Clinical Science (1992) 83, a 9 4 (Printed in Great Britain)
489
Ageing and the response of plasma insulin, glucose and Gpeptide concentrations to intravenous glucose in postmenopausal women Anthony J. PROUDLER, Carl V. FELTON and John C. STEVENSON Wynn Institute for Metabolic Research, London, U.K. (Received 18 December 199112 April 1992; accepted I I June 1992)
1. Eighty-six apparently healthy postmenopausal women not receiving hormone replacement therapy were given an intravenous glucose tolerance test. Plasma glucose, insulin and C-peptide concentrations were determined in fasting and post-glucose challenge samples. 2. Using a multivariate regression model, with predictor variables of chronological age, menopausal age and body mass index, neither chronological age nor menopausal age correlated with fasting or postchallenge plasma glucose or C-peptide concentrations. In contrast, menopausal age was positively associated with fasting plasma insulin concentration (P= 0.038, model r2=0.107), insulin area (P=0.01, model Vt =0.236) and incremental insulin area (P=0.024, model r2=0.243). This relationship could not be explained by differences in lifestyle variables of alcohol consumption, physical activity, previous duration of oral contraceptive usage, history of cigarette smoking or body mass index. 3. Our findings suggest that loss of ovarian function is associated with hyperinsulinaemia, possibly via alterations in the clearance of circulating insulin from the plasma. The hyperinsulinaemia observed may contribute to the increased risk of cardiovascular disease seen in postmenopausal women. INTRODUCTION
Loss of ovarian function appears to increase the risk of cardiovascular disease. Postmenopausal women have a higher incidence of cardiovascular disease than premenopausal women of the same age [1,2]. Furthermore, women with a bilateral ovariectomy who are not receiving oestrogen replacement therapy have an increased risk of coronary heart disease compared with premenopausal women [3, 41.However, the potential factors that may account for these effects have not been fully investigated. Menopause-asociated changes have been reported in risk factors for cardiovascular disease including total cholesterol, cholesterol lipoprotein fractions, triacylglycerol, apolipoprotein B, haemoglobin, fibrinogen, Factors VII, and VIII, antithrombin I11
antigen, a,-antitrypsin antigen and plasminogen activity [S-113. Few studies have investigated changes in carbohydrate metabolism in relation to the menopause. Natural and surgical menopause were not associated with a significant change in fasting plasma glucose concentration in the Framingham Heart study [S]. A similar result was obtained in a crosssectional study of pre- and post-menopausal women [7]. A longitudinal study of naturally menopausal women and age-matched premenopausal controls reported increases in fasting plasma insulin and glucose concentrations on follow up after 2.5 years. Since these increases were seen in both postmenopausal women and their controls, the results were interpreted as effects of ageing [S]. In order to examine further any interactions between chronological or menopausal age and carbohydrate metabolism, we have determined plasma glucose, insulin and C-peptide profiles after intravenous glucose tolerance testing in postmenopausal women of known menopausal age not receiving hormone replacement therapy. METHODs Subjects
Eighty-six apparently healthy postmenopausal Caucasian women were studied. They were recruited as part of a study into the effectiveness of hormone replacement therapy regimens [121. The women were all within 20% of their ideal body weight (Metropolitan Life Tables) and were not taking any medications known to affect carbohydrate metabolism. All women had undergone either a natural menopause or hysterectomy with the conservation of one ovary. The time of the menopause in the latter subjects was dated to the onset of symptoms typical of oestrogen deficiency. Procedures
The women attended our metabolic day ward on two occasions within a period of 2 weeks. At the
Key words: ageing, C-peptide, glucose, insulin, menopause. Correspondence: Dr I. C. Stevenson, Wynn Institute for Metabolic Research, 21 Wellington Road, St Johns Wood, London NW8 PSQ, U.K.
490
A. J. Proudler et al.
first visit routine biochemical, gonadotrophin and thyrotropin tests were performed on a fasting venous blood sample. This allowed confirmation of postmenopausal status and exclusion of occult hepatic, renal and thyroid disease. All women had a fasting plasma glucose level below 7.8 mmol/l. Before the second visit, in order to minimize the influence of dietary differences, the women were asked to consume a carbohydrate-rich diet ( > 200 g/ day) for 3 days, to fast overnight and to avoid smoking on the morning of the second visit. In each case, height, weight, age and time since menopause were recorded. Histories of oral contraception and cigarette consumption (for calculation of total years of oral contraceptive usage and calculation of total packs smoked) were taken. Questionnaires assessing current weekly alcohol intake and current physical activity [131 were completed. Full informed consent was obtained from each subject, and the study was approved by the local Ethical Committee. Intravenous glucose tolerance test
After the insertion of an indwelling plastic cannula into an antecubital vein, two fasting blood samples were taken 10min apart. Fifteen minutes before commencement of sampling, a heating pad was applied to the lower arm and hand in order to arterialize the venous blood in the sample arm. The pad remained in situ until the end of the test. A solution of 50% (w/v) D-glucose at a dosage of 0.5g of glucose/kg body weight was injected intravenously over 3min through a cannula inserted into an antecubital vein in the other arm, after which blood samples were taken at 3, 5, 7, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150 and 180min. Samples were collected into lithium-heparin tubes on ice and were centrifuged within 30 min. The plasma was divided into aliquots and the aliquots were either assayed for glucose or stored at -20°C for subsequent measurement of insulin and C-peptide. Plasma glucose concentration was measured using a glucose oxidase procedure with 4-aminophenazone as substrate. Plasma insulin and C-peptide immunoreactivities were assayed using double-antibody methods with materials supplied by Guildhay Ltd, Guildford, Surrey, U.K. Quality control was monitored by the use of freeze-dried plasmas and participation in national quality control schemes. Withinand between-batch coefficients of variation for each assay were 4.0% and 6.7% (insulin), 0.9% and 1.8% (glucose) and 6.0% and 8.1% (C-peptide). The ranges covered by quality control plasmas were 1 0 s 700 pmol/l (insulin), 4.7-10.6 mmol/l (glucose) and 0.32-1.69 nmol/l (C-peptide). Statistical methods
Data from the two fasting samples were combined to give a mean fasting value. The area under the curve was calculated using the trapezoidal rule for
Table 1. Age, time since menopause and lifestyle variables in 85 aowrentlv healthv postmenopausalwomen Mean
Range
Age (Yead Time since menopause (months) Body mass index (kg/m2) Cigarette packs smoked (packs) Physical activity (MJ/day) Alcohol intake (unitslweek) Duration of oral contraceptive usage (years)
n
52.4 43-61 31.1 !i-60 23.5 19.0-27.9 1130 0-21 900 10.2 7.15-16.8 5.2 0-28 2. I 0-20
the three analytes. Incremental area was calculated by subtracting the mean fasting value multiplied by 180 from the total area. A glucose decay constant was calculated using data from the glucose profile between 20 and 60min and the method of slope analysis [141. Kendall rank correlation analysis was used to examine the relationships of the lifestyle variables of current alcohol intake, cigarette consumption and duration of oral contraceptive use with chronological age and menopausal age and with the dependent variables of glucose, insulin and C-peptide listed below. Physical activity data were normalized by logarithmic transformation before bivariate regression analysis with menopausal and chronological age, and with glucose, insulin and C-peptide variables. Multivariate regression analysis used predictor variables of chronological age, menopausal age and body mass index. The dependent variables were fasting plasma level, total area or incremental area of glucose, insulin or C-peptide. Logarithmic transformation of data was performed to normalize the distribution of dependent variables with the exception of the incremental C-peptide area, which was subjected to square root transformation. RESULTS
