Effect of Intensive Blood Pressure Control on Subtypes of Mild Cognitive Impairment and Risk of Progression from SPRINT study

Abstract

BACKGROUND:

To examine the effect of intensive blood pressure control on the occurrence of subtypes of mild cognitive impairment (MCI) and determine the risk of progression to dementia or death.

METHODS:

Secondary analysis of a randomized trial of community-dwelling adults (≥50 years) with hypertension. Participants were randomized to a systolic blood pressure (SBP) goal of <120 mm Hg (Intensive treatment; n=4678) or <140 mm Hg (Standard treatment; n=4683). Outcomes included adjudicated MCI, MCI subtype (amnestic, non-amnestic, multidomain, single domain), and probable dementia. Multistate survival models were used to examine transitions in cognitive status accounting for the competing risk of death.

RESULTS:

Among 9361 randomized participants (mean age, 67.9 years; 3332 women [35.6%]), 640 participants met the protocol definition for MCI, with intensive treatment reducing the risk of MCI overall (Hazard Ratio (HR), 0.81 [95% Confidence Interval (CI), 0.69–0.94]), as previously reported. This effect was largely reflected in amnestic subtypes (HR, 0.78 [95% CI, 0.66–0.92]) and multi-domain subtypes (HR, 0.78 [95% CI, 0.65–0.93]). An adjudication of MCI, as compared to normal cognitive function, substantially increased the probability of progressing to probable dementia (5.9% [95% CI: 4.5% to 7.7%] versus 0.6% [95% CI: 0.3% to 0.9%]) and to death (10.0% [95% CI: 8.3% to 11.9%] versus 2.3% [95% CI: 2.0% to 2.7%]) within 2 years.

CONCULSIONS:

Intensive treatment reduced the risk for amnestic and multi-domain subtypes of MCI. An adjudication of MCI was associated with increased risk of progression to dementia and death, highlighting the relevance of MCI as a primary outcome in clinical and research settings.

Clinical Trial Registration:

NCT01206062 (https://clinicaltrials.gov/ct2/show/NCT01206062?term=NCT01206062&draw=2&rank=1)

INTRODUCTION

Mild Cognitive Impairment (MCI) represents a clinical state on the continuum of neurocognitive functioning between normal functioning and dementia.¹ MCI is a risk factor for dementia, with an overall annual conversion rate of approximately 7%.² Given the general lack of efficacy in treating dementia directly,^3,4 attention has turned toward identifying therapeutic strategies for the prevention and treatment of MCI. Phenotyping MCI has identified subtypes based on the presence or absence of memory deficits (amnestic versus non-amnestic), and on the breadth of deficits (single versus multiple cognitive domains).⁵ A meta-analysis of 41 studies showed that individuals with amnestic MCI and multidomain MCI have cumulative and annual conversion rates to dementia that are significantly higher than non-amnestic MCI, suggesting that MCI may represent several disease groups.⁶ Thus, it becomes important to know if therapuetic interventions differentially affect MCI subtypes, which could lead to greater treatment precision.

Blood pressure lowering as a treatment for hypertension, a modifiable risk factor for MCI and dementia,⁷ is the only pharmacologic strategy with randomized trial evidence demonstrating reductions in cognitive impairment.^8–10 The Systolic Blood Pressure Intervention Trial (SPRINT) demonstrated a significant reduction in the occurrence of MCI in adults with hypertension who were randomly assigned to intensive systolic blood pressure (BP) control (<120 mm Hg) compared with those assigned to standard BP control (<140 mm Hg).¹¹ Given the phenotypic heterogeneity inherent to MCI, questions remain about the possible differential effect of intensive BP treatment on subtypes of MCI, as well as the influence of treatment on the progression to more severe impairment. In the present work, we use data from SPRINT in a secondary analysis to examine these issues.

