Subscriptions: Information about subscribing to Hypertension is online at

Transcription

1 Essential hypertension and cognitive function. The role of hyperinsulinemia J Kuusisto, K Koivisto, L Mykkanen, EL Helkala, M Vanhanen, T Hanninen, K Pyorala, P Riekkinen and M Laakso Hypertension 1993;22; Hypertension is published by the American Heart Association Greenville Avenue, Dallas, TX Copyright 1993 American Heart Association. All rights reserved. Print ISSN: X. Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Subscriptions: Information about subscribing to Hypertension is online at Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters Kluwer Health, 351 West Camden Street, Baltimore, MD Phone: Fax: journalpermissions@lww.com Reprints: Information about reprints can be found online at

2 771 Essential Hypertension and Cognitive Function The Role of Hyperinsulinemia Johanna Kuusisto, Keijo Koivisto, Leena Mykkanen, Eeva-Liisa Helkala, Matti Vanhanen, Tuomo Hanninen, Kalevi Pybrala, Paavo Riekkinen, Markku Laakso The relation between hypertension and cognitive function is not well established. Therefore, we examined cognitive function in a random sample of 744 nondiabetic elderly inhabitants of Kuopio, East Finland. Five brief neuropsychological tests known to be sensitive to cognitive impairment due to dementia the Mini-Mental State Examination (MMSE), the Trail Making Test (TMT), the Buschke Selective Reminding Test (BSR), Russell's Adaptation of the Visual Reproduction Test (HVR), and the Verbal Fluency Test (VFT) were used to evaluate cognitive function. The performance of the hypertensive group (n=378) was impaired in almost all test items compared with that of the normotensive group (n=366), but the difference between these two groups was statistically significant in 5 of 19 test items only. Moreover, within the hypertensive group, hyperinsulinemic (fasting plasma insulin > 17.9 mu/l) hypertensive subjects (n=57) scored worse than normoinsulinemic hypertensive subjects (n=321) in 16 of 19 test items and worse than the normotensive subjects in the same 16 of 19 test items. The difference between the hyperinsulinemic hypertensive and normotensive groups was significant in 11 test items that reflected complex cognitive function such as calculation, language, semantic memory, and problem solving. This difference in neuropsychological tests among the three study groups (normotensive, normoinsulinemic hypertensive, and hyperinsulinemic hypertensive subjects) persisted after adjustment for fasting plasma glucose, age, sex, and education in 3 test items measuring calculation, copying, and semantic memory. Thus, essential hypertension in the elderly is associated with an impairment in complex cognitive function. Furthermore, hyperinsulinemia seems to identify a subgroup of hypertensive subjects with a particularly poor performance in neuropsychological tests requiring complex cognition such as semantic memory, problem solving, and abstraction. (Hypertension. 1993^22: ) KEY WORDS hypertension, essential cognition neuropsychological tests hyperinsulinemia Hypertension is known to be associated with an increased risk for cerebrovascular disease, including vascular dementia. However, the role of hypertension in cognitive function is not clear. Some previous studies 14 but not all 5 ' 6 have implied that neuropsychological performance is impaired in hypertension. Moreover, the only population-based studies on the topic have not verified the association between hypertension and impaired cognitive function. 5-6 In recent years it has been recognized that insulin resistance is a common feature of hypertension. 78 Of all hypertensive individuals, 30% to 50% are insulin resistant and hyperinsulinemic. 9 Theoretically, hyperinsulinemia could influence cognitive function by at least two mechanisms. First, hyperinsulinemia 10 and insulin resistance" are associated with an increased risk for atherosclerosis, and increased occurrence of cerebrovascular disease in hyperinsulinemic subjects could lead to impaired cognitive function. Second, it is also known that insulin is found in the brain, where it acts as a neuromodulator inhibiting synaptic activity. 12 In a recent study, 14 insulin reversibly reduced choline acetyltrans- Rcceived October 13, 1992; accepted in revised form June 25, From the Departments of Medicine (J.K., L.M., K.P., M.L.) and Neurology (K.K., E-L.H., M.V., T.H., P.R.), Kuopio University Hospital, Kuopio, Finland. Correspondence to Markku Laakso, MD, Department of Genetics, SK-50, University of Washington, Seattle, WA ferase activity in striatal neuron cultures, suggesting that insulin downregulates choline acetyltransferase. In rodent models, insulin also accelerates turnover of norepinephrine and other monoamines in the brain On the basis of these findings, it is reasonable to assume that insulin affects cognitive function. Indeed, high peak insulin response to glucose load has been found to correlate negatively with learning ability in rats. 18 However, cognitive function in humans in regard to variation in fasting insulin has not previously been studied. Therefore, in our population-based study, we investigated cognitive function in normotensive, normoinsulinemic hypertensive, and hyperinsulinemic hypertensive subjects using a wide battery of brief neuropsychological tests known to be sensitive in diagnosing impairment in cognitive function. Methods The baseline cross-sectional study was conducted in Kuopio, East Finland, between 1986 and The 1300 subjects who participated in the study at baseline were randomly selected from the Kuopio population aged 65 to 74 years. The formation of the original study population has been previously described in detail. 19 This follow-up study was performed during , on average 3.5 years (2.7±5.2 years) after the baseline cross-sectional study. After the baseline study 108 subjects had died. From 1192 subjects eligible for follow-up, 979 eventually participated in the follow-up study in

