The Lancet Neurology September 28, 2011

Effects of intensive glucose lowering on brain structure and function in people with type 2 diabetes (ACCORD MIND): a randomised open-label substudy

Lenore J Launer, Michael E Miller, Jeff D Williamson, et al.

Bottom Line

In older adults with type 2 diabetes and high cardiovascular risk, intensive glucose lowering reduced brain volume loss but did not improve cognitive outcomes compared to standard glycemic control over 40 months.

Key Findings

1. At 40 months, the primary cognitive outcome—the Digit Symbol Substitution Test (DSST) score—showed no significant difference between the intensive and standard glycemic control groups (mean difference 0.32; p=0.2997).

2. In the MRI sub-cohort (n=614), total brain volume decreased significantly less in the intensive management group (-13.0 cm³) compared to the standard management group (-17.7 cm³) over 40 months (mean difference 4.62 cm³, 95% CI 2.0 to 7.3; p=0.0007).

3. Secondary cognitive tests, including the Rey Auditory Verbal Learning Test, Stroop test, and Mini-Mental State Examination, showed no significant treatment effect at any time point.

4. Although intensive glucose control preserved total brain volume, it was associated with an increase in abnormal white matter volume (p=0.0156) and did not mitigate clinical cognitive decline.

Study Design

Design

Randomized Controlled Trial

Open-Label

Sample

2,977

Patients

Duration

40 mo

Median

Setting

Multicenter, North America

Population Adults aged 55 to 80 years with type 2 diabetes, HbA1c >7.5%, and a high risk for cardiovascular events.

Intervention Intensive glycemic control (targeting a glycated hemoglobin [HbA1c] level <6.0%).

Comparator Standard glycemic control (targeting a glycated hemoglobin [HbA1c] level of 7.0% to 7.9%).

Outcome Cognitive function, measured by the Digit Symbol Substitution Test (DSST) score, and total brain volume (TBV) assessed via MRI at 40 months.

Study Limitations

• The 40-month follow-up period was likely too brief to detect meaningful differences in cognitive decline, which typically unfolds over decades.

• The open-label design meant participants and investigators were unblinded to treatment allocations, potentially introducing bias in clinical and cognitive evaluations.

• The intensive glycemic control arm of the main ACCORD trial was terminated prematurely (after roughly 3.5 years) due to increased mortality, truncating the duration of the targeted intervention.

• The study enrolled older adults with long-standing type 2 diabetes (mean duration 10 years) and high cardiovascular risk; the findings may not generalize to younger patients or those with newly diagnosed diabetes.

Clinical Significance

ACCORD MIND definitively demonstrated that intensive glycemic control (targeting HbA1c <6.0%) in older adults with long-standing type 2 diabetes does not confer cognitive benefits over 3-4 years. While the intervention did modestly attenuate brain atrophy, this structural preservation failed to translate into measurable functional benefits and was overshadowed by the 22% increase in all-cause mortality observed in the parent ACCORD trial. These results caution against the use of intensive glucose lowering for the purpose of neuroprotection or preventing cognitive decline in older diabetics, shifting the paradigm toward individualized, safer HbA1c targets that prioritize avoiding severe hypoglycemia.

Historical Context

Observational and epidemiological studies had robustly linked type 2 diabetes with accelerated cognitive decline, brain atrophy, and an increased risk of dementia. Chronic hyperglycemia and microvascular disease were theorized to drive these neurological complications. The ACCORD MIND substudy was initiated to determine whether aggressively lowering HbA1c to near-normal levels (<6.0%) could halt this deterioration. The early cessation of the main ACCORD trial in 2008 due to increased mortality in the intensive arm fundamentally shook the diabetes management landscape. The 2011 publication of ACCORD MIND further diminished the case for tight control by showing a lack of cognitive benefit, reinforcing the modern clinical consensus that heavily emphasizes tailored glycemic targets and the avoidance of therapeutic harm in older patients.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

How does chronic hyperglycemia contribute to brain volume loss in type 2 diabetes, and physiologically, why might intensive glucose lowering preserve this structural volume but fail to improve cognitive function over a 40-month period?

Key Response

This questions tests basic pathophysiology, including microvascular damage, advanced glycation end-products (AGEs), and neuroinflammation. It also highlights the pathophysiological disconnect between macroscopic structural changes (brain volume) and functional neuroplasticity or synaptic function, emphasizing that structural preservation does not immediately guarantee functional cognitive improvement.

Resident

Given the findings of the ACCORD MIND substudy alongside the main ACCORD trial's early termination, how should you counsel an older patient with type 2 diabetes and high cardiovascular risk regarding their HbA1c goals for dementia prevention?

Key Response

Residents must learn to balance competing clinical risks. The main ACCORD trial demonstrated increased mortality with intensive control, and ACCORD MIND showed no cognitive benefit to offset that risk. Therefore, counseling should focus on standard, individualized A1c targets (e.g., 7.0-8.0% depending on comorbidities) rather than strict control for cognitive preservation, avoiding the harms of severe hypoglycemia.

Fellow

The ACCORD MIND study demonstrated a preservation of total brain volume (TBV) but no significant difference in the progression of white matter lesion (WML) volume. From a neurological and endocrinological perspective, how do these divergent trajectories inform our understanding of diabetic encephalopathy's dominant mechanisms (e.g., neurodegenerative vs. ischemic)?

Key Response

Fellows should be able to integrate nuanced structural MRI markers. TBV generally reflects whole-brain atrophy (often driven by metabolic or neurodegenerative pathways), whereas WML reflects microvascular ischemic disease. The fact that intensive control altered TBV but not WML suggests that glucose toxicity might drive generalized cellular atrophy independent of small vessel ischemic changes over this specific timeframe.

Attending

The dissociation between surrogate radiologic endpoints (brain volume) and patient-centered clinical endpoints (cognitive function) is a recurring theme in neuro-metabolic research. How do we best incorporate the ACCORD MIND findings into our clinical teaching regarding the limitations of relying on MRI to predict early cognitive trajectories?

Key Response

Attendings must teach the principles of evidence-based medicine and the pitfalls of surrogate markers. A 'better' MRI (less atrophy) does not necessarily mean the patient functions better clinically. This underscores the core teaching point of 'treating the patient, not the scan' and setting realistic expectations about metabolic interventions.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

The primary intensive intervention in the main ACCORD trial was halted prematurely due to increased mortality. How does informative censoring or survival bias potentially compromise the longitudinal mixed-effects models used to analyze cognitive decline and MRI changes in the ACCORD-MIND substudy?

Key Response

PhD-level analysis requires deep methodological critique. If patients in the intensive arm died early (and were excluded from later cognitive testing), those who survived and completed the 40-month follow-up might represent a healthier cohort (the 'survivor effect'). This selective attrition could skew the cognitive outcomes, potentially masking or artificially inflating differences between the arms.

Journal Editor

As an editor handling this manuscript, the finding of a structural benefit without a functional benefit raises questions of statistical power versus clinical relevance. What specific metrics regarding the minimal clinically important difference (MCID) for the Digit Symbol Substitution Test (DSST) would you demand the authors provide to justify their conclusion that no meaningful cognitive benefit occurred?

Key Response

Editors must ensure trial conclusions are robust and not merely artifacts of insensitive testing. Demanding the MCID provides context for whether the test (DSST) was too insensitive to detect subtle changes, whether the follow-up was too short, or if the lack of difference genuinely means intensive control is clinically futile for cognition in this window.

Guideline Committee

Current ADA and AGS guidelines recommend relaxed HbA1c targets (e.g., 7.5-8.5%) for older adults with multiple comorbidities to minimize hypoglycemia. Based on ACCORD-MIND's findings, what level of evidence does this study provide against using intensive glucose control for the primary prevention of cognitive decline in this demographic, and does it validate current guidelines?

Key Response

Guideline committees must synthesize trial outcomes to formulate clinical practice recommendations. ACCORD-MIND provides Level A evidence that targeting normal or near-normal HbA1c does not protect cognitive function over ~3.5 years in older high-risk patients. This directly validates and reinforces current guideline recommendations to avoid stringent targets in this population, as the lack of cognitive benefit cannot justify the mortality and hypoglycemic risks seen in the parent trial.

Clinical Landscape

Noteworthy Related Trials

2008

ACCORD Trial

n = 10,251 · NEJM

Tested

Intensive glucose lowering (target HbA1c < 6.0%)

Population

T2DM patients with high CV risk

Comparator

Standard glucose lowering (target HbA1c 7.0-7.9%)

Endpoint

Nonfatal MI, nonfatal stroke, or CV death

Key result: Intensive therapy increased all-cause mortality and did not significantly reduce major CV events, leading to premature termination of the intensive arm.

2008

ADVANCE Trial

n = 11,140 · NEJM

Tested

Intensive glucose control (gliclazide MR-based, target HbA1c <= 6.5%)

Population

T2DM patients with high risk of vascular events

Comparator

Standard glucose control

Endpoint

Composite of major macrovascular and microvascular events

Key result: Intensive control reduced the incidence of combined major macrovascular and microvascular events, primarily driven by a reduction in nephropathy, with no significant effect on mortality or cognitive decline.

2009

VADT Trial

n = 1,791 · NEJM

Tested

Intensive glucose control (target HbA1c < 6.0%)

Population

Veterans with poorly controlled T2DM

Comparator

Standard glucose control (target HbA1c 8.0-9.0%)

Endpoint

Time to first major CV event

Key result: Intensive glucose control had no significant effect on the rates of major cardiovascular events, death, or microvascular complications compared to standard therapy.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis