Influence of Baseline Diastolic Blood Pressure on the Effects of Systolic Blood Pressure Lowering on Cognitive Function in Type 2 Diabetes Mellitus

Abstract

Background

Lowering of systolic blood pressure (SBP) in patients with low diastolic blood pressure (DBP), can further lower DBP. This can potentially decrease cerebral perfusion and cognition. We examined the influence of baseline DBP on the effect of lowering SBP on cognition.

Methods

This is a post hoc analysis of the Memory in Diabetes (MIND) substudy (N = 1,430) of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study (NCT00000620). Standard neuropsychological tests (Digit Symbol Substitution Test [DSST], Mini-Mental State Examination [MMSE], Rey Auditory Verbal Learning Test [RAVLT], and Stroop test) were performed at baseline and months 20 and 40. We compared the effects of intensive (goal SBP <120 mm Hg) vs. standard (goal SBP <140 mm Hg) SBP control on the changes in the 4 test scores from baseline to the averages of months 20 and 40 across the range of baseline DBP using cubic spline terms.

Results

Mean age was 63 ± 6 years, 55% were women and 66% White. Participates with lower baseline DBP were older, had more cardiovascular events and a longer duration of diabetes. There was no difference in the change in DSST (−0.22; 95% CI −0.97, 0.52), MMSE (−0.14; 95% CI −0.34, 0.06), RAVLT (−0.12; 95% CI −0.29, 0.06), and Stroop interference (−0.47; 95% CI −1.76, 0.82) in the intensive vs. standard SBP intervention. There was no interaction between baseline DBP and change in scores with the SBP intervention.

Conclusions

Intensive SBP reduction does not adversely affect cognition, even in those with low baseline DBP.

Cognitive impairment and dementia negatively affect activities of daily living, quality of life, sense of well-being, morbidity, and mortality. Half of dementia cases are attributable to modifiable risk factors and even a modest 10%–25% reduction in these risk factors has been estimated to decrease the prevalence of dementia by nearly 500,000 cases in the United States.¹ Aggressive management of hypertension may thus represent an important public health strategy for lowering the risk of dementia. Both high systolic blood pressure (SBP) and diastolic blood pressure (DBP) are independently associated with impaired cognition.² Intensive SBP control reduces the incidence of mild cognitive impairment and dementia and slows cognitive decline.^3–7 However, lowering of SBP also decreases DBP and mean arterial pressure (MAP) estimated as 2/3rd of DBP + 1/3rd of SBP. As brain perfusion depends upon MAP, an intervention that lowers SBP could theoretically result in worsening of cognition in those with low DBP. Thus, the effects of intensive SBP lowering on cognition could be modified by the baseline DBP.

The Action to Control Cardiovascular Risk in Diabetes Blood Pressure (ACCORD BP) trial, in a two-by-two factorial randomized controlled study design, tested the effects of intensive glycemia control (glycated hemoglobin <6% vs. 7.0%–7.9%) and intensive SBP control (SBP goal <120 vs. <140 mm Hg) on major cardiovascular outcomes in 4,733 participants with type 2 diabetes mellitus.⁸ In a subgroup of ACCORD BP participants (N = 1,439) without dementia, cognitive function was assessed at baseline and months 20 and 40.^9,10 In this post hoc analysis of limited-access ACCORD BP trial data, we examined whether low baseline DBP adversely modifies the effect of SBP lowering on cognitive function.

METHODS

Standard protocol approvals, registrations, and patient consents

The current study is a secondary analysis of limited-access ACCORD BP dataset obtained from Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC) of the National Heart, Lung and Blood Institute (NHLBI).¹¹ The ACCORD study protocol was approved by the institutional review board or ethics committee at each participating center, as well as by a review panel at the NHLBI. All patients provided written informed consent (ClinicalTrials.gov number, NCT00000620).

Participants

Of the 4,733 ACCORD BP participants those aged ≥55 years with high risk for cardiovascular events, English or Spanish speaking (for testing cognitive function), and with no clinical evidence of cognitive impairment or dementia at baseline, were included in the Memory in Diabetes (MIND) substudy (N = 1,439) and underwent cognitive function assessment. In this analysis, we included 1,430 of these participants who had baseline data for Digit Symbol Substitution Test (DSST) (Figure 1).

Figure 1.

CONSORT flow diagram. Abbreviations: ACCORD BP, Action to Control Cardiovascular Risk in Diabetes Blood Pressure trial; ACCORD MIND, Action to Control Cardiovascular Risk in Diabetes Memory in Diabetes substudy; BP, blood pressure; DSST; Digit Symbol Substitution Test.