METHODS

Trial Design

The trial design and methodology have been published.^11,12 Participants were 50 years of age or older and had a screening systolic blood pressure (SBP) between 130 and 180 mm Hg. Participants qualified for the study if they had clinical or subclinical cardiovascular disease, chronic kidney disease (estimated glomerular filtration rate below 60 ml/min/1.73m²) or a Framingham cardiovascular disease (CVD) risk score of 15% or greater, or if they were 75 years of age or older. Individuals residing in a nursing home, persons with a diagnosis of dementia (based on medical record review) or treated with medications primarily used for dementia therapy were excluded, as were persons with prevalent diabetes mellitus or a history of stroke. Individuals were randomized by the Data Coordinating Center (1:1) to a SBP goal of less than 120 mm Hg (intensive treatment group, n = 4678) or a goal of less than 140 mm Hg (standard treatment group, n = 4683), with the randomization stratified by clinic site. The algorithms and formulary for the trial are listed in the published study protocol.^11,12 Race and ethnicity were collected via self-report using fixed categories in order to satisfy the National Institutes of Health Policy and Guidelines on The Inclusion of Women and Minorities as Subjects in Clinical Research.

Standard Protocol Approvals, Registrations, and Patient Consents

The study was approved by the institutional review board at each participating site and each participant provided informed consent. The ClinicalTrials.gov identifier is NCT01206062.

Ascertainment for Mild Cognitive Impairment and Probable Dementia

In SPRINT MIND ascertainment of cognitive status, which occurred at years 2 and 4 of follow-up, as well as at study closeout if that was >1 year removed from the 4-year follow-up visit, followed a three-step process.¹¹ In-person cognitive screening assessments were administered to all participants at baseline and during follow-up by centrally trained and certified examiners at each clinic site, and included a test of global cognitive function (Montreal Cognitive Assessment, MoCA),¹³ verbal learning and memory (Logical Memory I and II),¹⁴ and processing speed (Digit Symbol Coding).¹⁵ For participants scoring below pre-set race/ethnicity and education specific thresholds on the MoCA,¹¹ a pre-identified proxy was administered the Functional Activities Questionnaire (FAQ).¹⁶ In addition, participants underwent further testing with an extended cognitive battery that measured attention/concentration, verbal and non-verbal memory, language, and executive function.^11,17 A validated telephone battery was administered to participants who could not be assessed in-person during follow-up.¹⁸ For participants receiving the telephone battery, the FAQ was obtained if the participant scored below 31 on the Modified Telephone Interview for Cognitive Status (TICSm).¹⁹ If a participant had died or was otherwise unable to communicate by telephone, the Dementia Questionnaire (DQ)²⁰ was administered to their pre-identified proxy.

All participants were administered standardized measures of depressive symptoms, perceived health status, and quality of life,²¹ and reported current medications, medical problems, and current health behaviors. Hospitalizations were also recorded as part of a standardized protocol for the ascertainment of serious adverse events.¹² These data were independently reviewed by two clinicians experienced in the diagnosis of MCI and dementia and who were part of an expert adjudication panel that included neurologists, neuropsychologists, geriatricians, and geropsychologists. The adjudicators, masked to treatment assignment, classified each case as: no cognitive impairment, MCI, probable dementia using standardized diagnostic criteria,^22,23 or cannot classify. MCI classifications were further sub-typed as amnestic-single domain, amnestic-multi-domain, non-amnestic-single domain, or non-amnestic multi-domain.⁵ Classification agreements were considered final. Disagreements on primary classification (no impairment, MCI, probable dementia) were discussed by the full adjudication panel on regularly scheduled conference calls with the classification decision achieved by a majority vote. Disagreements on MCI subtype were discussed between the two primary adjudicators until consensus was reached or referred to committee for review with final classification decided by majority vote. No sub-classification of probable dementia was made.

Duration of Follow-up

Trial enrollment began in November 2010 and ended in March 2013. The trial planned cognitive assessments at baseline, 2 and 4 years of follow-up, as well as at study closeout if that was >1 year removed from the 4-year follow-up visit (Supplementary Figure S1). The trial intervention was stopped early on August 20, 2015 after the Director of the National Heart, Lung, and Blood Institute accepted the Data and Safety Monitoring Board recommendation to inform the investigators and participants of the cardiovascular results. Many of the planned year 4 cognitive assessments had not been completed as of this date, and so were completed at study closeout while the trial was still providing medication at no cost to the participant. After the trial was stopped and during the closeout visits, BP management decisions were returned to the participant’s primary care physician. After the closeout visit, medications were no longer provided by the trial. A final extended follow-up visit, which included an additional cognitive assessment, was conducted between October 2017 and July 2018. For this analysis, the final date of follow-up was 7/22/2018.

