Differential Effects of Nifedipine and Co-Amilozide on the Progression of Early Carotid Wall Changes
Background—Common carotid artery intima-media thickness (IMT) progression was compared between 4 years of treatment with nifedipine and diuretic.
Methods and Results—This study, ancillary to the International Nifedipine GITS Study: Intervention as a Goal in Hypertension Treatment (INSIGHT), involved nifedipine 30 mg or co-amilozide (hydrochlorothiazide 25 mg and amiloride 2.5 mg) with optional subsequent titration. Among 439 randomized hypertensive patients, 324 had ≥1 year of follow-up (intent-to-treat group), and 242 completed follow-up (until-end-of-study group). Ultrasonography was performed at baseline, 4 months later, and then every year. Central computerized reading provided far-wall IMT, diameter, and cross-sectional area IMT (CSA-IMT). The primary outcome was IMT progression rate (slope of IMT-time regression). Secondary outcomes were changes from baseline (Δ) in IMT, diameter, and CSA-IMT. In the until-end-of-study population, between-treatment differences existed in IMT progression rate (P=0.002), Δ IMT (P=0.001), and Δ CSA-IMT (P=0.006), because IMT progressed on co-amilozide but not on nifedipine. In the intent-to-treat population, treatment differences existed in Δ IMT (P=0.004) and Δ CSA-IMT (P=0.04) but not in IMT progression rate (P=0.09). Patients with ≥2, 3, or 4 years of follow-up showed treatment differences in IMT progression rate (P=0.04, 0.004, 0.007, respectively), Δ IMT (P=0.005, 0.001, 0.005), and Δ CSA-IMT (P=0.025, 0.013, 0.015). Diameter decreased more on co-amilozide than on nifedipine in the intent-to-treat population (P<0.05), whereas blood pressure decreased similarly on both treatments.
Conclusions—A difference in early carotid wall changes is shown between 2 equally effective antihypertensive treatments.
Preintrusive thickening of artery walls is detectable in hypertension1 by ultrasonographic measurement of intima-media thickness (IMT) in the carotid artery.2 3 4 5 6 It is relevant to assess the influence of antihypertensive treatment on IMT progression,7 which is faster in the presence of higher blood pressure than lower pressure.8 Previous trials have compared calcium antagonist and diuretic effects on IMT and shown differences favoring the former.9 10 This study,11 ancillary to the International Nifedipine GITS Study: Intervention as a Goal in Hypertension Treatment (INSIGHT),12 compares the effects of nifedipine (long-acting formulation) and co-amilozide on carotid IMT progression in hypertensive patients.
Participants were recruited by French INSIGHT centers with the same criteria as INSIGHT,12 plus 2 specific criteria: participation in INSIGHT and availability of a good IMT image in the right common carotid artery before randomization. Eligible subjects were men and women 55 to 80 years old with blood pressure of 150/95 mm Hg or systolic values of 160 mm Hg regardless of diastolic and ≥1 of the following risk factors: smoking, hypercholesterolemia, diabetes mellitus, left ventricular hypertrophy, history of cardiovascular disease, family history of myocardial infarction, and proteinuria. Patients with secondary or malignant hypertension, stroke, or myocardial infarction within the previous 12 months were not eligible. Of 469 subjects who met enrollment criteria, 439 fulfilled the blood pressure criteria to be randomized. Randomization was done centrally, including all centers participating, and respecting the randomization criteria of INSIGHT,12 taking into account age, sex, risk factors, and inclusion in the present study. Among randomized subjects, 324 had ≥1 year of follow-up and 3 serial IMT measurements, thus allowing IMT progression rate to be estimated (intent-to-treat population, Figure 1⇓). Of these, 242 subjects completed the planned follow-up and had ≥5 valid IMT measurements (until-end-of-study population, Figure 1⇓). Reasons why not all of the 439 randomized patients completed 1-year follow-up (intent-to-treat) or the planned follow-up (until-end-of-study) are given in Figure 2⇓.