3 772 Hypertension Vol 22, No 5 November 1993 Kuopio University and Kuopio University Hospital between March 1990 and June All diabetics (n=206) and subjects with previous stroke (n = 29) were excluded from the present study, so the final study population comprised 744 participants. All study subjects gave informed consent, and the study was approved by the ethics committee of Kuopio University Hospital. During the follow-up study, history of smoking habits and alcohol consumption was collected by a standardized interview. A standard electrocardiogram was taken at both the baseline and follow-up and coded according to the Minnesota Code. 20 A myocardial infarction was diagnosed if a major Q wave (Minnesota Codes 1.1 and 1.2) was found. If a history of severe chest pain, known or suspected myocardial infarction, or symptoms suggesting stroke was recorded during the interview at the baseline or follow-up, the hospital records were checked by a physician, and myocardial infarctions were classified according to World Health Organization (WHO) criteria. 21 At the baseline and follow-up, a stroke was recorded if, according to hospital records, a sudden neurological deficit lasting more than 24 hours was found by a physician and no other reason than stroke for the deficit existed (WHO criteria). 22 At the follow-up study, blood pressure was measured with subjects in a supine position with a standard mercury sphygmomanometer. Systolic and fifth-phase diastolic pressures were measured to the nearest 2 mm Hg in the right arm after 5 minutes of bed rest. Measurement was repeated after 1.5 minutes, and the second measurement was used in statistical analyses. A subject was considered to be hypertensive if he or she was receiving antihypertensive medication, systolic blood pressure was 160 mm Hg or higher, or diastolic blood pressure was 95 mm Hg or higher. At the follow-up study, the glucose tolerance status was measured in a 2-hour oral glucose tolerance test (75 g glucose in 10% solution) and classified according to WHO diagnostic criteria for diabetes mellitus and impaired glucose tolerance. 23 Venous blood samples for determinations of fasting plasma insulin were taken before the glucose load. Plasma glucose was determined by the glucose oxidase method (Glucose Auto & Stat HGA-1120 analyzer, Daichii, Kyoto, Japan). Plasma insulin was determined by radioimmunoassay (Phadeseph Insulin RIA 100, Pharmacia Diagnostics AB, Uppsala, Sweden). Cognitive function was evaluated 2 to 3 weeks after the clinical and laboratory investigations of the follow-up study. Cognitive function was measured by a psychologist, physician, or trained nurse during a single visit (taking about 1.5 hours) with the use of five neuropsychological tests: the Mini-Mental State Examination (MMSE), 24 Russell's Adaptation of the Visual Reproduction Test (HVR),^ the Trail Making Test (TMT), 26 the Verbal Fluency Test (VFT), and the Buschke Selective Reminding Test (BSR). 29 The MMSE assesses a combination of cognitive functions: orientation, registration, attention and calculation, recall, copying, language, and constructional ability. 24 Each item of the MMSE evaluates its own aspect of cognitive function. The total score of the MMSE is a sum of all other items of the test. Other neuropsychological tests were selected to assess visual, episodic and semantic memory, and abstraction and problem-solving ability. The HVR and BSR tests are memory tests and have been shown to be sensitive to memory impairment due to dementia The HVR consists of three items, assessing short- and long-term visual memory and copying after a visual model. The BSR consists of two items, one of which evaluates short-term and another long-term episodic verbal memory. The VFT, consisting of four items, grades the storage of semantic memory retrieval from semantic memory, and ability to sustain behavioral output. 32 The VFT has been shown to be impaired in different dementias and frontal lobe dysfunctions. 31 The TMT, which consists of three items, evaluates the ability to solve problems under time pressure. It also requires visuoperceptual and psychomotor abilities. The TMT parts B and C especially measure response inhibition and vulnerability to interference. The TMT (parts A and B) has been shown to be sensitive to impairment in cognitive function and inflexibility of response. 34 Neuropsychological tests used in this study have previously been validated 2434 and considered to evaluate different aspects of cognition. Furthermore, factor analysis was performed including all above-mentioned tests (with the exception of the MMSE, which by definition assesses a combination of different aspects of cognition). Principal-components analysis extracted three factors, explaining 65% of total variability, from a correlation matrix of test items (data not shown). After varimax rotation, the HVR items were positively grouped together and the TMT items negatively in factor 1, the VFT items in factor 2, and the BSR items in factor 3 (correlations between a factor and a test item >.5 or <-.5). Therefore, each test is presented separately in the "Results" section. Statistical analyses were conducted with the SPSSX and SPSS/PC+ programs. The results for continuous variables are given as a mean±sem. The differences