Mild Cognitive Impairment as a Clinical and Research Outcome: Ready for Prime Time?

Introduction

Mild cognitive impairment (MCI) was first introduced as a transitional stage between normal cognitive function and dementia.¹ While conceptually straightforward, MCI has resisted simple definitions and remains a complex, heterogeneous clinical entity.^1,2 Specifically, MCI is often sub-divided into single-domain versus multi-domain and amnestic versus non-amnestic MCI, with each of these subtypes showing differential risks of progression to dementia.^1,3,4 While patients across all MCI subtypes appear to have an elevated risk of progression to dementia compared to older adults with normal cognition, most MCI patients (regardless of subtype) do not progress to dementia or revert to normal cognition on follow-up.^3,5–10

To decrease the risk of progression from normal cognition to MCI and ultimately dementia, recent research has focused on modification of cardiovascular risk factors, such as hypertension and diabetes.¹¹ The Systolic Blood Pressure Intervention Trial-Memory and Cognition in Decreased Hypertension (SPRINT-MIND) study examined whether participants with cardiovascular risk factors randomized to more intensive blood pressure (BP) treatment had improved cognitive outcomes.¹² While intensive BP treatment led to a non-statistically significant reduction in all-cause probable dementia, the study found that intensive BP treatment led to a statistically significant reduction in MCI and the combined outcome of MCI or probable dementia.

Study Summary and Interpretation

In this issue of the Journal of the American Geriatrics Society, Gaussoin and colleagues provide an important addition to our knowledge about BP treatment and MCI.¹³ The authors examined the effect of intensive BP control on risk of progression to different subtypes of MCI among participants in SPRINT-MIND. They found that the beneficial effects of intensive treatment were largely in reducing the rate of amnestic MCI (hazard ratio (HR)=0.78, 95% CI 0.66–0.92) and multi-domain MCI (HR=0.78, 95% CI 0.65–0.93). Additionally, they examined transitions in cognitive status using a 5-state Markov survival model and found that a diagnosis of MCI when compared to normal cognitive function was associated with an increased two-year transition probability of progressing to probable dementia (5.9% vs. 0.6%) and death (10.0% vs. 2.3%). While intensive treatment reduced the transition rate from normal cognitive function to MCI, there were no other statistically significant associations between treatment groups which was consistent with the overall trial results.

These results have several important implications. First, while the authors suggest that the protective effect of intensive BP control largely applies to both amnestic MCI and multi-domain MCI, it may be more accurate to conclude that BP control decreases the rate of amnestic multi-domain MCI. Specifically, Supplementary Table S2 shows that 91% (443/487) of multi-domain MCI on first assessment was amnestic multi-domain MCI. Similarly, 80% (443/551) of amnestic MCI was amnestic multi-domain MCI. Given how much the amnestic and multi-domain results are driven by the large group of combined amnestic multi-domain MCI, it is not surprising that the HRs for amnestic and multi-domain MCI are nearly identical. Thus, this study suggests that intensive BP treatment decreases the risk of a common type of MCI: amnestic multi-domain MCI. This may be especially important as these individuals appear to be at highest risk for progression to dementia.³

Since the study showed that intensive BP treatment decreases amnestic multi-domain MCI, it is critical to determine how reliably we can differentiate between MCI subtypes. Supplementary Table S2 shows that only 8% of individuals with amnestic multi-domain MCI had a different type of MCI on follow-up cognitive assessment. However, for other MCI subtypes, over 50% had a different subtype of MCI on follow-up cognitive assessment. This suggests that while amnestic multi-domain MCI may be a reliable and stable outcome, other MCI subtypes appear less reliable and less likely to be useful for clinical or research purposes.

The multi-state model (MSM) confirms the conclusions from previous studies: while MCI increases the risk for dementia, most patients diagnosed with MCI remain in MCI or revert to normal.^3,5–10 Thus, while the authors focus on the 5.9% of participants with MCI who progress to dementia within 2 years, 84.1% of participants with MCI remain in the MCI category or revert to normal cognition (10% died). As the diagnosis of MCI is becoming more common in clinical settings, it is important that clinicians clearly communicate that MCI is a heterogenous entity that usually does not progress to dementia.