Study Outcomes

Study outcomes included the occurrence of adjudicated probable dementia and MCI. Protocol-defined MCI included two consecutive occurrences of an adjudicated classification of MCI, or a classification of MCI followed by an adjudication of probable dementia (Supplementary Figure S2). While all MCI cases were classified into four subtypes, due to data sparseness, we instead examined subtypes of MCI by collapsing across dimensions, contrasting amnestic versus non-amnestic subtypes of MCI, and single versus multi-domain MCI subtypes. Because the protocol definition of MCI required two consecutive classifications of cognitive impairment (with the first classification being MCI), the subtype from the initial MCI classification was used.

Statistical Methods

The occurrence of MCI overall and for subtypes of MCI was compared between treatment groups using Cox proportional hazards regression with the baseline hazard function stratified by clinical site.²⁴ Participants without MCI and probable dementia were censored on the date of their last cognitive assessment with the exception of those who were administered the DQ. If they were deceased, they were censored on their date of death. If they were alive, they were censored on the date of the DQ administration.

To examine transitions in cognitive status, we used a 5-state (normal cognitive function, MCI, intermittent MCI, probable dementia, or death) survival model based on first-order Markov assumptions to compare the risk for transition between these states between intensive and standard treatment (FIGURE 1). Intermittent MCI refers to cases where an adjudication of MCI is followed by the participant either not meeting screening criteria for adjudication or being formally adjudicated as having normal cognitive function. Since such individuals are at higher risk of dementia than individuals that are consistently cognitively normal,²⁵ we include it as a transition state rather than treat it as reversion. The multi-state model allowed bi-directional movement between intermittent MCI and MCI, treated probable dementia and death as absorbing states, and also included an adjustment for whether the time period was before or after the decision to stop the trial intervention on August 20, 2015. Probable dementia was treated as an absorbing state because participants adjudicated with probable dementia were not subsequently assessed for cognitive function. Note that the Markov assumption implies that MCI is modeled from the perspective of current status, and so does not entail the confirmation of MCI at a second assessment time point as in the trial’s protocol definition of MCI. Due to insurmountable model complexity, we were not able to consider multi-state models with additional states reflecting MCI subtypes. Because only the first transition in a multi-state framework represents a purely randomized comparison, these models evaluate the association between intensive treatment and transition risk. The multistate models were fit using the msm package for the R Statistical Computing Environment.²⁶

FIGURE 1.

Five-state model for cognitive status

Data Availability Statement

Data are available for investigators providing an IRB/Ethics approval or certification of exemption from IRB/Ethics review, and also agreeing to the terms and conditions of a data use agreement. Data are available in BioLINCC (https://biolincc.nhlbi.nih.gov/studies/sprint).

RESULTS

A total of 9361 participants were randomized between November 2010 and March 2013 (Supplementary Figure S3). Briefly, participants were a mean age of 67.9 years (SD, 9.4 years), with 28.2% of participants aged 75 years or older (Supplementary Table S1). Randomized participants were 35.6% female, 30.0% black, and 10.5% Hispanic with a mean SBP at baseline of 139.7 mm Hg (SD, 15.6 mm Hg). Median scores at baseline for the MoCA, Logical Memory II, and Digit Symbol Coding were 23 (interquartile range (IQR), 20–26), 8 (IQR, 6–11), and 51 (IQR, 41–61), respectively.