BP measurement and intervention

Detailed methods of BP measurement and intervention have been published previously.¹² Briefly, BP was measured with a standardized protocol using an automated device (Omron 907). BP was measured at baseline and then once every month for 4 months and every 2 months thereafter for participants in the intensive BP intervention arm, and at months 1 and 4 and every 4 months thereafter for participants in the standard BP intervention arm. At each of these visits, an average of 3 sitting BP measurements was recorded. Additional visits were permitted in both arms to achieve BP goals. For the standard SBP intervention arm, antihypertensive dose titration or addition of a new drug was recommended if SBP was >160 mm Hg at a single visit or >140 mm Hg at 2 successive visits. Down-titration was recommended if SBP was <135 mm Hg on 2 successive visits or <130 mm Hg at a single visit. For the intensive SBP intervention arm, antihypertensive dose titration or addition of a new drug was indicated if SBP was ≥120 mm Hg.

Cognitive function assessment

Cognitive function assessment was performed at baseline (1-month visit), 20 and 40 months after randomization with 4 standard neuropsychological tests, the DSST, the Mini-Mental State Examination (MMSE),¹³ the Rey Auditory Verbal Learning Test (RAVLT),¹⁴ and the Stroop interference test (Table 1).^15,16 Consistent with the primary outcome of the main ACCORD MIND study, DSST, a test of psychomotor function and speed¹⁷ was also our primary outcome. MMSE, RAVLT, and Stroop interference tests were secondary outcomes.

Table 1.

Neuropsychological tests used in ACCORD MIND

Test	Domain of cognition tested	Description of the test
Digit Symbol Substitution Test (DSST)	Psychomotor function, speed, learning, and working memory	Participants are given a key grid of numbers 1–9 with each number paired with a unique symbol. Below the key is a test section with randomly assigned numbers and empty boxes below each number. The participants fill the empty boxes with the symbol corresponding to the number (from the key above). The score is the number of correct number-symbol matches achieved in 120 s.
Mini-Mental State Examination (MMSE)	Global cognition	A ≈5 min, 30-point questionnaire including tests for orientation, attention, memory, language used for screening, and sensitive for moderate cognitive changes.
Rey Auditory Verbal Learning Test (RAVLT)	Verbal learning and memory	Participants are read a list of 15 unrelated words repeated over 5 different trials that they are asked to repeat as many words as possible (immediate recall). After this, an interference list of 15 unrelated words is presented, after which the participant is asked to recall words from the original list. After 10 min, the participant is asked to repeat the words from the original list (delayed recall).
Modified Stroop Color-Word Test	Executive function, ability to inhibit cognitive interference	Participants first read aloud words denoting colors (printed in black), then name colored bars, and finally read words denoting colors (printed in an incongruent color). Forty words and bars were used. A limit of 120 s is set for first 2 tasks and 180 s for the third. Time required and number of errors are recorded. Lower score indicates better cognitive function

Abbreviation: ACCORD MIND, Action to Control Cardiovascular Risk in Diabetes Memory in Diabetes substudy.

Statistical analysis

We grouped the ACCORD MIND study participants by tertiles of baseline DBP and compared baseline characteristics across tertiles. We used 1-way ANOVA for numeric values and χ² test for categorial variables. We used 2 sample t-tests to compare mean follow-up SBP, DBP, MAP, and pulse pressure between the tertiles.

We used ANOVA with margins to assess the association between baseline DBP groups and change in the average of 20- and 40-month test scores, and for 20- and 40-month scores for DSST, MMSE, RAVLT, and Stroop interference with the randomized SBP intervention as a predictor variable. We used cubic spline terms for neuropsychological test scores with main effects for the SBP intervention and cubic spline terms for baseline DBP, plus interactions between the SBP intervention and the cubic spline terms, adjusting for age, gender, race, and glycemic control. We also used linear regressions to compare the effect of intensive SBP intervention on DSST (and other neuropsychological tests) between the lowest baseline DBP tertile and the upper 2 tertiles.

All analyses were performed in STATA 15.1 or SAS 9.4 using a 2-sided α = 0.05.

Data availability statement

Anonymized data not published within this article will be made available by request from any qualified investigator.

RESULTS

The mean age of the 1,430 participants in the current analysis was 63 ± 6 years, with 55% women and 66% White. Mean baseline SBP and DBP were 139 ± 16 and 76 ± 10 mm Hg, respectively. Mean DSST, MMSE, RAVLT, and Stroop scores were 52 ± 16, 27 ± 3, 7.5 ± 2.5, and 32 ± 16, respectively. Compared with the participants in the highest baseline DBP tertile (≥81 mm Hg), those in the lowest DBP tertile (≤72 mm Hg) were older, had more cardiovascular events and a longer duration of diabetes (Table 2). While DSST and MMSE scores were not different, those with lower baseline DBP appeared to have lower cognitive function as evidenced by lower RAVLT and higher Stroop interference scores (Table 2).