After 1 month of treatment with placebo (baseline), patients were randomized in a double-blind fashion to receive nifedipine gastrointestinal-transport-system (GITS) (30 mg) or co-amilozide (amiloride 2.5 mg and hydrochlorothiazide 25 mg) (monotherapy dose 1). Medication titration could subsequently be adapted to obtain blood pressure of 140/90 mm Hg. Three titration levels were performed: (1) monotherapy dose 2, ie, twice monotherapy dose 1; (2) bitherapy, adding atenolol 25 or 50 mg (or enalapril 5 or 10 mg if atenolol was contraindicated); and (3) tritherapy, adding to the bitherapy any other antihypertensive drug exclusive of calcium antagonists and diuretics. Duration of follow-up of the until-end-of-study population was on average 48 months, range 37 to 53 months (Figure 1⇑).
Investigation was performed in 13 centers with high-resolution imagers,11 with 5- to 10-MHz probes, by sonographers who were vascular physicians trained in IMT measurement (Appendix). Baseline ultrasound detection of plaque was performed in both carotid arteries. The presence of plaque, defined as a focal echogenic structure encroaching into the lumen with a thickness of 50% of that of the surrounding walls, was positive if ≥1 plaques were found.13 Longitudinal scans of the right common carotid artery, 3 cm before the bifurcation (flow divider), were performed at baseline, 4 months after randomization, and every 1 year for the duration of follow-up. This segment, close and parallel to the skin surface, is easy to image, with a lumen-intima interface regular and parallel to the adventitia. The same sonographer performed all scans of each patient throughout the study. Minimum gain was adjusted to visualize the lumen-intimal and media-adventitial interfaces defining IMT in the far wall along ≥1 cm of length. Then, the vessel image was frozen in telediastole by ECG R-triggering and transferred to a computer (Apple Macintosh) with a program of image acquisition (Iôtec, IôDP). This program allowed the image to be digitized and stored.2 6 The computerized generation of the anatomic situation (mask) of the investigated area at baseline allowed superimposition of the real-time echographic image under treatment and the mask of the first investigation to improve the repeated identification of the carotid sector.2 6
Images stored on disk were sent to the reading center for off-line analysis by 1 reader blinded to treatment assignment. Far-wall IMT was measured with an automated computerized edge-detection program (Iô, IôDP) on the basis of gray-level density and tissue-recognition algorithms,2 3 6 locating echoes arising from IMT interfaces and calculating their distance every 100 μm along ≥1 cm length. The program also located the leading edges of far- and near-wall blood-intima interfaces defining lumen diameter (D) and calculated their distance along the same length as IMT.6 The cross-sectional area of IMT (CSA-IMT) was calculated as follows14 : π×IMT×(IMT+D).
Quality Control Program
Centers were visited before the study and annually until its termination, with examination of a tissue-equivalent phantom allowing the initial value of the settings to be maintained throughout the study. Because the same phantom was examined, all units and transducers were matched between centers. Images were read after termination of the study, but the reader was not blind to the sequence time of the tracings. Images were read on a field of measure that had the same situation, with regard to the artery and surrounding structure, at baseline and during treatment and was reader-independent thanks to the automated program. Reproducibility of measurement was assessed during the placebo period in 39 random subjects; the mean absolute difference between repeat measures of IMT at 15-day intervals by the same sonographer was 0.026 mm, in agreement with the precision obtained on the basis of previous studies.5 6
Primary outcome was the IMT progression rate estimated in each patient as the slope of the least-squares regression line relating IMT to the time since baseline ultrasonography throughout the follow-up. A minimum of 3 serial IMT assessments (ie, 1 year of follow-up) was required for obtaining a valid estimation of slope. In the until-end-of-study population, the slope was estimated from IMT measurements (≥5) during the entire follow-up period. In the intent-to-treat population, the slope was estimated independently of whether the patient completed the planned follow-up. To analyze how the slope may change as a function of follow-up duration, intention-to-treat analysis was performed in patients who were ≥2 years, ≥3 years, or ≥4 years on treatment. Secondary outcomes consisted of absolute change from baseline (last value under treatment minus baseline value) in IMT, CSA-IMT, and diameter. They were determined in the until-end-of-study population, in the intent-to-treat population independently of whether the patient completed the planned follow-up, and in patients with ≥2, 3, or 4 years on treatment.