Intervention Effects of MCI and its Subtypes

During follow-up, 4278 (91.4%) participants in the intensive treatment group completed at least one cognitive assessment, as compared to 4285 (91.5%) in the standard treatment group (Supplementary Figure S3). Over a median follow-up of 5.06 years (interquartile range 3.82 to 5.96 years), 640 participants met the protocol-definition of MCI which required two consecutive adjudications of cognitive impairment. Supplementary Table S2 displays the frequency of the initial and second adjudication for cases of MCI by subtype. Based on the first adjudication of MCI, the majority of cases were amnestic multi-domain (69.2%) or amnestic single domain (16.9%), while non-amnestic multi- and single-domain accounted for 6.8% and 7%, respectively. For those initially adjudicated as amnestic multi-domain, the majority were either adjudicated again as amnestic multi-domain at the second assessment (77.9%) or had progressed to probable dementia (14.5%). With respect to treatment group differences, MCI occurred in 287 participants in the intensive treatment group and 353 participants in the standard treatment group (14.6 versus 18.3 per 1000 person-years, Hazard Ratio (HR), 0.81 [95% CI, 0.69–0.94], FIGURE 2), as reported previously. There was a similar beneficial effect of intensive treatment on the occurrence of amnestic subtypes of MCI (HR = 0.78; 95% CI, 0.66–0.92; p-value = 0.003), as well as multi-domain subtypes of MCI (HR = 0.78, 95% CI, 0.65–0.93, p-value = 0.007, FIGURE 3). In contrast, the occurrence of non-amnestic and single domain subtypes of MCI was similar between the treatment groups.

FIGURE 2.

Occurrence of Mild Cognitive Impairment by Treatment group

Shaded regions indicate 95% confidence intervals.

FIGURE 3.

Occurrence of Subtypes of mild cognitive impairment by treatment group

Shaded regions indicate 95% confidence intervals.

Transitions in Cognitive Status

As shown in Supplementary Table S3, transitions indicative of progressive cognitive impairment were generally less frequent in the intensive treatment group. Participants randomized to intensive treatment less frequently transitioned from normal cognition to MCI (8.4% versus 9.4%), and less frequently progressed from MCI to probable dementia (4.6% versus 7.1%) compared to participants receiving standard treatment. Transitions from indeterminate adjudications represented a small proportion of transitions in both treatment groups (0.7% for both treatment groups). Supplementary Table S4 examines model fit for the multistate survival model, comparing the observed versus expected prevalence of each cognitive state at various points of follow-up. The multistate model generally fit well through 4.5 years of follow-up, though it slightly over-estimated the frequency of MCI and under-estimated the frequency of death. Model fit was poorer at 5.25 years of follow-up, though this is not unexpected as this largely reflects the extended, observational follow-up period, when many participants were not accessed. The estimated 2-year transition probabilities from the multistate model are shown in TABLE 1. Overall, from a state of normal cognitive function, the most frequent transition was to MCI (estimated 2-year transition probability (P₂ = 4.3%, 95% CI: 3.9% to 4.9%), with transitions to probable dementia (P₂ = 0.6%, 95% CI: 0.5% to 0.9%) or death (P₂ = 2.3%, 95% CI: 2.0% to 2.7%) being less frequent. From a state of MCI, the most likely transition was to intermittent MCI (P₂ = 31.6%, 95% CI: 28.9% to 34.5%). Compared to a current status of normal cognitive function, the risk of transitioning to probable dementia or death was higher coming from MCI, with 2-year transition probabilities for probable dementia and mortality increasing to 5.9% (95% CI: 4.5% to 7.7%) and 10.0% (95% CI: 8.3% to 11.9%), respectively. Consistent with the overall trial results,¹¹ the risk of transitioning from normal cognitive function to MCI was lower with intensive treatment, though there were no other significant associations between treatment group and transition risk within the multistate model.

TABLE 1.