Age, time since menopause and lifestyle variables are shown in Table 1. Intravenous glucose tolerance test data are shown in Table 2. Neither chronological nor menopausal age was significantly associated with the lifestyle variables of current physical activity, current alcohol intake, duration of oral contraception usage and cigarette packs smoked. Duration of oral contraceptive usage was not significantly associated with plasma fasting level or any area variable of glucose, insulin or C-peptide. Cigarette packs smoked was negatively associated with incremental glucose area ( P 0.005). Weekly alcohol consumption was associated with total and incremental C-peptide areas ( P < 0.01). Physical activity was positively associated with total insulin area and mean fasting plasma glucose level ( P 0.05
Chronological age Menopausal age BMI
>0.05
Incremental glucose area
Chronological age Menopausal age
Glucose decay constant
Chronological age Menopausal age BMI
>0.05 >0.05 >0.05 >0.05 >0.05
Fasting plasma glucose level
Glucose area
BMI
>0.05 0.016
>0.05 0.046
>0.05
Regression coefficient f SEM
r2
-0.00088f0.0028 -0.00099f0.00052 0.0098 fO.Oo40 0.0038f0.0044 0.00053f0.00083 0.013 f 0.0063 0.013 fO.012 O.oOol4 f0.0023 0.018 f0.017 -0.0031 fO.012 -O.ooOO23& 0.0022 -0.0093 f0.017
0.131
0.095
0.043
0.007
Table 4. Multivariate regression analyses using fasting plasma Gpeptide levels and C-peptide areas as single dependent variables and chronological age, menopausal age and body mass index as independent variables. Data are from 86 apparently healthy postmenopausal women given an intravenous glucose tolerance test. Abbreviation: BMI, body mass index. Dependent variable
Independent variable
P (independent
Regression coefficient f SEM
r2
0.118
>0.05
0.0081 k0.014 0.0043f0.0026 0.037 f 0.020 0.0IZk 0.016 0.0017+0.0030 0.040 0.023 0.059& 0.080 0.0085f0.015
>0.05
0.21 *O*I I
variable) Fasting plasma C-peptide level
C-peptide area
Incremental C-peptide area
Chronological age Menopausal age 1MI
>0.05 >0.05
>0.05
Chronological age Menopausal age BMI
>0.05
Chronological age Menopausal age BMI
>0.05
areas, were positively correlated (P0.05 >0.05
0.074
0.075
jects. This was also found for fasting plasma C-peptide level and C-peptide areas. Body mass index was associated with fasting plasma glucose level and total glucose area. In contrast, fasting plasma insulin level and insulin areas were positively associated with menopausal age and body
A. J. Proudler e t al.
492
Table 5. Multivariate regression analyses using fasting plasma insulin level and insulin areas as single dependent variables and chronological age, menopausal age and body mass index as independent variables. Data are from 86 apparently healthy postmenopausal women given an intravenous glucose tolerance test. Abbreviation: BMI, body mass index. ~
~~~
Dependent variable
Independent variable
P (independent variable)
Fasting plasma insulin level
Chronological age Menopausal age BMI
Insulin area
Chronological age Menopausal age
Incremental insulin area
Chronological age Menopausal age BMI
BMI
mass index, but not chronological age. This relationship of insulin area with menopausal age was preserved ( P =0.048) when physical activity was added as a dependent variable to the multivariate analysis.
DISCUSSION
Our results show a significant association between plasma insulin levels, expressed either as fasting plasma level or post-glucose tolerance insulin areas, and menopausal age. Despite the high degree of association between chronological and menopausal age in the subjects, an association of similar magnitude between plasma insulin level and chronological age was not found. Neither plasma glucose level nor plasma C-peptide level was significantly associated with menopausal age. This is in spite of the observed association between plasma insulin and C-peptide indices. These differences were not explained by physical activity, cigarette or alcohol consumption, or past duration of oral contraceptive usage. Previous studies of carbohydrate metabolism have demonstrated that chronological age is associated with a decrease in glucose tolerance, together with a tendency of plasma insulin concentration to increase [15-171. Such studies of ageing and carbohydrate metabolism have tended to ignore menopausal status and to utilize age groups that would encompass both pre- and post-menopausal women. Thus it is dificult to use such results to investigate the effects of the menopause on carbohydrate metabolism. A longitudinal study of healthy women that had undergone a natural menopause reported that fasting plasma insulin and glucose concentrations were seen to rise when tested 12-15 months after the menopause [S]. It was concluded that these were effects of ageing rather than of the menopause, as similar changes were seen in chronological agematched premenopausal controls. However, over the age range studied, it is possible that the control group consisted mostly of perimenopausal women. Consequently the changes observed in the fasting