between two groups were assessed by the Student's t test or x 2 test when appropriate. Correlations were calculated by using Pearson correlation coefficients. Analysis of variance was used in testing the differences between more than two group means. Adjustments were done by analysis of covariance (ANCOVA). Results Table 1 shows the characteristics of the study subjects. Of 744 nondiabetic subjects without a history of stroke, 366 were normotensive and 378 hypertensive. The normotensive and hypertensive groups were comparable with respect to education, but the percentage of women was greater in the hypertensive group than in the normotensive group, and the hypertensive subjects were slightly older. The prevalence of current smokers was greater in the normotensive group; the two groups did not differ in the prevalence of current alcohol users. The prevalence of a previous myocardial infarction did not significantly differ between the normotensive and hypertensive groups. The prevalence of abnormal electrocardiogram was greater in the hypertensive group, mainly because of the greater prevalence of left ventricular hypertrophy (15.6% versus 5.7%). There was no difference in the prevalence of atrial fibrillation or major Q waves (Minnesota Codes 1.1 and 1.2) between the normotensive and hypertensive groups (data not shown). Impaired glucose tolerance was more prevalent in the hypertensive than the normotensive group. In addition, fasting plasma glucose, 2-hour plasma glucose

4 Kuusisto et al Hypertension and Cognitive Function 773 TABLE 1. Characteristics of Study Subjects by Hypertension Status Characteristic Men/women Age, y Education, y Current smokers, % Alcohol users, % Systolic blood pressure, mm Hg Diastollc blood pressure, mm Hg Use of /3-blockers, % Use of diuretics, % Use of calcium antagonists, % Myocardial infarction, % Abnormal electrocardiogram, % Impaired glucose tolerance, % Fasting plasma glucose, mmol/l 2-Hour plasma glucose, mmol/l Fasting plasma insulin, mu/l No Hypertension (n=366) 155/ ± ± ± ± ± Values are mean+sem. *P<.001, tp<05, *P<.01, detennined with Student's f test or x 2 test. A history of myocardial infarction according to electrocardiogram or hospital records. Hypertension (n=378) 116/262* 73.1±0.2t 6.7± * ±1.0* 86.2±0.5* 34.1* 28.6* 15.1* * 5.7±0.0* 7.0±0.1* * in the 2-hour glucose tolerance test, and fasting plasma insulin were higher in the hypertensive group. To investigate the role of hypertension and plasma glucose and insulin in cognitive function in the whole study population, we calculated the correlations of systolic and diastolic blood pressures, fasting plasma glucose, and fasting plasma insulin with neuropsychological test scores (Table 2). The test scores correlated significantly and inversely with current systolic blood pressures in 13 of 19 test items. The correlations were strongest in the MMSE (the attention and calculation part, total score) and in tests assessing visual, episodic verbal and semantic memory, and problem solving. The test scores correlated significantly also with diastolic blood pressure in 9 of 19 test items. The correlations between fasting plasma glucose and the test scores were statistically significant in 6 test items only. Interestingly, fasting insulin correlated inversely with neuropsychological test scores in 12 test items, showing almost as strong an association with cognitive tests as systolic blood pressure. The correlations between the neuropsychological tests and current blood pressure and parameters of glucose metabolism were studied separately in the hypertensive group (only correlations with fasting insulin are shown in Table 2). Among the hypertensive subjects, current systolic blood pressure correlated significantly with neuropsychological test scores in one test item only (VFT, animal category, r=.12, P<.05) and with current diastolic blood pressure in two test items only (MMSE, language, r=-.19, / > <.O1 and VFT, A words, r=-.ll, P<.05). Correlations with fasting glucose were significant in one test item only (TMT, part C, r=.ll, / > <.05). Table 3 shows the neuropsychological test scores separately in the normotensive and hypertensive groups. The hypertensive group scored worse than the normotensive group in all neuropsychological test items with the exception of those in orientation, registration, and copying parts of the MMSE and short-term memory in the BSR. The difference between the normotensive and hypertensive groups was statistically significant in the attention and calculation item of the MMSE, the immediate recall in the HVR, on letter P in the VFT, and parts A and C in the TMT. Moreover, when adjusted for age, sex, education, and fasting plasma glucose, the difference between the normotensive and hypertensive groups remained statistically significant in the following test items: attention and calculation in the MMSE (F=.O14) and the VFT on letter P (/>=.O29). The hypertensive group was divided into two subgroups according to fasting plasma insulin. The hyperinsulinemic hypertensive group (n=57) was defined as those hypertensive subjects with fasting plasma insulin more than one standard deviation above the whole study population mean (11.5±6.4 mu/l=17.9 mu/l). Other hypertensive subjects were regarded as normoinsulinemic (n=321). Fasting plasma insulin levels were 10.2±0.2 mu/l (mean±sem) in the normoinsulinemic hypertensive group and 25.2±1.0 mu/l in the hyperinsulinemic hypertensive group. The normoinsulinemic hypertensive and hyperinsulinemic hypertensive groups were comparable with respect to age (73.0±0.2 versus 73.6±0.4 years), education (6.8±0.2 versus 5.9±0.4 years), prevalence of smokers (5.9% versus 0.0%) and alcohol users (19.6% versus 8.8%), prevalence of a previous myocardial infarction (14.6% versus 14.5%), and prevalence of pathological electrocardiogram including atrial fibrillation, left ventricular hypertrophy, and pathological Q waves (data not shown). Use of /3-blockers and diuretics was equally prevalent in both hypertensive groups (33.3% versus 38.6%, f=, and 26.8% versus 38.6%, P=). The normoinsulinemic and hyperinsulinemic hypertensive groups differed from