2-year transition probabilities from five state model for cognitive status

Transition			Intensive	Standard	Hazard Ratio
From	To	Overall	Treatment	Treatment	(95% CI)
Normal Cognition	MCI	4.3% (3.9%, 4.9%)	4.1% (3.6%, 4.6%)	4.6% (4.1%, 5.2%)	0.87 (0.79, 0.96)
Normal Cognition	Probable Dementia	0.6% (0.5%, 0.9%)	0.6% (0.3%, 0.9%)	0.6% (0.5%, 0.9%)	1.26 (0.76, 2.09)
Normal Cognition	Death	2.3% (2.0%, 2.7%)	2.2% (1.9%, 2.7%)	2.4% (2.0%, 2.9%)	0.96 (0.73, 1.25)
MCI	Intermittent MCI	31.6% (28.9%, 34.5%)	32.8% (29.1%, 36.4%)	30.5% (27.2%, 34.0%)	1.04 (0.87, 1.25)
MCI	Probable Dementia	5.9% (4.5%, 7.7%)	4.9% (3.4%, 7.0%)	6.9% (5.0%, 9.5%)	0.70 (0.46, 1.07)
MCI	Death	10.0% (8.3%, 11.9%)	9.3% (7.1%, 12.1%)	10.5% (8.3%, 13.3%)	0.88 (0.62, 1.24)
Intermittent MCI	MCI	25.1% (20.7%, 30.2%)	24.4% (18.4%, 31.5%)	25.6% (20.0%, 32.2%)	0.92 (0.59, 1.45)

CI denotes confidence interval and MCI Mild Cognitive Impairment. Estimates are from multistate survival model as shown in FIGURE 1. In SPRINT, 63% of follow-up was during the active intervention phase, the rest consisting of observational follow-up. Above estimates similarly assume 63% of 2-year interval consists of active intervention.

DISCUSSION

Hypertension is a well-established risk factor for cognitive impairment including dementia.^27–29 As it is modifiable, it represents an important interventional target against a leading cause of disability and death in modernized societies. SPRINT has previously shown that intensive BP control to a goal of <120 mm Hg, as compared to a target of <140 mm Hg, significantly reduced the risk of MCI,¹¹ a precursor to dementia, and was associated with smaller increases in white matter hyperintensities based on magnetic resonance imaging.³⁰ This study extends those findings with two primary observations. First, the protective effect of intensive BP treatment on MCI was largely reflected in amnestic and multi-domain subtypes of MCI, subtypes which generally exhibit higher risk of progression to dementia.⁶This signals that the response to other treatments also may not be consistent across subtypes. Second, following an adjudication of MCI of any subtype, the risk of transitioning to probable dementia or death greatly increased compared to normal cognitive function. While the increased risk of progression associated with MCI is well established for Alzheimer’s dementia,⁹ these results suggest that the risk may be similar among individuals with significant cardiovascular risk.

Our findings are consistent with prior studies showing that amnestic and multi-domain MCI are the most prevalent subtypes in persons being assessed for cognitive decline.^31–33 In the longitudinal Sydney Memory and Ageing Study, 84% of participants who were diagnosed with incident dementia were previously diagnosed with MCI, most commonly the amnestic and multi-domain subtypes.²⁵ Similarly, in the Framingham Heart Study and Mayo Clinic Study of Aging, the risk of progression from MCI to dementia was higher for amnestic compared to non-amnestic MCI, and for multi-domain compared to single domain MCI.⁶ The finding that intensive treatment provided benefit for these high-risk subtypes suggests it may ultimately help to prevent or delay more severe cognitive impairment including dementia. SPRINT did not show a significant effect of intensive treatment on probable dementia (HR = 0.83; 95% CI, 0.67–1.04),¹¹ however, the relative effect was very similar to the significant estimated treatment effect reported for MCI (HR = 0.81; 95% CI, 0.69–0.95). While we cannot rule out that we may have been underpowered, it appears that the protective effect of intensive treatment may occur most strongly at a place on the continuum of cognitive impairment beyond relatively minor cognitive deficits (i.e. single domain), and more in the range of broader, more severe deficits (i.e., multi-domain), especially in the presence of memory deficits.

We also observed that an adjudication of MCI conferred increased risk of progressing to probable dementia or death, further suggesting that intensive treatment not only prevents clinically meaningful MCI, but could also ultimately prevent or delay the onset of dementia, an hypothesis that is also supported by recent meta-analyses of blood pressure lowering trials.^10,34 These results suggest that MCI should be considered a primary clinical outcome in care settings as well as in future prevention trials of cognitive impairment as it likely represents an earlier stage in disease progression.