Table 2.

Baseline characteristics by baseline diastolic blood pressure (DBP) tertiles (n = 1,430)

	Tertile 1	Tertile 2	Tertile 3	P
	DBP 49–72	DBP 73–80	DBP 81–97
	n = 519	n = 448	n = 463
Age (y)	64.5 ± 6.2	62.5 ± 5.7	61.5 ± 5.2	<0.001
Female gender (%)	54	57	53	0.49
White race (%)	69	67	63	0.16
Highest level of education (%)				0.87
Less than high school graduate	14	13	13
High school graduate	24	27	26
Some college or technical school	38	35	36
College graduate or more	24	26	25
Current or history of smoking (%)	55	50	57	0.13
History of clinical CV events (%)	36	27	19	<0.001
History of depression (%)	26	26	27	0.87
Intensive BP arm (%)	53	50	51	0.61
Intensive glycemia arm (%)	48	51	48	0.54
Systolic blood pressure (mm Hg)	131 ± 14	138 ± 13	148 ± 15	<0.001
BMI (kg/m2)	32.0 ± 5.2	33.0 ± 5.3	33.5 ± 5.4	<0.001
Hemoglobin A1c (%)	8.2 ± 0.9	8.3 ± 1.0	8.3 ± 1.1	0.18
Duration of diabetes (y)	13 ± 8	11 ± 8	9 ± 6	<0.001
DSST score	52 ± 16	52 ± 16	53 ± 16	0.19
MMSE score	28 (26, 29)	28 (26, 29)	28 (26, 29)	0.18
RAVLT score	7.3 ± 2.5	7.6 ± 2.5	7.7 ± 2.5	0.03
Stroop interference score	30 (23, 40)	29 (22, 40)	28 (20, 38)	0.01

Abbreviations: BMI, body mass index; BP, blood pressure; CV, cardiovascular; DSST, Digit Symbol Substitution Test; MMSE, Mini-Mental State Examination; RAVLT, Rey Auditory Verbal Learning Test.

The achieved mean follow-up SBP was similar within the standard and intensive SBP arms across baseline DBP tertiles (Figure 2). The achieved mean follow-up DBP was lower in the lowest baseline DBP tertile compared with the highest baseline DBP tertile in both the standard (66 ± 6 vs. 77 ± 6 mm Hg, P < 0.001) and the intensive (62 ± 6 vs. 69 ± 6 mm Hg, P < 0.001) SBP arms. Similarly, the achieved mean follow-up MAP was lower in the lowest baseline DBP tertile compared with the highest baseline DBP tertile in the standard (88 ± 5 vs. 96 ± 5 mm Hg, P < 0.001) and the intensive (81 ± 6 vs. 87 ± 6 mm Hg, P < 0.001) SBP arms.

Figure 2.

Mean follow-up blood pressures in each tertile of baseline diastolic blood pressure (DBP) by intensive and standard systolic blood pressure (SBP) intervention. Boxplots display the median, 25th and 75th percentiles of participants follow-up blood pressure values for (a) SBP, (b) DBP, and (c) mean arterial pressure (MAP) in the different tertiles of DBP in the intensive and standard BP intervention arms.

Effects of the intervention on cognitive function by baseline DBP tertiles and the range of baseline DBP

In the entire cohort, the DSST scores decreased by −1.6 (95% CI −1.9, −1.2) from baseline to the average of 20- and 40-month scores. Each 5 mm lower baseline DBP was associated with a nonsignificant increase of 0.1 (95% CI −0.1, 0.3) in the DSST score. In the standard SBP arm the DSST scores decreased by −1.46 (95% CI −1.99, −0.93) and in the intensive SBP arm by −1.66 (95% CI −2.21, −1.16). There was no difference in randomized comparisons across the baseline DBPs in the 2 arms (−0.22; 95% CI −0.97, 0.52) (Figure 3). Interaction product term of baseline DBP as a continuous variable and SBP intervention was nonsignificant (P = 0.32). There was no evidence of effect modification in a spline regression analysis relating the range of baseline DBP with the effects of the intervention on DSST scores (Figure 4). Also, there was no evidence that SBP intervention in the lowest DBP tertile lowered DSST scores when compared with standard SBP arm (regression coefficient 0.03; 95% CI −1.15, 1.22 for randomized comparison). Furthermore, regression coefficients of the effects of the SBP intervention on DSST scores in the lowest and highest baseline DBP tertiles were not different (P = 0.67). These results were consistent at 20- and 40-month follow-up (Supplementary Figures S1 and S5 online).

Figure 3.

Forest plot of SBP intervention effects on the change in Digit Symbol Substitution Test (DSST) scores from baseline to average of 20- and 40-month scores in the entire cohort and within each baseline DBP tertile. Abbreviations: DBP, diastolic blood pressure; SBP, systolic blood pressure.

Figure 4.

Spline regression curves for baseline diastolic blood pressure (DBP) with systolic blood pressure (SBP) intervention effects on change in Digit Symbol Substitution Test (DSST) scores from baseline to average of 20- and 40-month scores in the entire cohort.

The results remained consistent for the MMSE, RAVLT, and Stroop inference scores. There was no evidence that the SBP intervention decreased cognitive function in the entire cohort or within randomized comparison by tertiles (Supplementary Figures S2–S4 and S6 online).

DISCUSSION

In this post hoc analysis of the ACCORD MIND study, we found that low baseline DBP did not modify the effect of SBP lowering on cognitive function. While cognitive function decreased with time in both the standard and intensive SBP intervention arms, there were no differences in randomized comparisons between intensive and standard BP arms for changes in scores of DSST, MMSE, RAVLT, and Stroop interference tests in the entire cohort or within each of the baseline DBP tertiles.

Overall, cognitive function declined in the study cohort. Lower baseline DBP was associated with a nonsignificant lower decline in cognitive function. The SBP intervention, despite lowering DBP even in participants with low baseline DBP, did not result in a higher decline in cognitive function test scores when compared with participants in the standard SBP arm with similar DBPs. The spline regression models showed that the effect of lowering SBP on cognitive function was not modified by baseline DBP. This was true across the wide range of baseline DBP in ACCORD MIND.

These results have clinical implications. These data indicate that SBP can be lowered even in patients with low DBP without adversely affecting cognition. Both high SBP and DBP are known risk factors for cognitive decline. However, low DBP is also associated with a faster decline in cognitive function.¹⁸ Despite multiple studies indicating benefits of lowering BP, aggressive SBP lowering is often limited in clinical practice, especially in patients with low DBP due to concerns for further lowering DBP and MAP. Theoretically, lowering SBP could decrease cerebral perfusion by decreasing MAP, thereby worsening cognition.^19–22 Recent studies, however, indicate that BP lowering, even in participants with severe small vessel cerebral disease²³ does not decrease cerebral perfusion.^24,25 This may be due to the ability to maintain cerebral perfusion even with lowering of SBP and MAP due to cerebral autoregulation. Our findings support these data. In fact, hesitancy in lowering SBP in patients with low DBP, could deprive patients of the benefits of intensive SBP lowering in reduction of cardiovascular events and mortality.²⁶ Future studies evaluating cerebral blood flow with lowering SBP in participants with low DBP can help understand cerebral autoregulation in these patients.

Participants with lower baseline DBP were older, and with higher prevalence of cardiovascular disease and a longer duration of diabetes. Patients with diabetes have a higher prevalence of cognitive impairment²⁷ and a faster decline in cognition,²⁸ especially with older age and a longer duration of diabetes²⁹ as were the participants with the lowest DBP. Among the 4 neuropsychological tests (DSST, MMSE, RAVLT, and Stroop interference test) used in ACCORD MIND, participants in the lowest DBP tertile performed worse on the RAVLT and the Stroop interference test, while there was no difference in baseline DSST and MMSE scores across baseline DBP tertiles. This may be because RAVLT and the Stroop interference scores³⁰ generally remain unaffected in patients with diabetes whereas MMSE, a screening test for dementia is less sensitive in detecting mild impairment in executive function,³¹ a domain preferentially affected in diabetes and vascular cognitive impairment.³²

Major strengths of this study include a large sample size with a wide range of DBP, a randomized comparison arms and assessment of cognitive function with multiple neuropsychological tests. A limitation of the study is the post hoc nature of the analysis, as the main study was not designed to assess this outcome. In addition, neuropsychological tests such as MMSE may not be sensitive enough to detect subtle changes in cognition and may thus diminish the power of the study.

In conclusion, in ACCORD MIND participants with type II diabetes, SBP lowering in participants with low baseline DBP did not adversely affect cognitive function. Although a risk factor for faster decline in cognition, low DBP does not modify the effect of lowering SBP on cognitive function.

FUNDING

This work was supported by the NIH grant K23AG055666 from the National Institute of Aging.

PRIOR PRESENTATIONS

The results were presented at the ASN week 2020.

DISCLOSURE

The authors declared no conflict of interest.