5 774 Hypertension Vol 22, No 5 November 1993 TABLE 2. Pearson Correlation Coefficients of Systolic and Diastollc Blood Pressures and Fasting Plasma Glucose and Insulin With Neuropsychologlcal Test Scores In the Whole Study Population Cognitive Subsection MMSE Orientation Registration Attention and calculation Recall Copying Language Total Visual memory and construction HVR, copying HVR, immediate HVR, delayed Episodic verbal memory BSR, short-term BSR, long-term Semantic memory VFT, number of P words S words A words VFT, animal category Problem solving and abstraction TMT, seconds Part A Part B PartC Systolic o.iot 0.03 * -1.00t -0.09* 0.08* -1.15t -out -o.iot 0.08* 0.10* 0.14t Blood Pressure Dlastolic * t -0.13t * out -0.12t -o.iot t Fasting Glucose * o.iot t All t -0.09* * -0.12* -0.10* * -o.iot -0.09* 0.13t t Fasting Insulin Hypertensives* * * -0.13* -0.17t * -0.14t -0.11* MMSE indicates Mini-Mental State Examination; HVR, Russell's Adaptation of the Visual Reproduction Test; BSR, Buschke Selective Reminding Test; VFT, Verbal Fluency Test; and TMT, Trail Making Test. *Correlations in the hypertensive group. tp<.01,*p< t * each other in sex distribution (men/women ratio, 105/ 216 versus 11/46, P=.O4) and systolic blood pressure (168.3±1.1 versus 174.3±3.1 mm Hg, F=.O4) but not in diastolic blood pressure (86.2±0.5 versus 85.7±1.2 mm Hg, P=). Impaired glucose tolerance was more frequent among the hyperinsulinemic hypertensive subjects than the normoinsulinemic hypertensive subjects (57.9% versus 28.0%, P<.001). Similarly, the mean fasting plasma glucose and 2-hour plasma glucose levels were higher in the hyperinsulinemic group (5.6±0.0 versus 6.1 ±0.9 mmol/l, P<.001, and 6.8±0.1 versus 8.0±0.2 mmol/l, /><.001). Table 3 gives the results of neuropsychological tests separately for the normoinsulinemic and hyperinsulinemic hypertensive subjects. The normoinsulinemic hypertensive group scored almost equally compared with the normotensive group. The only statistically significant differences between these two groups were found in the attention and calculation item of the MMSE and the C part of the TMT. Normoinsulinemic hypertensive subjects scored even significantly better than normotensive subjects in the copying part of the MMSE. In contrast, the hyperinsulinemic hypertensive subjects scored worse than the normoinsulinemic hypertensive subjects in all but 3 of 19 neuropsychological test items and also worse than the normotensive subjects in these same test items. The difference in test scores between the normotensive and hyperinsulinemic hypertensive groups was statistically significant in the MMSE (attention and calculation item, language item, and total score), long-term memory of the BSR, and all parts of the VFT and TMT. Finally, after adjustment for age, sex, education, and fasting glucose by ANCOVA, the difference between the normotensive, normoinsulinemic hypertensive, and hyperinsulinemic hypertensive groups remained statistically significant in two items of the MMSE (the attention and calculation items and language item) and in the VFT on letter P. In the Figure, scores are shown adjusted for age, sex, education, and fasting glucose in the attention and calculation items of the MMSE, parts A and B in the TMT, and the VFT on letter P in the normotensive,

6 Kuusisto et al Hypertension and Cognitive Function 775 TABLE 3. Neuropsychologlcal Test Scores In Normotensive, All Hypertensive, Normolnsulinemic Hypertensive, and Hyperlnsullnemlc Hypertensive Subjects Cognitive Subsection MMSE Orientation Registration Attention and calculation Recall Copying Language Total Normotenslves (n=366) 9.8± ± ± ±0.1 Visual memory and construction HVR, copying 15.5±0.1 HVR, Immediate 9.1 ±0.2 HVR, delayed 6.1 ±0.2 Episodic verbal memory BSR, short-term 8.4±0.3 BSR, long-term 24.2±0.6 Semantic memory VFT, number of P words 11.5±0.3 S words 11.4±0.3 A words 9.2±0.2 VFT, animal category 17.0±0.3 Problem solving and abstraction TMT, seconds Part A PartB 187.5±5.1 PartC 169.0±4.2 Hypertensives (n=378) 9.8± ±0.1t 1.6± ± ± ± ± ± ± ± ± ±0.3t 11.0± ± ± ± ± ±4.4 No (n=321) 9.8± ±0.1t 1.7± ± ± ± ± ± ± ± ± ± ± ± ± ± ±4.8 Hypertensives, Hyperlnsullnemla Yes (n=57) 9.9± ±0.2t ± ± ±0.4t 15.0± ± ± ± * 9.2±0.6* 9.6±0.7t 8.0± ± ± ±11.2t ANCOVA P Value* ANCOVA indicates analysis of covariance; MMSE, Mini-Mental State Examination; HVR, Russell's Adaptation of the Visual Reproduction Test; BSR, Buschke Selective Reminding Test; VFT, Verbal Fluency Test; and TMT, Trail Making Test. Values are mean±sem. *Analysis of covariance was performed over the normotensive, normoinsullnemic hypertensive, and hyperinsulinemic hypertensive groups (covariates: age, sex, education, and fasting blood glucose). tp<-001, $P<.01, P<.05, compared with normotensive group, Student's Mest normoinsulinemic hypertensive, and hyperinsulinemic hypertensive groups. After the adjustment, there were no significant differences between the normotensive and normoinsulinemic hypertensive groups. Again, the hyperinsulinemic hypertensive group scored significantly worse than the normotensive group in the attention and calculation items of the