While the results from SPRINT support the choice of MCI as a clinical target, they also raise a number of questions about how best to operationalize it. We observed, as others have, that a sizable proportion of persons initially adjudicated with MCI subsequently did not meet criteria for adjudication, or were classified in cross-section as having normal cognitive function.^9,25,35,36 This was not unexpected given that MCI represents an early stage in the progression of cognitive impairment; the boundaries between pre-clinical deficits and evident mild impairment often blur. Using cognitive test thresholds designed to be more sensitive for MCI and repeated cognitive testing to demonstrate persistent cognitive impairment for defining MCI, as was done in SPRINT, could reduce misclassifications. It is worth noting, individuals with more variable cognitive status (i.e. intermittent MCI) exhibited consistently lower MoCA scores compared to participants with normal cognition in SPRINT (Supplementary Figure S4), and are known to be at higher risk for developing more severe impairment.²⁵

Strengths and Limitations

There are several limitations to this study. First, loss to follow-up with the extended follow-up visits, though not different for each treatment group, could have been related to cognitive status and led to under-ascertainment of cognitive outcomes. Second, the trial did not adjudicate baseline cognitive status, therefore we cannot exclude nor examine the influence of prevalent MCI at the time of randomization. Third, we only observed a significant association between treatment and the transition between normal cognitive function and MCI. The lack of associations with other specific transitions likely reflects, to some extent, a lack of statistical power. Multistate models naturally reflect more complex outcomes, with greater data requirements for observing specific transitions between states as compared to analyses of time to the first event, i.e., time to event analyses do not typically differentiate between transitioning to probable dementia from normal cognitive function versus MCI. A similar observation with respect to physical activity was made in the Lifestyle Interventions and Independence for Elders trial. While a structured physical activity program reduced the incidence of the first occurrence of major mobility disability, the intervention was not significantly associated with specific transitions between mobility states.³⁷ The strengths of this study are large sample size, the use of multiple measures for cognitive domains, blinded adjudication, length of participant follow up, and the trial’s stringent protocol definition of MCI.

CONCLUSIONS

Hypertension is one of the most modifiable risk factors for cognitive impairment. Blood pressure control is relatively low cost, readily available and easy to administer, making it an attractive choice for preventing cognitive decline in the population. Intensive blood pressure treatment reduced the risk for amnestic and multi-domain subtypes of MCI, a transitional state between normal cognitive functioning and dementia. An adjudication of MCI was associated with increased risk of progression to dementia and death among individuals at cardiovascular risk, highlighting the relevance of MCI as a clinical outcome. Intensive blood pressure control represents a viable option in preventing or delaying a major public health threat in the US and beyond.

Supplementary Material

Supplementary Online Content

Supplementary Table S1. Baseline characteristics of trial participants

Supplementary Table S2. Comparison of initial and subsequent subtypes for cases meeting protocol definition of Mild Cognitive Impairment

Supplementary Table S3. Number and percentage of transitions between cognitive states across all follow-up visits by treatment group

Supplementary Table S4. Observed and Expected Prevalence at Selected Follow-up Times

Supplementary Figure S1. Timeline for follow-up cognitive assessments in SPRINT

Supplementary Figure S2. Schematic depicting possible combinations of adjudication decisions including Mild Cognitive Impairment (MCI)

Supplementary Figure S3. Consolidated Standards of Reporting Trial (CONSORT) flow diagram of participants in the Systolic Blood Pressure Intervention Trial (SPRINT)

Supplementary Figure S4. MoCA scores by adjudicated cognitive status during follow-up

Key Points:

• Among ambulatory older men and women with hypertension, more aggressive systolic blood pressure control reduced the risk of mild cognitive impairment, primarily in the amnestic and multidomain subtypes of MCI.

• Having MCI substantially increased the likelihood of progression to dementia or death compared to having no cognitive impairment.

• Lower systolic blood pressure control could ultimately reduce the likelihood of dementia and death

Why does this paper matter?

Subtyping MCI combined with intensive blood pressure control could lead to greater treatment efficacy and precision. MCI should be considered a primary outcome in clinical and research contexts.