Sample size was chosen to provide 90% power (with 2P=0.05) to detect a 0.0375-mm difference in change from baseline in IMT after 3 years between treatment groups, with an expected SD of the change of 0.105 mm.11 Accordingly, 165 patients were required per group, ie, a total of 330, which was increased by ≈30% (430 patients) to take the withdrawals into account. It was decided before unblinding that the primary analysis would be done in the until-end-of-study population because the precision of the primary outcome was greater than in the intent-to-treat population. All patients of the until-end-of-study population had ≥5 valid IMT assessments, whereas some patients in the intent-to-treat population had only 3 IMT assessments. Between–treatment group differences in parameters were tested with Student’s t test for independent groups. Within–treatment group differences in parameters from zero were tested with a paired t test. Multivariate analysis was used to test between–treatment group differences after adjustment for (1) ultrasound center; (2) baseline covariates including age, sex, body mass index, associated risk factors, ultrasound plaque, and diastolic pressure; and (3) absolute change in diastolic pressure (last value under treatment minus baseline value). Statistical analysis was performed with an SAS software package (V12) for Windows operating system.
The intent-to-treat, until-end-of-study, and randomized populations did not differ at baseline. Treatment groups did not differ in baseline parameters in either study population (Table 1⇓). The percentages of participants on monotherapy (dose 1 or 2) or bitherapy (atenolol or enalapril) did not differ between treatment groups (Table 2⇓). Decreases in pressure were similar in the until-end-of-study and intent-to-treat populations at different durations of follow-up (Figure 3⇓).
IMT Progression Rate
The IMT progression rate differed between treatment groups before (P=0.002) and after (P=0.003) adjustment for center, baseline covariates, and diastolic pressure change. Within-group comparison to zero showed that IMT progressed significantly on co-amilozide (P<0.001) but not on nifedipine (Table 3⇓).
In the whole intent-to treat population, the difference in IMT progression rate between treatments did not reach statistical significance (P=0.09) (Table 3⇑), but in the subgroup of individuals who received only the randomization medications (monotherapy dose 1 or 2), the difference between nifedipine and co-amilozide was significant (−0.0015±0.0033 versus 0.0078±0.0026 mm/y, P<0.05). In patients with ≥2, ≥3, or ≥4 years of follow-up, the difference in IMT progression rate was significant between treatments before adjustment (P=0.04, 0.004, 0.007, respectively) and after adjustment (P=0.04, 0.003, 0.010) (Table 3⇑). In the subgroup of patients receiving the randomization medications (monotherapy), the difference between nifedipine and co-amilozide was also significant in subjects with ≥2 years of follow-up (−0.0016±0.0026 versus 0.0065±0.0022 mm/y, P<0.05), ≥3 years of follow-up (−0.0027±0.0026 versus 0.0065±0.0023 mm/y, P<0.01), or ≥4 years of follow-up (−0.0031±0.0028 versus 0.0063±0.0022 mm/y, P<0.01). Within-group comparison to zero showed that IMT progressed significantly on co-amilozide in the whole intent-to-treat population (P<0.05) and in patients with ≥2, 3, or 4 years of follow-up (P<0.01, P<0.001, P<0.001) but did not on nifedipine (Table 3⇑).
Changes From Baseline in IMT and CSA-IMT
Changes in IMT and CSA-IMT differed between treatments before adjustment (P=0.001, P=0.006) and after adjustment (P=0.002, P=0.005) (Table 4⇓). Within-group comparison to zero showed that IMT and CSA-IMT increased on co-amilozide (P<0.001) but not on nifedipine (Table 4⇓).
In the whole intent-to-treat population, changes in IMT and CSA-IMT differed between treatments before adjustment (P=0.004, P=0.037) and after adjustment (P=0.008, P=0.036) (Table 4⇑). Changes in IMT and CSA-IMT differed between treatments in patients with ≥2 years (P=0.005, P=0.025), 3 years (P=0.001, P=0.013), or 4 years (P=0.005, P=0.015) of follow-up, and such differences persisted after adjustment (Table 4⇑). Within-group comparisons showed that IMT increased from baseline on co-amilozide in the whole intent-to-treat population and in patients with ≥2, 3, or 4 years of follow-up (P<0.001) but did not on nifedipine (Table 4⇑); CSA-IMT increased from baseline on co-amilozide only in patients with ≥3 or 4 years of follow-up (P<0.05) but did not on nifedipine (Table 4⇑).
Change in Diameter From Baseline
Diameter decreased on nifedipine (P<0.05) and co-amilozide (P<0.001), and diameter changes did not differ between the 2 treatments (Figure 4⇓).
Diameter decreased on co-amilozide in the whole intent-to-treat population (P<0.001) and in patients with different duration of follow-up (P<0.001) but did not change on nifedipine except in patients with ≥4 years of follow-up (P<0.05) (Figure 4⇑). Diameter changes were different between drugs in the whole intent-to-treat population (P<0.05) and in patients with ≥3 years of follow-up (P<0.05) but not in patients with ≥1 or ≥4 years of follow-up (Figure 4⇑).
In the until-end-of-study population, IMT progression rate and changes from baseline in IMT and CSA-IMT (an estimate of arterial wall mass) were different between treatments. In the intent-to-treat population, treatment difference in IMT rate did not reach statistical significance, whereas changes in IMT and CSA-IMT were significantly different between treatments; but in patients with ≥2, 3, or 4 years of follow-up, significant treatment differences existed in IMT progression rate as well as in changes in IMT and CSA-IMT. Thus, the nonsignificant treatment difference in IMT progression rate of the intent-to-treat population was due to patients who were only 1 year on treatment. This follow-up is not sufficiently long to determine IMT progression rate with good precision, especially because measurement at 4 months was taken too early. Different treatment effects on IMT result because IMT progressed on co-amilozide, whereas it did not on nifedipine. The possibility that atenolol and hydrochlorothiazide both have negative effects on IMT progression (ie, by increasing insulin resistance) has been excluded by the analysis of those individuals who received only the randomization medications (monotherapy) and who exhibited the same different drug effects on IMT progression rate as the whole population.
Several points of these findings have to be considered. First, previous trials (especially MIDAS and VHAS9 10 ) have shown beneficial effects of calcium antagonists on carotid IMT compared with diuretic agents. The MIDAS study concluded that there was no difference in IMT effects of isradipine and hydrochlorothiazide, because there was no divergence in the IMT progression slope between treatments,9 but when data were analyzed as IMT change from baseline, differences existed favoring isradipine.9 The VHAS study found a greater effect of verapamil than chlorthalidone on IMT progression when the slope of IMT change was corrected by the initial IMT value.10 The VHAS analysis, however, took into account thicker lesions in the bifurcation and internal carotid artery; this may explain why we found progression with diuretics and no change with nifedipine but VHAS found regression with both drugs and greater regression with verapamil.10 Another point is the physiological meaning of the treatment effects on IMT change. This latter may be a result of medial hypertrophy or intimal thickening reflecting early atherosclerosis.15 We cannot distinguish between these mechanisms, because ultrasonography does not measure medial and intimal thickness separately. Thus, our interpretation is dependent on indirect arguments. Those for medial hypertrophy are that IMT has been measured in distal common carotid artery free of atherosclerotic plaque and that medial hypertrophy is specifically related to hypertension.1 Conversely, the implication of atherogenesis is supported (1) by the fact that IMT, even measured locally, may reflect a widespread atherosclerosis-related phenomenon (in particular in elderly high-risk patients)16 and (2) by previously demonstrated antiatherogenic properties of calcium antagonists.17 Arguments in favor of an atherogenic nature of IMT change would have been strengthened if investigation had been extended to plaques. Although the presence of plaque was analyzed at baseline as a part of risk stratification, the thickness of plaque and its change throughout the study were not assessed because our computerized measurement was inappropriate. Thus, our work does not permit us to assess whether the effects of treatments on IMT may reflect different effects on vascular hypertrophy rather than on atherosclerosis. A third point to discuss is the meaning of different effects on carotid IMT when the effects of 2 treatments on mortality-morbidity are equal in INSIGHT,12 as well as in our patients, with limitations due to the smallness of the sample. An explanation may be that early wall change, such as IMT, cannot predict relatively short-term occurrence of complications. In consideration of the well-recognized predictive value of IMT,16 the beneficial effects of nifedipine on IMT may announce a smaller incidence of complications in the longer term. The last findings are diameter changes that may influence wall thickness. They occur at the beginning of treatment (when pressure is reduced) and remain relatively stable without progression as the IMT changes, suggesting that they did not confound IMT progression over time. Furthermore, the decrease in diameter is more pronounced on co-amilozide than on nifedipine, probably because the latter vasodilates the carotid artery18 and counteracts the passive diameter decrease due to pressure lowering. Such a difference, however, disappears in patients with the longest follow-up.
In summary, we report clear-cut differences in early carotid wall changes between 2 equally effective antihypertensive strategies. The general applicability of the findings may be limited by the facts that our population was small and that the sample (until-end-of-study population) from which our main conclusions were drawn was smaller than the one estimated to give a 90% power to detect. Therefore, our findings would need to be confirmed by larger studies.
Sonographers: Angers, M. Davinroy; Blois, H. Ducoux; Bordeaux, V. Dehant, J.C. Saby; Cavaillon, P. Giraud, J.Y. Brunet; La Rochelle, J.P. Chantereau, G. Desprairie; Nantes, G. Fauvel; Nimes, M. Dauzat; Paris Nord, F. Luizy; Paris Sud, J. Gariépy; Roanne, P. Drevon; Thionville, S. Kownator; Toulouse, B. Chamontin, J. Amar; Toulon, N. Dagorn.
Reader: J. Gariépy.
Technical assistance: M. Massonneau.
Bayer project leaders: J.C. Provost, A. Deverly.
Statistician: D. Moyse.
Management Committee: A. Simon, J. Levenson, J. Gariepy, J.L. Megnien, D. Moyse, A. Deverly, M. Massonneau.
This study was supported by Bayer Pharma. We thank the INSIGHT investigators for recruiting participants, M. Crichi for secretarial assistance, and J. Menard for advice in the writing of the manuscript.
- Received January 18, 2001.
- Revision received March 28, 2001.
- Accepted March 30, 2001.
- Copyright © 2001 by American Heart Association
Gariepy J, Massonneau M, Levenson J, et al. Evidence for in vivo carotid and femoral wall thickening in human hypertension. Groupe de Prévention Cardio-vasculaire en Médecine du Travail. Hypertension. 1993;22:111–118.
Armentano RL, Graf S, Barra JG, et al. Carotid wall viscosity increase is related to intima-media thickening in hypertensive patients. Hypertension. 1998;31:534–539.
Lakka TA, Salonen R, Kaplan GA, et al. Blood pressure and the progression of carotid atherosclerosis in middle-aged men. Hypertension. 1999;34:51–56.
Simon A, Gariepy J, Provost JC, et al. Nifedipine, hypertension, and arterial wall thickness in extracoronary arteries: design features of a clinical trial using non invasive ultrasound end points. J Cardiovasc Pharmacol. 1996;28:S27–S32.
Brown MJ, Palmer CR, Castaigne A, et al. Morbidity and mortality in patients randomized to double-blind treatment with a long-acting calcium-channel blocker or diuretic in the International Nifedipine GITS study: Intervention as a Goal in Hypertension Treatment (INSIGHT). Lancet. 2000;356:366–372.
Simon A, Giral P, Levenson J. Extracoronary atherosclerotic plaque at multiple sites and total coronary calcification deposit in asymptomatic men: association with coronary risk profile. Circulation. 1995;92:1414–1421.
Linhart A, Gariepy J, Giral P, et al. Carotid artery and left ventricular structural relationship in asymptomatic men at risk for cardiovascular disease. Atherosclerosis. 1996;15:127:103–112.
Levenson JA, Safar ME, Simon AC, et al. Systemic and arterial hemodynamic effects of nifedipine (20 mg) in mild-to-moderate hypertension. Hypertension. 1983;5(suppl V):V-57–V-60.