MMSE and the VFT on letter P. Table 4 shows the effect of antihypertensive medication on neuropsychological tests. Hypertensive subjects with /3-blocker medication scored better or equally compared with hypertensive subjects without antihypertensive medication in all but one test item, but the difference was statistically significant only in the attention and calculation and copying parts of the MMSE. Hypertensive subjects receiving diuretics scored worse than the hypertensive subjects without antihypertensive treatment in four items of the MMSE, all items of the HVR, one item of the BSR, one item of the VFT, and all items of the TMT. The group receiving diuretic therapy scored better than the group without antihypertensive treatment in one item of the BSR and in the VFT animal category. However, the difference between the diuretic group and the group without antihypertensive medication was significant in one test item only (TMT, part C). After adjustment for age, sex, education, and glucose tolerance status, the difference between hypertensive subjects without antihypertensive medication, with /3-blockers, and with diuretics remained statistically significant in the copying item and total score of the MMSE and in the items of copying and immediate recall of the HVR. Discussion The pathogenesis of hypertension, especially its association with hyperinsulinemia and insulin resistance, has recently been a target of keen investigation. 79

7 776 Hypertension Vol 22, No 5 November 1993 MN-MENTAL STATE EXAMMAHON (attention, calculation) TRAIL MAKING TEST (part A) TRAIL MAKING TEST (part B) VERBAL FLUENCY TEST (P-words) I I Normotentives i H Hypertens/normoins I B Hypertens/hyperins Bar graphs show neuropsychological screening test scores adjusted for age, sex, education, and fasting plasma glucose in normotensive, normoinsulinemic hypertensive (Hypertens/normoins), and hyperinsulinemic hypertensive (Hypertens/ hyperins) groups. *P<.05, compared with normotensive group by analysis of covariance. Essential hypertension has been found to be linked with hyperinsulinemia and insulin resistance in 30% to 50% of hypertensive individuals. 9 Although the association between essential hypertension and impaired cognitive function has previously been reported, 14 no studies have directly dealt with cognitive function in normoinsulinemic and hyperinsulinemic hypertensive subjects. Thus, the main result of the present study indicating that hyperinsulinemia identifies a subgroup of hypertensive subjects with a particularly poor cognitive function is a novel finding. Cognitive function in hypertension has been found to be impaired in some 14 but not in all 5 ' 6 previous clinical studies. This population-based study confirmed that cognitive function is disturbed in elderly hypertensive subjects. However, the cognitive disturbance in hypertension was largely confined to hyperinsulinemic hypertensive subjects. In fact, the difference between the normotensive and normoinsulinemic hypertensive groups in cognitive tests was practically insignificant. Moreover, the difference between normotensive and hypertensive subjects was not evident in all neuropsychological tests. Cognitive disturbance in hypertension was found in tasks requiring attention and calculation, problem solving, and abstraction, all complex cognitive functions. Finally, in contrast to previous studies, the study subjects were elderly and not middle-aged. All these factors may account for the lack of consistency with other studies. Why is cognitive function disturbed in essential hypertension? Several explanations have been proposed. Hypertension is known to be associated with an increased risk for stroke. However, in the present study, subjects with clinical stroke were excluded, so classic strokes do not explain the findings of this study. Hypertension is also known to be associated with two other types of cerebrovascular disease lacunar brain infarcts caused by occlusion of perforating arteries and diffuse white matter demyelination associated with arteriolosclerosis. 35 These latter two forms of cerebrovascular disease are often silent without overt clinical neurological symptoms or findings. On the other hand, they are the leading cause of vascular dementia. 35 In a recent study, impaired cognitive performance in hypertension was associated with confluent brain white matter lesions on magnetic resonance imaging. 3 * These white matter lesions have been found to represent mainly diffuse demyelination of the white matter and only in a small part lacunar infarcts. 37 Interestingly, in the study of van Swieten et al, 3 * the MMSE and TMT scores, also particularly impaired in the hypertensive group in the present study, were worse in the hypertensive group with confluent white matter lesions than in the normotensive group and hypertensive group without confluent white matter lesions. In the present study, magnetic resonance imaging or computed tomographic scanning of the brain was not performed, preventing further confirmation of the role of white matter lesions