>0.05 0.038
0.048 >0.05 0.0I0 0.05 0.024
489
Ageing and the response of plasma insulin, glucose and Gpeptide concentrations to intravenous glucose in postmenopausal women Anthony J. PROUDLER, Carl V. FELTON and John C. STEVENSON Wynn Institute for Metabolic Research, London, U.K. (Received 18 December 199112 April 1992; accepted I I June 1992)
1. Eighty-six apparently healthy postmenopausal women not receiving hormone replacement therapy were given an intravenous glucose tolerance test. Plasma glucose, insulin and C-peptide concentrations were determined in fasting and post-glucose challenge samples. 2. Using a multivariate regression model, with predictor variables of chronological age, menopausal age and body mass index, neither chronological age nor menopausal age correlated with fasting or postchallenge plasma glucose or C-peptide concentrations. In contrast, menopausal age was positively associated with fasting plasma insulin concentration (P= 0.038, model r2=0.107), insulin area (P=0.01, model Vt =0.236) and incremental insulin area (P=0.024, model r2=0.243). This relationship could not be explained by differences in lifestyle variables of alcohol consumption, physical activity, previous duration of oral contraceptive usage, history of cigarette smoking or body mass index. 3. Our findings suggest that loss of ovarian function is associated with hyperinsulinaemia, possibly via alterations in the clearance of circulating insulin from the plasma. The hyperinsulinaemia observed may contribute to the increased risk of cardiovascular disease seen in postmenopausal women. INTRODUCTION
Loss of ovarian function appears to increase the risk of cardiovascular disease. Postmenopausal women have a higher incidence of cardiovascular disease than premenopausal women of the same age [1,2]. Furthermore, women with a bilateral ovariectomy who are not receiving oestrogen replacement therapy have an increased risk of coronary heart disease compared with premenopausal women [3, 41.However, the potential factors that may account for these effects have not been fully investigated. Menopause-asociated changes have been reported in risk factors for cardiovascular disease including total cholesterol, cholesterol lipoprotein fractions, triacylglycerol, apolipoprotein B, haemoglobin, fibrinogen, Factors VII, and VIII, antithrombin I11
antigen, a,-antitrypsin antigen and plasminogen activity [S-113. Few studies have investigated changes in carbohydrate metabolism in relation to the menopause. Natural and surgical menopause were not associated with a significant change in fasting plasma glucose concentration in the Framingham Heart study [S]. A similar result was obtained in a crosssectional study of pre- and post-menopausal women [7]. A longitudinal study of naturally menopausal women and age-matched premenopausal controls reported increases in fasting plasma insulin and glucose concentrations on follow up after 2.5 years. Since these increases were seen in both postmenopausal women and their controls, the results were interpreted as effects of ageing [S]. In order to examine further any interactions between chronological or menopausal age and carbohydrate metabolism, we have determined plasma glucose, insulin and C-peptide profiles after intravenous glucose tolerance testing in postmenopausal women of known menopausal age not receiving hormone replacement therapy. METHODs Subjects
Eighty-six apparently healthy postmenopausal Caucasian women were studied. They were recruited as part of a study into the effectiveness of hormone replacement therapy regimens [121. The women were all within 20% of their ideal body weight (Metropolitan Life Tables) and were not taking any medications known to affect carbohydrate metabolism. All women had undergone either a natural menopause or hysterectomy with the conservation of one ovary. The time of the menopause in the latter subjects was dated to the onset of symptoms typical of oestrogen deficiency. Procedures
The women attended our metabolic day ward on two occasions within a period of 2 weeks. At the
Key words: ageing, C-peptide, glucose, insulin, menopause. Correspondence: Dr I. C. Stevenson, Wynn Institute for Metabolic Research, 21 Wellington Road, St Johns Wood, London NW8 PSQ, U.K.
490
A. J. Proudler et al.
first visit routine biochemical, gonadotrophin and thyrotropin tests were performed on a fasting venous blood sample. This allowed confirmation of postmenopausal status and exclusion of occult hepatic, renal and thyroid disease. All women had a fasting plasma glucose level below 7.8 mmol/l. Before the second visit, in order to minimize the influence of dietary differences, the women were asked to consume a carbohydrate-rich diet ( > 200 g/ day) for 3 days, to fast overnight and to avoid smoking on the morning of the second visit. In each case, height, weight, age and time since menopause were recorded. Histories of oral contraception and cigarette consumption (for calculation of total years of oral contraceptive usage and calculation of total packs smoked) were taken. Questionnaires assessing current weekly alcohol intake and current physical activity [131 were completed. Full informed consent was obtained from each subject, and the study was approved by the local Ethical Committee. Intravenous glucose tolerance test
After the insertion of an indwelling plastic cannula into an antecubital vein, two fasting blood samples were taken 10min apart. Fifteen minutes before commencement of sampling, a heating pad was applied to the lower arm and hand in order to arterialize the venous blood in the sample arm. The pad remained in situ until the end of the test. A solution of 50% (w/v) D-glucose at a dosage of 0.5g of glucose/kg body weight was injected intravenously over 3min through a cannula inserted into an antecubital vein in the other arm, after which blood samples were taken at 3, 5, 7, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150 and 180min. Samples were collected into lithium-heparin tubes on ice and were centrifuged within 30 min. The plasma was divided into aliquots and the aliquots were either assayed for glucose or stored at -20°C for subsequent measurement of insulin and C-peptide. Plasma glucose concentration was measured using a glucose oxidase procedure with 4-aminophenazone as substrate. Plasma insulin and C-peptide immunoreactivities were assayed using double-antibody methods with materials supplied by Guildhay Ltd, Guildford, Surrey, U.K. Quality control was monitored by the use of freeze-dried plasmas and participation in national quality control schemes. Withinand between-batch coefficients of variation for each assay were 4.0% and 6.7% (insulin), 0.9% and 1.8% (glucose) and 6.0% and 8.1% (C-peptide). The ranges covered by quality control plasmas were 1 0 s 700 pmol/l (insulin), 4.7-10.6 mmol/l (glucose) and 0.32-1.69 nmol/l (C-peptide). Statistical methods
Data from the two fasting samples were combined to give a mean fasting value. The area under the curve was calculated using the trapezoidal rule for
Table 1. Age, time since menopause and lifestyle variables in 85 aowrentlv healthv postmenopausalwomen Mean
Range
Age (Yead Time since menopause (months) Body mass index (kg/m2) Cigarette packs smoked (packs) Physical activity (MJ/day) Alcohol intake (unitslweek) Duration of oral contraceptive usage (years)
n
52.4 43-61 31.1 !i-60 23.5 19.0-27.9 1130 0-21 900 10.2 7.15-16.8 5.2 0-28 2. I 0-20
the three analytes. Incremental area was calculated by subtracting the mean fasting value multiplied by 180 from the total area. A glucose decay constant was calculated using data from the glucose profile between 20 and 60min and the method of slope analysis [141. Kendall rank correlation analysis was used to examine the relationships of the lifestyle variables of current alcohol intake, cigarette consumption and duration of oral contraceptive use with chronological age and menopausal age and with the dependent variables of glucose, insulin and C-peptide listed below. Physical activity data were normalized by logarithmic transformation before bivariate regression analysis with menopausal and chronological age, and with glucose, insulin and C-peptide variables. Multivariate regression analysis used predictor variables of chronological age, menopausal age and body mass index. The dependent variables were fasting plasma level, total area or incremental area of glucose, insulin or C-peptide. Logarithmic transformation of data was performed to normalize the distribution of dependent variables with the exception of the incremental C-peptide area, which was subjected to square root transformation. RESULTS
Age, time since menopause and lifestyle variables are shown in Table 1. Intravenous glucose tolerance test data are shown in Table 2. Neither chronological nor menopausal age was significantly associated with the lifestyle variables of current physical activity, current alcohol intake, duration of oral contraception usage and cigarette packs smoked. Duration of oral contraceptive usage was not significantly associated with plasma fasting level or any area variable of glucose, insulin or C-peptide. Cigarette packs smoked was negatively associated with incremental glucose area ( P 0.005). Weekly alcohol consumption was associated with total and incremental C-peptide areas ( P < 0.01). Physical activity was positively associated with total insulin area and mean fasting plasma glucose level ( P 0.05
Chronological age Menopausal age BMI
>0.05
Incremental glucose area
Chronological age Menopausal age
Glucose decay constant
Chronological age Menopausal age BMI
>0.05 >0.05 >0.05 >0.05 >0.05
Fasting plasma glucose level
Glucose area
BMI
>0.05 0.016
>0.05 0.046
>0.05
Regression coefficient f SEM
r2
-0.00088f0.0028 -0.00099f0.00052 0.0098 fO.Oo40 0.0038f0.0044 0.00053f0.00083 0.013 f 0.0063 0.013 fO.012 O.oOol4 f0.0023 0.018 f0.017 -0.0031 fO.012 -O.ooOO23& 0.0022 -0.0093 f0.017
0.131
0.095
0.043
0.007
Table 4. Multivariate regression analyses using fasting plasma Gpeptide levels and C-peptide areas as single dependent variables and chronological age, menopausal age and body mass index as independent variables. Data are from 86 apparently healthy postmenopausal women given an intravenous glucose tolerance test. Abbreviation: BMI, body mass index. Dependent variable
Independent variable
P (independent
Regression coefficient f SEM
r2
0.118
>0.05
0.0081 k0.014 0.0043f0.0026 0.037 f 0.020 0.0IZk 0.016 0.0017+0.0030 0.040 0.023 0.059& 0.080 0.0085f0.015
>0.05
0.21 *O*I I
variable) Fasting plasma C-peptide level
C-peptide area
Incremental C-peptide area
Chronological age Menopausal age 1MI
>0.05 >0.05
>0.05
Chronological age Menopausal age BMI
>0.05
Chronological age Menopausal age BMI
>0.05
areas, were positively correlated (P0.05 >0.05
0.074
0.075
jects. This was also found for fasting plasma C-peptide level and C-peptide areas. Body mass index was associated with fasting plasma glucose level and total glucose area. In contrast, fasting plasma insulin level and insulin areas were positively associated with menopausal age and body
A. J. Proudler e t al.
492
Table 5. Multivariate regression analyses using fasting plasma insulin level and insulin areas as single dependent variables and chronological age, menopausal age and body mass index as independent variables. Data are from 86 apparently healthy postmenopausal women given an intravenous glucose tolerance test. Abbreviation: BMI, body mass index. ~
~~~
Dependent variable
Independent variable
P (independent variable)
Fasting plasma insulin level
Chronological age Menopausal age BMI
Insulin area
Chronological age Menopausal age
Incremental insulin area
Chronological age Menopausal age BMI
BMI
mass index, but not chronological age. This relationship of insulin area with menopausal age was preserved ( P =0.048) when physical activity was added as a dependent variable to the multivariate analysis.
DISCUSSION
Our results show a significant association between plasma insulin levels, expressed either as fasting plasma level or post-glucose tolerance insulin areas, and menopausal age. Despite the high degree of association between chronological and menopausal age in the subjects, an association of similar magnitude between plasma insulin level and chronological age was not found. Neither plasma glucose level nor plasma C-peptide level was significantly associated with menopausal age. This is in spite of the observed association between plasma insulin and C-peptide indices. These differences were not explained by physical activity, cigarette or alcohol consumption, or past duration of oral contraceptive usage. Previous studies of carbohydrate metabolism have demonstrated that chronological age is associated with a decrease in glucose tolerance, together with a tendency of plasma insulin concentration to increase [15-171. Such studies of ageing and carbohydrate metabolism have tended to ignore menopausal status and to utilize age groups that would encompass both pre- and post-menopausal women. Thus it is dificult to use such results to investigate the effects of the menopause on carbohydrate metabolism. A longitudinal study of healthy women that had undergone a natural menopause reported that fasting plasma insulin and glucose concentrations were seen to rise when tested 12-15 months after the menopause [S]. It was concluded that these were effects of ageing rather than of the menopause, as similar changes were seen in chronological agematched premenopausal controls. However, over the age range studied, it is possible that the control group consisted mostly of perimenopausal women. Consequently the changes observed in the fasting
>0.05 0.038
0.048 >0.05 0.0I0 0.05 0.024