While the MSM is an interesting approach to the SPRINT MIND data, it is less clear whether the added complexity of the MSM leads to new insights. The primary result from the MSM was that intensive BP treatment decreased the risk of MCI (HR=0.87, 95% CI 0.79–0.96). This is similar to the previously reported results from SPRINT-MIND (HR=0.81, 95% CI 0.69–0.95).¹² This is not surprising given that SPRINT-MIND had at most 3 full cognitive assessments, meaning that only 2 transitions were possible. With few transitions per study participant, it is unlikely that a MSM result will differ substantially from a survival analysis.

The MSM appears reasonably well calibrated up to the 3.25-year timepoint. However, as the authors acknowledge, substantial miscalibration is seen at the 4.25 and 5.25-year timepoints, with 30–40% overestimation of the rate of MCI and similar underestimation of the rate of dementia. This miscalibration may be due to the “Intermittent MCI” state that can only transition to the MCI state. The rationale for the “Intermittent MCI” state is that these patients are at higher risk for dementia compared to those with normal cognition.¹⁴ Yet the model allows for a direct transition from Normal Cognition to Dementia but does not allow for a transition between “Intermittent MCI” to Dementia, forcing all individuals to go through the MCI state. This structure, where patients assigned to the “Intermittent MCI” state get trapped and are only able to escape by transitioning to the MCI state, likely contributes to model miscalibration (overestimating MCI and underestimating dementia rate). Adding transitions between “Intermittent MCI” and Dementia and “Intermittent MCI” and Normal Cognition may result in a better calibrated MSM.

Implications and Next Steps

Given their findings, the authors conclude that MCI should be considered a primary outcome in clinical settings and future prevention trials of cognitive impairment. However, the unreliability and heterogeneity of MCI suggests this conclusion is premature. In clinical settings, diagnosis of MCI may provide several benefits, such as in-depth assessments for reversible causes (e.g., depression, polypharmacy), emphasizing cardiovascular risk factor modification, implementing nonpharmacologic interventions (e.g., diet, exercise), and discussing advance care planning.^10,15 However, this must be balanced against potential harms. Over-aggressive BP treatment may have negative effects (e.g., acute kidney injury, syncope, orthostatic hypotension), particularly among those with multimorbidity and advanced frailty that were excluded from SPRINT.¹⁶ Additionally, given the strong stigma associated with dementia, the MCI label can create anxiety for some patients when most will never progress to dementia or even revert to normal cognition on repeat testing. The instability in MCI diagnosis is particularly prominent with early assessments and single-domain MCI as individuals may “bounce around” cognitive testing cut-offs due to the subtle nature of early impairment.¹ If MCI is widely adopted in clinical settings, it is critical that patients are appropriately informed that while a diagnosis of MCI increases the risk of dementia, it is by no means a sign that further cognitive decline is inevitable. Terms such as “a form of mild dementia” and “predementia” should be avoided, and diagnosis should be accompanied by longitudinal follow-up and support.¹⁷

Similarly, we believe it is premature to use all-type MCI as a primary outcome in research settings as it captures a heterogenous population at varying risk of progression to dementia. However, this and other studies have shown that amnestic multi-domain MCI may be especially likely to progress.³ Thus, for future prevention trials considering MCI as a primary outcome, focusing on the amnestic multi-domain MCI subtype may have the most utility as a reliable surrogate of those at highest risk for progressing to dementia.

Despite the possible advantages of a timely diagnosis of MCI, these are largely based on expert opinion rather than rigorous studies.¹⁸ Until recently, a primary advantage of earlier diagnosis, namely earlier treatment with disease-modifying therapies, was theoretical. However, with the approval of aducanumab in MCI due to Alzheimer’s disease (AD) and breakthrough therapy designation given to other anti-amyloid antibodies (donanemab, lecanemab, and gantenerumab), it is likely that greater emphasis will be placed on promptly diagnosing MCI.^15,19 This will increasingly lead to further work-up (e.g., amyloid positron emission tomography, lumbar puncture, genetic testing) to identify the cause as potentially due to AD.¹⁹ The full ramifications of aducanumab’s approval are unclear, although the Centers for Medicare and Medicaid Services has indicated that the increase in 2022 Medicare Part B premiums is due in part to contingency planning if Medicare ultimately covers aducanumab.²⁰ When targeting general prevention of dementia, these developments highlight the importance of identifying MCI subtypes that are most likely to progress to dementia. The study by Gaussoin and colleagues adds to our evidence base suggesting that we should focus on amnestic multi-domain MCI rather than all subtypes of MCI.