8 Kuusisto el al Hypertension and Cognitive Function 777 TABLE 4. Neuropsychologlcal Test Scores in Hypertensive Subjects Without AntI hypertensive Medication or WHh 0-Blockers or Diuretics Cognitive Subsection MMSE Orientation Registration Attention and calculation Recall Copying Language Total Visual memory and construction HVR, copying HVR, immediate HVR, delayed Episodic verbal memory BSR, short term BSR, long term Semantic memory VFT, number of P words S words A words VFT, animal category Problem solving and abstraction TMT, seconds Part A Part B Part C No (n=130) 9.8± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±7.7 Antihypertensive Medication 0-Blockere (n=64) 9.9± ±0.1t 1.8± ±0.0* 7.3± ± ±0.3t 9.3±0 4t 6.1 ± ± ± ± ± ± ± ± ±9.9 Diuretics (n=40) 9.7± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±15.2+ ANCOVA P Value* ANCOVA indicates analysis of covariance; MMSE, Mini-Mental State Examination; HVR, Russell's Adaptation of the Visual Reproduction Test; BSR, Buschke Selective Reminding Test; VFT, Verbal Fluency Test; and TMT, Trail Making Test. Values are mean±sem. *Analysis of covariance was performed over the group wrthout antihypertensive medication, the 0-blocker group, and the diuretic group (covariates; age, sex, education, and glucose tolerance status). tp<-05, $P<.01, compared with normotensive group, Student's f test and lacunar infarcts in impaired cognitive function in hypertension. Sex and especially age and education affect cognitive performance Could these demographic factors explain the cognitive difference between the normotensive and hypertensive subjects? First, in the present study, women formed the majority of hypertensive subjects. However, in all ANCOVAs, sex was a covariate, excluding the effect of sex on the results. In addition, there were no systematic differences in neuropsychological test results between the genders in the present study; moreover, women performed better than men in many tests (data not shown). Also, in a recent study by Koivisto et al 3 * 1 on an elderly population, the effect of sex on the performance in a neuropsychological test battery similar to the one used in this study was found only in some test items of the battery. Second, the hypertensive group was slightly older than the normotensive group; however, age adjustment did not abolish the difference between the study groups. Third, no difference in education between the normotensive and hypertensive groups was found in the present study. Differences in smoking and drinking habits 35 might also explain cognitive differences between normotensive and hypertensive subjects. In the present study, smoking and alcohol consumption were less prevalent among hypertensive subjects and therefore cannot explain the impairment in cognitive performance among hypertensive subjects. Some studies 40 have implied that antihypertensive medication might worsen cognitive function. In the present study, the use of /3-blockers and diuretics was more prevalent among hypertensive subjects and in the hypertensive group was slightly more common in the hyperinsulinemic than in the normoinsulinemic hypertensive subjects. However, as shown in Table 3, antihy-

9 778 Hypertension Vol 22, No 5 November 1993 pertensive medication did not systematically worsen cognitive function. In fact, hypertensive subjects receiving /3-blockers scored better than those receiving no antihypertensive medication in most cognitive tests, a finding consistent with a few previous studies Diuretic therapy was associated with only a nonsignificant tendency toward impaired cognitive performance. Moreover, cognitive test items most impaired in hyperinsulinemic hypertensive subjects were not significantly affected by /3-blocker or diuretic therapy. In fact, scores in tasks requiring semantic memory and problem solving were even better in the hypertensive group receiving 0-blocker therapy than in the hypertensive group without antihypertensive medication. Thus, antihypertensive medication does not seem to explain the difference in cognitive function between the normotensive and hypertensive groups. Finally, impaired glucose tolerance is known to be associated with hypertension. 9 In the present study, impaired glucose tolerance was more prevalent and fasting plasma glucose and fasting plasma insulin were higher in the hypertensive than in the normotensive group. In fact, high plasma insulin identified a hypertensive subgroup with a particularly impaired performance in cognitive tests. This difference in cognition between normoinsulinemic and hyperinsulinemic hypertensive subgroups remained significant even after adjustment for fasting glucose, suggesting that insulin, not glucose, is an important factor affecting cognitive function in essential hypertension. Among the hypertensive subjects, fasting glucose correlated significantly with one neuropsychological test item only (data not shown), and fasting insulin correlated significantly with nine items (Table 2), which further supports the importance of insulin. The normoinsulinemic and hyperinsulinemic hypertensive groups did not differ with respect to age, education, smoking or drinking habits, or clinical manifestations of atherosclerosis. Although mean systolic blood pressure was somewhat higher in the hyperinsulinemic hypertensive than the nonnoinsulinemic hypertensive group, among the hypertensive group the neuropsychological test scores correlated significantly with current blood pressure in one test item only. This suggests that some factors other than current blood pressure predict cognitive function in hyperinsulinemic hypertensive subjects. How could high insulin level affect cognitive function? On the basis of this study, the association between hyperinsulinemia and impaired cognitive function in hypertension cannot be explained. However, at least two possibilities for this association can be proposed. Hyperinsulinemia is associated with an increased risk for atherosclerosis, particularly coronary heart disease Moreover, in a recent study, asymptomatic atherosclerosis in both femoral and carotid arteries has been found to be associated with insulin resistance measured by the euglycemic glucose clamp technique. 