|Risk factors and carotid atherosclerosis in hypertensive and control subjects|
|Prev||Chapter 6. Discussion||Next|
Attempts have been made to standardize the methodological variability in the IMT measurements between different laboratories during the recent years (Berglund et al. 1994). Although near-wall measurements have been theoretically considered unreliable, near-wall measurements continue to be commonly used even in recent studies (ARIC, Vitelli 1997, The Rotterdam study (Bots et al. 1999), the CHS study (Polak et al. 1996).). The argument for the choice of both the near and the far wall is the clinical findings of progression trials, including the finding that near wall measurements reduce the variability of IMT progression by 30–40%. Furthermore, the lower variability means that a smaller sample size is sufficient (Furberg et al. 1994). In the present study, the ultrasonographic examinations and the IMT measurements were performed by one trained radiologist at five distinct points in both the right and the left carotid arteries, including the near and far walls, i.e. altogether 20 sites. The mean values of several measurements were used, which gives a more reliable IMT value than the maximal IMT value alone (Wikstrand & Wendelhag 1994). Furthermore, IMT measurements with and without plaque thickness were used, due to the fact that if a plaque is located on the site of measurement, it may be difficult to measure the IMT value separately. In the present research, the variabilitity of the IMT measurements was in good agreement with that of the other studies (Riley et al. 1992), and the variabilities were also similar for near-wall and far-wall values. In general, high precision and good accuracy make ultrasonography a feasible method for assessing the IMT of arterial walls as a valid estimate of the degree of early atherosclerosis in population studies (Berglund et al. 1994).
In the present study, IMT was independently associated with the established risk factors for atherosclerosis: age, LDL cholesterol, systolic blood pressure and smoking. These results are consistent with some earlier studies (Craven et al. 1990, Howard et al. 1990, Salonen et al. 1991). We did not find any significant associations between BMI, WH ratio, physical activity and IMT, which is contradictory to the ARIC study (Folsom et al. 1994), but agrees with Salonen et al. (Salonen et al. 1991). The independent predictors of IMT explained only 18% of the variance in IMT, which is in accordance with some earlier reports (Salonen et al. 1991). The possible reasons for only this small amount of variance being explained are the possible weaker associations with carotid atherosclerosis than coronary atherosclerosis, the measurement error of IMT, genetic susceptibility, haemostatic factors, imprecise measurement of risk factors and unknown confounding biases (Folsom et al. 1994).
The mean IMT values were approximately 0.15 mm higher in our study compared to a previous study (Howard et al. 1993). Compared to other Finnish studies, the IMT values were 0.17–0.22 mm higher (Salonen & Salonen 1991) or 0.15 mm lower in corresponding samples (Rauramaa et al. 1994). Using US, Salonen et al. (Salonen et al. 1988) detected atherosclerotic lesions in 48.8% of the 42– to 60-year-old men in their ischaemic heart disease risk factor study, but 37% had intima-media thickening, while only 10% had plaques and 1.8% had carotid stenosis. Plaques were more frequent in our series. Since the prevalence of atherosclerosis is higher or at least equal in eastern Finland, the difference is unexpected and may be due to the different methods for detecting plaques and IMT. To a great extent, the discrepancy between the different studies may have been due to the different methods of detecting the IMT and the different cardiovascular risk profiles of the samples.
Carotid atherosclerosis measured as CPIMT and carotid plaques was significantly more prevalent in the hypertensive cohorts than in the controls in our study. This is in concordance with the results of several studies (Roman et al. 1992, Crouse et al. 1987, Tell et al. 1989, Bots et al. 1993), but contrary results have also been observed (Roman et al. 1992, Handa et al. 1990, Poli et al. 1993, Haapanen et al. 1989), and the effect of hypertension on IMT has been consistently less powerful than the effect of the other CV risk factors, i.e., age, smoking and cholesterol (Giral et al. 1991, Handa et al. 1990).
However, the present finding was limited only to the hypertensive men. This gender difference agrees with the previous observations (Tell et al. 1989, Fabris et al. 1994, Handa et al. 1990, Prisant et al. 1993). In Japanese subjects (Handa et al. 1990), age and male gender were independent risk factors for a plaque score, but hypertension did not reach statistical significance. In this particular study, the patient selection was not randomized, the prevalence of cardiovascular risk factors was very high, e.g., in controls the prevalences of hypertension, DM and smoking were 63%, 30% and 41%, respectively.
In the present study, office systolic BP was the most powerful predictor of maximal CPIMT among the different measures of blood pressure. This is in concordance with the recent studies (Ebrahim et al. 1999, Sun et al. 2000, Saba et al. 1999). However, ambulatory BP levels have been shown to predict target organ damage more consistently than office BP levels (Mancia et al. 1996, Muiesan et al. 1996). This finding was recently supported in the study by Pauletto et al. (1999), who showed an independent association between IMT and ambulatory BP measures, but not with office BP. The role of hypertension in the development of LDL cholesterol mediated atherosclerosis measured by CCAIMT was confirmed by the study of Sun et al. (2000). They observed that elevated LDL cholesterol was associated with increased IMT in the upper tertile of SBP but not in the lower tertiles after adjustment for the other risk factors. Total cholesterol and DBP yielded a comparable pattern of associations, but the magnitude of the interactions was reduced. This supports the response-to-injury model of hypertension-induced atherosclerosis. Another explanation for the IMT thickening along increasing LDL levels occurring in the middle of the SBP tertile was suggested to be adaptive thickening of the intima and the media (Chopanian 1990). Such thickening is characterized by remodeling to counteract the rise in wall tension observed as medial hypertrophy in the presence of hypertension. In contrast, maladaptive thickening involving monocyte recruitment and lipid accumulation in the intima occurs in the high BP tertile group, in which endothelial damage is more likely to be sufficient to initiate atherogenesis (Sun et al. 2000). These findings were supported in the ACAPS study (Furberg et al. 1994) where the effect of the lipid-lowering lovastatin intervention was larger in hypertensive patients than in the nonhypertensive group. However, alternative explanations for the interaction between BP and LDL related to carotid IMT have also been proposed. Elevated BP may increase the diffusion of LDL into the subendothelial space (Fry 1987) or prolong the retention of LDL in the intima (Williams & Tabas 1995). Furthermore, the interaction between BP and LDL may be due to other factors, as is indicated by the presence of both risk factors.
The relationship between IMT and hypertension was not consistently significant at the different sites of carotid arteries. Indeed, the IMT differences were significant only in the ICA and not in the CCA or BIF. This is in agreement with the results of Tell et al. (1989), who found that the risk factors were not uniformly associated with the plaque thickness at different sites of the carotid arteries. Furthermore, Giral et al. (1991) found different risk profiles according to site of atherosis in the carotid, aortic and femoral arteries. Hypertension was significantly associated with plaques in the aortic and femoral arteries but not in the carotid artery in hypercholesterolemic men without CVD. Moreover, LDL cholesterol predicted independently arterial plaques only in the carotid artery, but not at other locations. Finally, age, smoking, SBP and fasting glucose were significantly associated with the number of arterial sites affected by plaque.
Data reported in the literature indicate that different risk factors, such as lipid levels and BP, may have different roles during the pathogenesis of structural changes of the carotid artery wall. More advanced atherosclerotic lesions, such as mean maximal IMT, have been shown to be more influenced by plasma cholesterol and triglycerides than by BP in multivariate analysis (Pauletto et al. 1999). This hypothesis is supported by Giral et al. (1991) and by the finding that the prevalences of plaques or marked intimal lesions were much lower in Italian borderline hypertensives than in Swedish borderline hypertensives with higher cholesterol levels (Lemne et al. 1995). Furthermore, parallel results were observed in recent analyses of the British Regional Heart Study. Ebrahim et al. (1999) found that IMTCCA, IMTBIF and plaque correlate with each other but show different patterns of association with risk factors and prevalent disease. IMTCCA was strongly associated with the risk factors for stroke, such as age and SBP, whereas IMTBIF and plaque were more directly associated with ischemic heart disease risk factors, such as smoking, and prevalent ischemic heart disease. Their analyses suggest that the presence of plaque, rather than the thickness of IMTBIF, appears to be the major criterion of high-risk disease. The relative impact of the various risk factors on different structural changes of the carotid artery requires further evaluation. Indeed, the inclusion of plaque thickness in the IMT measurement may be confounding because of the possibly two qualitatively different pathological processes: smooth muscle hypertrophy and plaque formation.
In the present, study we found inconsistent associations between different insulin measures and IMT. In simple correlation analysis, the area under the curve of the insulinemic response in OGTT had a significant association with IMT in control women, but the association did not remain significant in regression analyses. These findings are in concordance with the results of Haffner et al. (1998), who found that the correlations between fasting and 2-hour insulin and IMT varied within 0.06–0.08 (p = 0.1 and p = 0.02, respectively), indicating a similar magnitude in the relationship. However, the correlation was significant only in non-Hispanic whites and Hispanics, but not in African Americans. In that study, proinsulin and split proinsulin had a significant association with IMT in non-Hispanic whites and Hispanics, but the effect did not remain after adjustment for CV risk factors. Proinsulin correlated most strongly correlated with the PAI-1 (r = 0.41, p < 0.001) levels, and further analyses suggested a possible mediating role for PAI-1 between proinsulin and IMT.
In the ARIC study (Folsom et al. 1994), 14 430 persons, both blacks (27.3% of the reported sample) and whites, in four US communities were evaluated for intimal-medial carotid wall thickness and cardiovascular risk factors. The investigators found a positive univariate association between BMI, WH ratio, work physical activity and diabetes in all the four regression analyses with IMT, and fasting insulin in three of the four regression analyses with IMT. After adjustment of fasting serum insulin for the established cardiovascular risk factors, the effect of insulin was reduced to a non-significant level in women (p = 0.06) and to borderline significance in men (p = 0.04). An increase of 100 mmol/L in fasting insulin represented an increase of about 3% in average carotid wall thickness, which shows the effect of insulin to be rather modest. The main result of the present study remained essentially unchanged, even when the insulin values were adjusted for age, pack-years of smoking, BMI, hypertension and LDL and HDL cholesterol, as in the ARIC study. Contrary results were obtained in a Japanese study, which used OGTT and the sum of insulin values to determine insulin resistance (Fujii et al. 1997). The sum of insulin values and the the sum of insulin/the sum of glucose ratio had independent association with the maximal CCA IMT in regression analyses in normotensive, non-diabetic men. The risk factor profile was different compared with our study, e.g., the mean HDL cholesterol level was 2.6 mmol/L (SD 0.7 mmol/L and the prevalence of smoking was 50.4%. Also, the prevalence of IGT and DM assessed by OGTT in a recent study (Ohmura et al. 1994) was surprisingly high, being 25.6% and 12.4%, respectively.
Results closely comparable to the present findings were obtained in a Swedish sample, in which insulin resistance was determined by homeostasis model assessment (Hedblad et al. 2000).
Three studies have assessed the relationship between IMT and insulin resistance or sensitivity by direct methods, such as the euglycemic glucose clamp technique and frequently sampled intravenous glucose tolerance test (FSIGT). In the IRAS study, insulin sensitivity was determined by FSIGT and OGTT and a significant association was observed between insulin sensitivity and ICA IMT in non-Hispanic whites and Hispanics, but not in black subjects. The relationship attenuated after adjustment for the traditional risk factors and only remained significant in non-Hispanic whites (Howard et al. 1996). However, the fasting and 2-hour insulin levels did not reach any notable relation with IMT, which is in agreement with our study. Agewall et al. (1995) found a significant association between CCA IMT and insulin sensitivity determined by the clamp technique in elderly Swedish men at high and low coronary risk. Laakso et al. (1991) investigated the relationship between atherosclerosis determined by the presence of plaque in the carotid and/or femoral artery and insulin resistance assessed by clamp technique in middle-aged non-obese men. Subjects with atherosclerosis had 20% reduced whole-body glucose uptake compared to subjects without atherosclerosis. Patients with atherosclerosis in either the femoral or carotid arteries were equally insulin resistant as patients having atherosclerosis in both of these arterial beds. Their results indicate that the defect in insulin action in atherosclerosis lies in the peripheral tissues, most likely in muscle. Interestingly, glucose oxidation, lipid oxidation, suppression of FFA levels and promotion of potassium uptake did not differ between patients and controls, resembling the profile of hypertension. However, the study subjects were normotensive without hyperinsulinemia. Thus, the results suggest that insulin resistance in atherosclerosis may be similar to that in hypertension.
All in all, the previous studies assessing insulin resistance by serum insulin concentrations yield comparable results, indicating that the association between insulin and atherosclerotic vascular disease may be weak and thus far only significant in middle-aged Caucasian men and women (Folsom et al. 1997, Fontbonne 1994), not in African Americans (Howard et al. 1996). Indeed, the results of a recent meta-analysis by Ruige et al. (1998) support this interpretation between insulin levels and CVD. An increase of 7.95 mU/L (50 pmol/L) in fasting insulin resulted in a summary relative risk (RR)(95% CI) of 1.18 (1.08 to 1.29). Because the effect of insulin is attenuated after adjustment for traditional risk factors, at least part of the effect is mediated by the interaction between insulin and the traditional risk factors.
The effect of insulin on atherosclerosis is also less powerful than that of the established cardiovascular risk factors, such as age, LDL cholesterol and smoking. However, an insulin-resistant state in an individual may develop by at least two different mechanisms, namely insulin resistance can be a primary phenomenon without any elevation in serum insulin levels or a secondary event usually leading to hyperinsulinemia. According to Laakso et al. (1991), directly measured insulin resistance is associated with atherosclerosis. Insulin resistance per se has been associated with a variety of CV risk factors, including hypertension and lipid abnormalities, which are accepted to enhance atherosclerosis.
Some observed differences between hypertensives and controls in the present study may be partly explained by certain types of antihypertensive medication affecting the relation between insulin and IMT. Hypertensive subjects under medical control may be more prone to receive other therapies including cholesterol lowering medication.