already under debate when UKPDS 34 began. A 2010 Cochrane systematic review of 347 comparative trials found no cases of fatal or nonfatal lactic acidosis attributable to metformin. The UKPDS long-term data were consistent with this finding: no excess lactic acidosis events were reported in the metformin cohort.

Vitamin B12 deficiency. Long-term metformin use is associated with reduced B12 absorption. The HOME trial (2010) confirmed that 4.3 years of metformin therapy reduced B12 levels by 19%, with 7.2% of metformin-treated patients developing frank deficiency versus 2.3% on placebo. UKPDS did not systematically measure B12, so the long-term prevalence in that cohort is unknown. Current ADA Standards of Care recommend periodic B12 monitoring in patients on long-term metformin.

GI tolerability. The well-known gastrointestinal side effects (diarrhea, nausea, abdominal discomfort) were reported in approximately 20% to 30% of patients during the active trial. Extended-release formulations, not widely available during UKPDS, have since reduced these effects. The current FDA label for metformin lists GI adverse events as the most common reason for discontinuation.

What Changed in Clinical Practice

The UKPDS 34 plus UKPDS 80 combination fundamentally shaped diabetes treatment guidelines. Before these data, clinicians debated whether any glucose-lowering agent could reduce macrovascular events. The ADA/EASD consensus now positions metformin as a first-line agent for most patients with type 2 diabetes, a recommendation that rests primarily on UKPDS evidence.

The legacy effect concept also influenced trial design for newer agents. The DCCT/EDIC trial in type 1 diabetes showed a similar pattern of durable benefit from early intensive therapy. Together, these programs established that early glycemic control carries long-term dividends, even if control later deteriorates. This principle now underpins the clinical urgency around early T2D treatment.

Limitations the Authors Acknowledged

The UKPDS investigators were transparent about several weaknesses:

  1. The overweight subgroup definition (>120% ideal body weight) captured a relatively homogeneous population. Results may not generalize to lean patients with type 2 diabetes or to populations with different ethnic backgrounds. The trial was conducted almost entirely in white British patients.
  2. Metformin doses (up to 2 to 550 mg/day) were higher than what many patients tolerate in clinical practice. The intention-to-treat benefit includes patients who reduced or stopped metformin during the trial.
  3. The metformin-plus-sulfonylurea mortality signal, while likely a statistical artifact, could not be fully resolved with the available data. This generated years of clinical confusion, particularly in the UK, where some clinicians hesitated to combine metformin with sulfonylureas.
  4. The post-trial monitoring was observational. While the original randomization was preserved in the analysis, the absence of ongoing treatment allocation means UKPDS 80 is, strictly speaking, a prospective cohort study of a previously randomized population.

The 30-Year and Beyond View

The UKPDS cohort remains the longest continuously followed population of type 2 diabetes patients from diagnosis. Some patients have now been tracked for over 30 years. Intermittent updates from the UKPDS Group have confirmed that the mortality curves continue to separate, though statistical power diminishes as the cohort ages and shrinks. The most recent data presented at EASD conferences suggest the MI benefit persists with point estimates similar to those reported in UKPDS 80.

No other metformin trial has come close to replicating the UKPDS follow-up duration. The SPREAD-DIMCAD trial (2013) showed metformin reduced recurrent cardiovascular events versus glipizide over 3 years in patients with established coronary disease, providing supportive but shorter-term evidence. The ongoing debate about whether metformin's benefits are purely glycemic or partly pleiotropic (anti-inflammatory, AMPK-mediated) remains unresolved, though the legacy effect data lean toward a contribution beyond glucose lowering alone.

Frequently asked questions

›
›
›
›
›
›
›
›
›
›

References

  1. UK Prospective Diabetes Study (UKPDS) Group. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). Lancet. 1998;352(9131):854-865. PubMed
  2. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes (UKPDS 80). N Engl J Med. 2008;359(15):1577-1589. PubMed
  3. de Jager J, Kooy A, Lehert P, et al. Long term treatment with metformin in patients with type 2 diabetes and risk of vitamin B-12 deficiency: randomised placebo controlled trial (HOME trial). BMJ. 2010;340:c2181. PubMed
  4. Salpeter SR, Greyber E, Pasternak GA, Salpeter EE. Risk of fatal and nonfatal lactic acidosis with metformin use in type 2 diabetes mellitus. Cochrane Database Syst Rev. 2010;(4):CD002967. PubMed
  5. Hong J, Zhang Y, Lai S, et al. Effects of metformin versus glipizide on cardiovascular outcomes in patients with type 2 diabetes and coronary artery disease (SPREAD-DIMCAD). Diabetes Care. 2013;36(5):1304-1311. PubMed
  6. FDA. Metformin hydrochloride prescribing information. AccessFDA