11 In our study, a history of myocardial infarction was not more frequent among the hyperinsulinemic hypertensive than the normoinsulinemic hypertensive subjects, and subjects with clinical stroke were excluded. However, as mentioned above, the two other manifestions of cerebrovascular atherosclerotic disease lacunar infarcts and white matter demyelination are often clinically silent. Because subjects of the present study did not undergo magnetic resonance imaging or computed tomography, the possibility cannot be confirmed or excluded that the poorer cognitive performance in hyperinsulinemic hypertensive compared with normoinsulinemic hypertensive subjects is due to a higher rate of silent lacunar infarcts or white matter lesions caused by accelerated atherosclerosis associated with hyperinsulinemia or insulin resistance. Insulin itself can also be found in the brain, where it acts as a neuromodulator inhibiting synaptic activity. 12 Insulin receptors have been found in the hypothalamic 46 and hippocampal 47 areas. It is believed that insulin from plasma has access to the brain at circumventricular areas lacking the blood-brain barrier. Insulin is also transported across the blood-brain barrier via specific receptors to enter neural tissue directly or is taken up into neural tissue from cerebrospinal fluid. 12 In recent studies, insulin has also been found to reversibly reduce cholinergic activity of striatal neuron cultures 14 and to accelerate turnover of monoamines in the brain It is also known that cholinergic neurons are destroyed in Alzheimer's disease, leading to dementia. These facts suggest that insulin might interfere with cognitive function, possibly directly as a synaptic inhibitor, by decreasing cholinergic activity or by affecting monoamine metabolism in the brain. If insulin affects cognition, does our cutoff point of more than one standard deviation above the mean fasting insulin in the whole population reflect hyperinsulinemia? First, the distribution of insulin levels in old populations has not been systematically studied. Second, normal values are dependent on the insulin assay used. Subsequently, the criteria for hyperinsulinemia drawn from other studies are difficult to justify. Instead, one standard deviation above mean fasting insulin was calculated in our study population (excluding diabetics and subjects with stroke) and selected as the cutoff point. The results of our study remained essentially unchanged if the cutoff point was defined as a median fasting insulin in the whole population, which indicates that even a milder degree of hyperinsulinemia (>10.0 mll/l) affects cognitive function. In our previous study investigating the association of insulin with insulin resistance, 48 insulin levels were measured in the same laboratory with the same method as in the current study. In that previous study, in the age group of 45 to 74 years (mean, 65 ± 1 years), more than 90% of subjects with normal or impaired glucose tolerance and fasting insulin above the cutoff point of 17.9 mu/l used in the current study belonged to the most insulin-resistant fertile. 48 Thus, insulin levels greater than 17.9 mu/l represent hyperinsulinemia. In conclusion, our large population-based study showed that in elderly hypertensive subjects, cognitive functions requiring attention, calculation, semantic memory, and problem solving are impaired. In addition, hyperinsulinemia identifies a subgroup with particularly impaired cognition. Although hyperinsulinemia is associated with atherosclerotic complications and changes in brain neurotransmitters, the possible mechanisms explaining the association between hyperinsulinemia and cognitive function cannot be solved on the basis of this study and require further investigations. Acknowledgment Supported by a grant from the Medical Research Council of the Academy of Finland.

10 Kuusisto et al Hypertension and Cognitive Function 779 References 1. Wilkie F, Eisdorfer C. Intelligence and blood pressure in the aged. Science. 1971;172: Boiler F, Vrtunski PB, Mack JL, Kim Y. Neuropsychological correlates of hypertension. Arch NeuroL : Elias MF, Robbins MA, Schultz NR Jr, Streeten DHP, Elias PK. Clinical significance of cognitive performance by hypertensive patients. Hypertension. 1987;9: Waldstein SR, Manuck SB, Ryan CM, Parkinson DK, Bromet EJ. Learning and memory function in men with untreated blood pressure elevation. J Consult Clin PsychoL 1991^9: Farmer ME, White LR, Abbott RD, Kittner SJ, Kaplan E, Wolz MM, Brody JA, Wolf PA. Blood pressure and cognitive performance: the Framingham study. Am J Epidemiol. 1987;126: Farmer ME, Kittner SJ, Abbott RD, Wolz MM, Wolf PA, White LR. Longitudinally measured blood pressure, antihypertensive medication use, and cognitive performance: the Framingham study. J Clin Epidemiol 1990;43: Welborn TA, Breckenridge A, Rubinstein AH, Dollery CT, Fraser TR. Serum-insulin in essential hypertension and in peripheral vascular disease. Lancet. 1966;1: Ferrannini E, Buzzigoli G, Bonadonna R, Giorico MA, Oleggini M, Graziadei L, Pedrinelli R, Brandi L, Bevilacqua S. Insulin resistance in essential hypertension. N Engl J Med. 1987;317: Black HR. The coronary artery disease paradox: the role of hyperinsulincmia and insulin resistance and implications for therapy. J Cardiovasc Pharmacol. 1990;15(suppl 5): Stout RW. The relationship of abnormal circulating insulin level to atherosclerosis. Atherosclerosis. 1977;27: Laakso M, Sarlund H, Salonen R, Suhonen M, Py6ralS K, Salonen JT, KarhapSa P. Asymptomatic atherosclerosis and insulin resistance. Arteriosclerosis. 1991;ll: Baskin DG, Figlewicz DP, Woods SC, Porte D Jr, Dorsa DM. Insulin in the brain. Annu Rev Physiol 1987;49: LeRoith D, Hendricks SA, Lesnaik MA, Rishi S, Becker KL. Insulin in brain and other extrapancreatic tissues of vertebrates and non-vertebrates. Adv Metab Disorders. 1983;10:3O4-34O. 14. Brass BJ, Nonner D, Barrett JN. Differential effects of insulin on choline acetyltransferase and glutamic acid decarboxylase activities in neuron-rich striatal cultures. J Neurochem, 1992;59: Kwok RPS, Juorio AW. Facilitating effect of insulin on brain 5-hydroxytryptamine metabolism. Neuroendocrinology. 1987;45: Kwok RPS, Juorio AW. The effect of insulin on rat brain noradrenaline. Neurochem Res. 1988;13: Sauter A, Goldstein M, Engel J, Ueta K. Effect of insulin on central catecholamines. Brain Res. 1983;260: Long JM, Davis BJ, Garofalo PG, Spangler EL, Ingram DK. Complex maze performance in young and aged rats: response to glucose treatment and relationship to blood insulin and glucose. Physiol Behav. 1992^1: MykkSnen L, Laakso M, Uusitupa M, PyoralS K. Prevalence of diabetes and impaired glucose tolerance in elderly subjects and their association with obesity and family history of diabetes. Diabetes Care. 1990;11:1099-U Prineas RJ, Crow RS, Blackburn H. The Minnesota Code Manual of Electrocardiographic Findings. Bristol, UK: John Wright PSG Inc; World Health Organization. Proposal for the Multinational Monitoring of Trends and Determinants in Cardiovascular Disease and Protocol (MONICA-Project). Geneva, Switzerland: World Health Organization; WHO/MNC/82.1, Rev Weinfeld FD, ed. The national survey of stroke Stroke. 1981; 12(suppl I):I-32-I World Health Organization. Diabetes Mellitus: Report of a WHO Study Group. Geneva, Switzerland: World Health Organization; Tech Rep Ser, No Folstein MF, Folstein SE, McHugh PR. 'Mini-mental state': a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12: Lezak MD. Neuropsychological Assessment. 2nd ed. New York, NY: Oxford University Press; Reitan RM. Validity of the Trail Making Test as an indicator of organic brain damage. Percept Mot Skills. 1958;8: Borkowski JG, Benton AL, Spreen O. Word fluency and brain damage. Neuropsychologia. 1967;5:135-14O. 28. Butters N, Granholm E, Salmon DP, Grant I. Episodic and semantic memory: a comparison of amnesic and demented patients. J Clin Exp NeumpsychoL 1987;9: Buschke H, Altman Fuld P. Evaluating storage, retention and retrieval in disordered memory and learning. Neurology. 1974;24: Storandt M, Botwinick J, Danziger WL, Ber L, Hughes CP. Psychometric differentiation of mild senile dementia of the Alzheimer type. Arch NeuroL 1984;41: Miller BL, Cummings JL, Villanueva-Meyer J, Boone K, Mehringer CM, Lesser IM, Mena I. Frontal lobe degeneration: clinical, neuropsychological and SPECT characteristics. Neurology. 1991;41: Weintraub S, Mcsulam MM. Mental state assessment of young and elderly adults in behavioral neurology. In: Mesulam MM, ed. Principles of Behavioral Neurology. Philadelphia, Pa: FA Davis Co; 1985: Greencliff CL, Margolis RB, Erker GJ. Application of the Trail Making Test in differentiating neuropsychological impairment of elderly persons. Percept Mot Skills. 1985;61: Lezak MD. Executive functions and motor performance. In: Lezak M, ed. Neuropsychological Assessment. New York, NY: Oxford University Press; 1983: Meyer JS, McClintic KL, Rogers RL, Sims P, Mortel KF. Aetiological considerations for multi-infarct dementia. J Neurol Neurosurg Psychiatry. 1988^1: van Swieten JC, Geyskes GG, Derix MMA, Peeck BM, Ramos LMP, van Latum JC, van Gijn J. Hypertension in the elderly is associated with white matter lesions and cognitive decline. Ann NeuroL 1991;30: van Swieten JC, van der Hout JHW, van Ketel BA, Hijadra A, Wokke JHJ, van Gijn J. Periventricular lesions in the white matter on magnetic resonance imaging in the elderly. Brain. 1991;114: Koivisto K, Helkala E-L, Reinikainen KJ, Hanninen T, Mykkanen L, Laakso M, PyorSla K, Riekkinen PJ. Population-based dementia screening program in Kuopio: the effect of education, age and sex on brief neuropsychological tests. J Genatr Psychiatry Neurol. 1992^: O'Connor DW, Pollitt PA, Treasure FP, Brook CPB, Reiss BB. The influence of education, social class and sex on Mini-Mental State scores. Psycho! Med, 1989;19: Solomon S, Hotchkiss E, Saravay EM, Bayer C, Ramsay P, Blum RS, Blum RS. Impairment of memory function by antihypertensive medication. Arch Gen Psychiatry. 1983;4O:l Streufert S, DePadova A, McGlynn T, Pogash R, Piasecki BS. Impact of betablockade on complex cognitive functioning. Am Heart J. 1988;116: Shapiro AP, Nixon P, Miller RE, Manuck SB, Jennings R, King HE. Behavioural consequences of hypertension: effects of age and type of antihypertensive agent. J Hum Hypertens. 1989;3: Pyorala K. Relationship of glucose tolerance and plasma insulin to the incidence of coronary heart disease: results from two population studies in Finland. Diabetes Care. 1972;2: Ducimetiere P, Eschwege E, Papoz L, Richard JL, Claude JR, Rosselin G. Relationship of plasma insulin level to the incidence of myocardial infarction and coronary heart disease mortality in a middle-aged population. Diabetotogta. 1980;19: Welborn TA, Wearne K. Coronary heart disease incidence and cardiovascular mortality in Busselton with reference to glucose and insulin concentration. Diabetes Care. 1979;2: Shibata S, Liou SY, Ueki S, Oomura Y. Inhibitory action of insulin on suprachiasmatic neurons in rat hypothalamic slice preparation. Physiol Behav. 1985^6: Palovick RA, Phillips MI, Kappy MS, Raizada MI. Insulin inhibits pyramidal neurons in hippocampal slices. Brain Res : Laakso M. How good a marker is insulin level for insulin resistance? Am J Epidemiol 1993; 137: