OUP user menu

Antioxidant enzymes, inflammatory indices and lifestyle factors in older men: a cohort analysis

C.W. Mulholland, P.C. Elwood, A. Davis, D.I. Thurnham, O. Kennedy, J. Coulter, A. Fehily, J.J. Strain
DOI: http://dx.doi.org/10.1093/qjmed/92.10.579 579-585 First published online: 1 October 1999


We examined the relationship between blood antioxidant enzyme activities, indices of inflammatory status and a number of lifestyle factors in the Caerphilly prospective cohort study of ischaemic heart disease. The study began in 1979 and is based on a representative male population sample. Initially 2512 men were seen in phase I, and followed-up every 5 years in phases II and III; they have recently been seen in phase IV. Data on social class, smoking habit, alcohol consumption were obtained by questionnaire, and body mass index was measured. Antioxidant enzyme activities and indices of inflammatory status were estimated by standard techniques. Significant associations were observed for: age with α-1-antichymotrypsin (p<0.0001) and with caeruloplasmin, both protein and oxidase (p<0.0001); smoking habit with α-1-antichymotrypsin (p<0.0001), with caeruloplasmin, both protein and oxidase (p<0.0001) and with glutathione peroxidose (GPX) (p<0.0001); social class with α-1-antichymotrypsin (p<0.0001), with caeruloplasmin both protein (p<0.001) and oxidase (p<0.01) and with GPX (p<0.0001); body mass index with α-1-antichymotrypsin (p<0.0001) and with caeruloplasmin protein (p<0.001). There was no significant association between alcohol consumption and any of the blood enzymes measured. Factor analysis produced a three-factor model (explaining 65.9% of the variation in the data set) which appeared to indicate close inter-relationships among antioxidants.


Ischaemic heart disease continues to be a major cause of death in Europe and the USA.1 In recent years there has been much interest in the pathological role of free radicals in the development of atherosclerosis.2–,5 The body has evolved a complex defence strategy to minimize the damaging effects of various oxidants. Central to this defence are the antioxidant enzymes of the blood. They include superoxide dismutase (SOD, EC, glutathione peroxidase (GPX, EC, catalase (EC, and caeruloplasmin, which act in concert to protect the organism from oxidative damage.6 Erythrocyte superoxide dismutase is copper-dependent (though zinc has a structural role) and accelerates the dismutation of superoxide O2 to H2O2 and O2.7 The selenium-containing enzyme glutathione peroxidase detoxifies H2O2 by utilizing reduced glutathione (GSH) and H2O2 as substrates to yield H2O and oxidized glutathione (GSSG);8 the GSSG, in turn, is reduced to GSH by the enzyme glutathione reductase which requires NADPH as reductant; the latter is supplied via the enzyme glucose-6-phosphate dehydrogenase (G-6-PDH, EC Catalase is a large enzyme containing haem-bound iron at its active sites. It removes H2O2 by breaking it down directly to O2.9 In plasma, caeruloplasmin (oxidase) acts as an important antioxidant in the protection of the organism from oxidative stress. This latter enzyme, together with alpha-1-antichymotrypsin, is also an acute-phase reactant and can be used to indicate inflammatory status.

Accumulating evidence suggests that alterations or disturbances in these defensive enzyme systems may contribute to, or exacerbate, the development of atherosclerosis.10–,12 Other studies examining the effect of ageing on myocardial antioxidant enzyme activities and lipid peroxidation in rats have shown that myocardial antioxidant capacity is weakened during ageing, increasing the possibility of oxidative damage.13 In humans, it has been demonstrated that smoking, a major risk factor for atherosclerosis depresses plasma glutathione peroxidase activity.14 The authors of the study argue that the finding is consistent with the view that smokers are under a sustained oxidative load, and that an inadequate antioxidant status combined with the increased free radical load caused by smoking, may exacerbate the oxidation of LDL and increase its atherogenic properties. This may be particularly true in a population at high risk of coronary heart disease. Decreased glutathione peroxidase, superoxide dismutase and catalase activities have also been observed in platelets from individuals with coronary heart disease,15 and in individuals with acute myocardial infarction.16–,17

Most of the studies examining antioxidant enzyme activities and heart disease have been in highly selected subjects, using relatively small samples. The Caerphilly prospective cohort study of ischaemic heart disease, stroke and cognitive decline, gave an opportunity to study blood antioxidant enzyme activities in a representative male population sample and explore possible interrelations between antioxidant enzyme activities, age, social class, smoking habit and alcohol consumption.


A full description of the Caerphilly cohort and the collection of data relevant to heart disease, stroke and other conditions has been given elsewhere;18–,21 only a brief description is presented here. Initially 2512 men were seen in phase I (1979–1983), when they were aged 45–59 years. The men were subsequently followed-up after each 5-year period (in phases II and III) and have recently been seen in phase IV. When the men were aged 55–69 years (phase III), fasting blood samples were taken for the estimation of blood antioxidant vitamins and enzymes. Unfortunately, the examinations of the men had commenced some months before the opportunity for these estimations arose and blood was not obtained from a few hundred men for the antioxidant estimations. Plasma was separated from erythrocytes by low-speed centrifugation, and the erythrocytes were washed with physiological saline. Both plasma and washed erythrocytes were stored at −70 °C prior to analysis.

Antioxidant enzymes were estimated by standard methods as follows. Superoxide dismutase, glutathione peroxidase and glucose-6-phosphate dehydrogenase activities were estimated on a Cobas Fara auto-analyzer using commercially available kits supplied by Randox Laboratories. Erythrocyte catalase activity was estimated by the method of Aebi22 using a Varian Cary 1 spectrometer. Erythrocyte haemoglobin (Hb) was measured by a modification of the method described by Dacie and Lewis23 using a dedicated haemoglobinometer (Coulter Electronics). The assay was calibrated using cyanmethaemoglobin (HiCN) standard, conforming to B.S. 3985 (BDH) as recommended by the manufacturer. The acute-phase proteins alpha-1-antichymotrypsin and caeruloplasmin were estimated using immunoturbidometric methods and commercial kits supplied by Dako, with the exception that the aliquots used for these analyses were diluted manually, instead of automatically prior to assay on a Cobas Fara auto-analyzer. Caeruloplasmin oxidase (CpO) activity was estimated on the Cobas Fara using p-phenylenediamine dihydrochloride as a substrate.24 Intra- and inter-batch variations were <6.5% for all analytes measured.

Additional data, including smoking habit, social class, alcohol consumption and body mass index were collected by questionnaire and physical measurement, respectively; full details have been described previously.19

Statistical analysis

The distributions of the various parameters measured were examined for normality using the Kolmogorov Smirnov test. Results are presented as means, medians and standard deviations (SDs), together with the mean and 95% ranges obtained after transformation of data if appropriate. Inter-relationships among the various data sets were examined by correlation, regression in a correlation matrix, and by factor analysis, and for these, the variates were transformed, usually to logarithms, to achieve normality or near normality, where appropriate. Factor analyses were performed in an attempt to extract a relatively small number of factors or underlying constructs which would simplify the interrelations of the variables and so facilitate understanding of the complex processes.


There were 2154 men in the cohort at the time of phase III examinations (1989–1993). Fasting blood samples were obtained from 1800 of these, representing 84% of the total cohort.

Means, medians and SDs for the antioxidant enzymes studied in the Caerphilly cohort of men are shown in Table 1. Only SOD and GPX were found to be normally distributed. Results are presented for all enzymes as raw data and also following log transformation. Inter-relationships (shown as correlation coefficients), among the various antioxidant enzymes in the cohort are shown in Table 2. Significant correlations were observed between a number of the measures, particularly those which are linked through inflammatory status, i.e. alpha-1-antichymotrypsin and caeruloplasmin (protein and oxidase). Associations of blood enzyme levels were also compared with a number of variates, specifically: age, smoking habit, social class, body mass index and alcohol consumption. The results are shown in Table 3. Data were both examined as raw data and standardized to allow for possible effects of confounders. There were no major differences between the two. Therefore only standardized data are presented. Significant associations were observed for: age with alpha-1-antichymotrypsin and with caeruloplasmin, both protein and oxidase; smoking habit with alpha-1-antichymotrypsin, with caeruloplasmin (protein and oxidase) and with GPX; social class with alpha-1-antichymotrypsin, with caeruloplasmin (protein and oxidase) and with GPX; body mass index with alpha-1-antichymotrypsin and with caeruloplasmin protein. No significant association was observed between alcohol consumption and any of the blood enzymes measured.

View this table:
Table 1 

Distributions of blood levels of enzymes in approximately 1780 men aged 55–69 years

VariateMean, median and standard deviation of the raw data
Mean (95% range) after log transformation
α-1-Antichymotrypsin (g/l)Mean 0.310; median 0.308; SD 0.074
0.302 (0.180–0.469)
Caeruloplasmin: total (g/l)Mean 0.229; median 0.230; SD 0.047
0.224 (0.130–0.320)
Caeruloplasmin: oxidase (U/l)Mean 627; median 615; SD 130
614 (402–915)
Superoxide dismutase (U/g Hb)Mean 995; median 958; SD 145
943 (684–1225)
Glutathione peroxidase (U/g Hb)Mean 38.0; median 37.8; SD 11.0
36.2 (16.7–60.0)
Glucose-6-phosphate dehydrogenase (U/g Hb)Mean 1.32; median 1.16; SD 0.700
1.15 (0.409–3.000)
Catalase (K/g Hb)Mean 52.9; median 53.9; SD 17.1
50.1 (23.0–86.2)
View this table:
Table 2 

Inter-relationships, shown as correlation coefficients among the various blood enzyme concentrations or activities of 1689 men

α-1-Antichymotrypsin (g/l)Caerulo-plasmin Total (g/l)Caerulo-plasmin Oxidase (U/l)SOD (U/g Hb)GPX (U/g Hb)G-6-PDH (U/g Hb)Catalase (K/gHb)
Data were all log-transformed except for superoxide dismutase (SOD) and glutathione peroxidase (GPX) which were normally distributed. Two-tailed tests of significance: **p<0.01; ***p<0.001. G-6-PDH, glucose-6-phosphate dehydrogenase.
α-1-Antichymotrypsin (g/l)1
Caeruloplasmin: total (g/l)0.722***1
Caeruloplasmin: oxidase (U/l)0.453***0.5831
SOD (U/g Hb)0.093***0.103***0.0371
GPX (U/g Hb)−0.030−0.003−0.0570.0591
G-6-PDH (U/g Hb)0.0340.0190.0580.115***0.116***1
Catalase (K/g Hb)0.116***0.117***0.078**0.284***0.087***0.193***1
View this table:
Table 3a 

Associations of blood α-1-antichymotrypsin and caeruloplasmin with a range of variates

Variaten*α-1-Antichymotrypsin (g/l)Caeruloplasmin: total (g/l)Caeruloplasmin: oxidase (U/l)
Data shown are means and 95% ranges after log transformation in original units. aMaximum number of men in each group. Each variate has been standardized for the effects of the other factors, except for age, which is shown unstandardized.
Age (years)
0.20–0.52 0.14–0.33 437–942
0.20–0.45 0.13–0.32 426–892
0.156–0.45 0.12–0.31 381–920
p <0.0001<0.0001<0.0001
Smoking habit
0.16–0.39 0.12–0.29 352–780
0.17–0.41 0.12–0.29 384–818
0.17–0.47 0.13–0.32 445–927
p <0.0001<0.0001<0.0001
Social class
0.15–0.41 0.12–0.29 378–820
0.18–0.44 0.13–0.31 400–832
p <0.0001<0.001<0.01
Body mass index
Highest 1/36150.270.20570
0.15–0.41 0.12–0.29 378–809
Middle 1/36080.280.21572
0.16–0.42 0.12–0.29 402–811
Lowest 1/35610.290.22578
0.20–0.43 0.14–0.30 382–869
p <0.01<0.0001NS
View this table:
Table 3b 

Associations of blood superoxide dismutase, glutathione peroxidase, glucose-6-phosphate dehydrogenase and catalase with variates from Table 3a

VariateSuperoxide dismutase (U/g Hb)Glutathione peroxidase (U/g Hb)G-6-PDH (U/g Hb)Catalase (k/g Hb)
Data shown are means and 95% ranges after log transformation in original units (except for superoxide dismutase (SOD) and glutathione peroxidase (GPX) which are normally distributed and for which standard deviations are also shown). G-6-PDH, glucose-6-phosphate dehydrogenase. Each variate has been standardized for the effects of the other factors, except for age, which is shown unstandardized.
>65955 SD 15237 SD 11.51.1250.0
658–1225 15.8–59.8 0.40–2.88 23.3–1.7
60–65956 SD 13638.4 SD 10.91.1450.6
706–1225 18.4–59.4 0.42–3.1 22.6–85.9
≤60954 SD 14837.9 SD 10.81.1849.8
682–1225 16.6–60.5 0.41–3.0 22.8–92.0
Smoking habit
Never951 SD 148340.8 SD 11.31.1048.7
675–1204 17.6–64.1 0.39–2.93 22.9–85.1
Ex-smoker933 SD 14140.5 SD 11.11.1448.2
667–1189 18.8–63.0 0.42–2.88 23.3–81.1
Smoker947 SD 15137.4 SD 10.41.1348.9
669–1220 17.7–58.3 0.35–3.05 20.7–88.9
p NS<0.0001NSNS
Social class
Non-manual133 SD 13340.5 SD 11.31.1448.2
686–1190 17.4–61.9 0.42–2.93 22.8–85.5
Manual946 SD 15138.3 SD 10.81.1949.7
658–1214 18.2–60.9 0.43–3.13 23.0–82.6
p NS<0.001NSNS
Body mass index
Highest 1/3930 SD 14340.1 SD 11.41.1948.9
647–1188 19.1–63.8 0.43–3.01 24.2–82.4
Middle 1/3933 SD 14240.5 SD 10.81.1448.2
664–1180 19.2–60.7 0.43–2.92 22.6–84.7
Lowest 1/3933 SD 15139.5 SD 10.61.1448.8
660–1236 17.9–60.8 0.38–3.10 20.3–83.0

When factor analysis was applied to the seven plasma variables (n=1689), two common factors were derived with eigen values of 2.24 and 1.40, which together explained 52.1% of the total variation in the data set. Factor 1 explained 32.0% of the total variation in the data set. It was most closely associated with alpha-1-antichymotrypsin, caeruloplasmin protein and oxidase (and to a less important degree with SOD and catalase). Factor 2 explained a further 20.1% of the total variation. It was most closely associated with SOD, catalase, G-6-PDH and GPX.

The measures of adequacy of the model were adequate, with the notable exception of low communalities of three of the variables (0.42 for SOD, 0.34 for G-6-PDH, 0.18 for GPX) which made the model unacceptable.

A three-factor model was also constructed. Although the third factor explains marginally less of the total variation in the data set than an individual variable would, the communalities of the variables (and other measures of the adequacy of the model) improve, indicating that this model provides a more fitting description of the interrelationships among the observed variables. Factor 3 explains a further 13.8% of the total variation in the data set, giving a total for the three-factor model of 65.9%. Factor 1 is still most closely associated with alpha-1-antichymotrypsin, caeruloplasmin protein and oxidase. Factor 2 is most closely associated with SOD and catalase (and to a much lesser extent G-6-PDH). Factor 3 is most closely associated with GPX and G-6-PDH.

The most striking improvement in the adequacy of the three-factor model as opposed to the two-factor model is the improvement in the communality of GPX (from 0.18 to 0.78). The communalities of SOD and G-6-PDH also improved (to 0.63 and 0.42). It would appear that GPX has a particular role which is not contained within the three-factor model.


The Caerphilly cohort of men has been extensively studied since the project's inception in 1979.25–,29 One of the strengths of the present data is that they have been derived from a representative population sample. Caerphilly, in south Wales, is a typical industrial/residential area, although the social class distribution suggests that there may be some bias towards manual occupations. The aim of the present study was to obtain descriptive statistics and examine possible inter-relationships among blood antioxidant enzyme activities, using factor analysis as a data reduction and hypothesis- creating technique. In addition, possible relationships between antioxidant enzymes and a number of lifestyle variates were investigated. Although there are currently no definitive reference ranges for the antioxidant enzymes SOD, GPX and catalase, the results were initially compared with previously published data before possible relationships between the antioxidant enzymes and lifestyle factors were explored.

SOD activity is quantified indirectly in a sample by the ability of that sample to reduce the detected free radical flux. SOD activities vary between assay systems, and therefore the kit manufacturers do not recommend reference ranges. However, the data obtained on the Caerphilly sample is comparable with previously published results using the same analytical kit.30 Activities of GPX may also vary within and between assay systems. It is thought that oxidized GPX is susceptible to deactivation by cyanide, which is contained in the Drabkins reagent used to stabilize haemoglobin in the samples and prevent other enzyme reactions from occurring.31 This problem was overcome by the addition of dithiothreitol/dithionite in the dilution buffer. Results from the Caerphilly cohort (38.0±11.0SD IU/g Hb) are similar to a recently published study (36.1±6.5 IU/g Hb).32 The latter study, however, was based on a very small number (n=10) of myocardial infarction patients. Mean catalase activities in the Caerphilly cohort were comparable with a previous, though much smaller, study30 from our laboratory of a group of 50–54-year-old males (52.9±17.1 cf. 49.0±20 K/g Hb respectively, n=18), who in turn had higher catalase activities than a much earlier study (of 31.3±9.6 K/g Hb).33 The manufacturer of the G-6-PDH kit used quotes a `normal' value of 1.31±0.13 (mean±SD) U/gHb which is comparable to the results obtained in this study of 1.32±0.7 U/g Hb.

Acute-phase proteins are markers of disease. Those measured in this study are known as positive acute-phase proteins (i.e. they increase in disease states). Both mean alpha-1-antichymotrypsin and caeruloplasmin (protein) levels fell within previously established reference ranges34 of 0.3–0.6 and 0.15–0.6 g/l, respectively. They were also noted to be comparable to those in a cohort of the UK arm of the EPIC study,35 while caeruloplasmin oxidase activities were similar to those of a 1994 investigation.36 Table 3 shows that, on the whole, blood antioxidant enzymes changed little with age. As expected, acute-phase proteins (a measure of inflammatory status) were positively associated with age. Both acute-phase proteins displayed significant positive associations with smoking habit. Only GPX demonstrated a significant association (negative) with smoking habit. Erythrocyte GPX activities have previously been shown to be lower in smokers compared with non-smokers,37 most probably reflecting the increased oxidative stress thought to occur in this group, owing to the free radicals present in smoke.38 The results for social class showed a trend towards increasing inflammatory status (as judged by increasing levels of acute-phase proteins), with decreasing social class. GPX was the only enzyme to show a significant association with social class, activities being decreased in the manual group. BMI exhibited a similar pattern to the other lifestyle factors examined, however, the associations were not as strong. The trends and associations observed for the lower social classes were not unexpected, with environmental circumstances and lifestyle factors such as smoking, diet, etc. probably contributing to higher levels of inflammation and increased BMIs.

Factor analysis of the data produced a three-factor model that explained 65.9% of the variation in the data set. A number of interesting aspects were revealed by factor analysis. Factor 1 of the model gathers together three measures which are linked through inflammatory mechanisms. Factor 2 includes two `main-stream' antioxidants, namely SOD and catalase. The observation that factor 3 is most closely associated with GPX and G-6-PDH is what would be predicted by theory, as the latter enzyme produces reduced cofactors for the reduction of glutathione that, in turn, may be utilized by the former enzyme. Evidence indicates that glucose-6-phosphate is a central molecule in cellular glucose metabolism which critically influences pentose phosphate cycle activity and, via NADPH generation, regulates GPX activity for radical detoxification and also cholesterol and triglyceride synthesis.39

The Cu Zn SOD enzyme converts O2 to H2O2. Erythrocyte GPX can remove H2O2 at a high rate, and oxidizes GSH to GSSG. The GSSG can be reduced to GSH by NADPH produced from G-6-PD. Catalase has a greater capacity than GPX to destroy H2O2 (produced by, inter alia, the respiratory burst of neutrophils) by breaking it down directly to H2O; in terms of molecules of H2O2 destroyed per min per molecule of enzyme, it is one of the most active enzymes known. However, its affinity for H2O2 is also low, and it requires high H2O2 concentrations for high activity, dealing only slowly with H2O2 at low concentrations.

The H2O2 molecule has the ability to cross cell membranes readily and, if produced extracellularly, can diffuse into the nearest cell (including erythrocytes) for metabolism by antioxidant enzymes. Therefore, erythrocytes and other cell types, could act as `sinks' for H2O2. This inter-linking of antioxidant defence mechanisms combined with indices of inflammation (or oxidative stress) may be what is being portrayed in the three-factor model. This supports the concept of a balanced, co-operative, antioxidant defence system in blood, which depends on optimum levels of a number of antioxidant systems rather than on the concentration or activity of a single antioxidant, but does not preclude or diminish the importance of particular antioxidants, in particular situations or conditions.

In conclusion, the major findings from this study are that caeruloplasmin (total and oxidase) is increased in smokers only in the lower social classes. Changes in caeruloplasmin were reflected in changes in the acute-phase reactant alpha-1-antichymotrypsin, and were probably related to inflammation or disease status. GPX was the only antioxidant enzyme to change significantly with lifestyle, increasing with age, but was lower in smokers and those in lower socio-economic classes and with lower BMI. Taken together these results indicate a decreased antioxidant protection paralleling oxidative/inflammatory stress. Factor analysis resulted in a three-factor model explaining 65.9% of the variation in the data set, which appeared to indicate close inter-relationships among the antioxidant enzymes.


View Abstract