Among patients with diabetes mellitus, elevated blood. Hypertension

Hypertension Effects of Visit-to-Visit Variability in Systolic Blood Pressure on Macrovascular and Microvascular Complications in Patients With Type 2 Diabetes Mellitus The ADVANCE Trial Jun Hata, MD, PhD; Hisatomi Arima, MD, PhD; Peter M. Rothwell, MD, PhD; Mark Woodward, PhD; Sophia Zoungas, MD, PhD; Craig Anderson, MD, PhD; Anushka Patel, MD, PhD; Bruce Neal, MD, PhD; Paul Glasziou, MD, PhD; Pavel Hamet, MD, PhD; Giuseppe Mancia, MD, PhD; Neil Poulter, MD, PhD; Bryan Williams, MD, PhD; Stephen MacMahon, PhD; John Chalmers MD, PhD on behalf of the ADVANCE Collaborative Group Background Recent evidence suggests that visit-to-visit variability in systolic blood pressure () and maximum are predictors of cardiovascular disease. However, it remains uncertain whether these parameters predict the risks of macrovascular and microvascular complications in patients with type 2 diabetes mellitus. Methods and Results The Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) was a factorial randomized controlled trial of blood pressure lowering and blood glucose control in patients with type 2 diabetes mellitus. The present analysis included 8811 patients without major macrovascular and microvascular or death during the first 24 months after randomization. variability (defined as standard deviation) and maximum were determined during the first 24 months after randomization. During a median 2.4 years of follow-up from the 24-month visit, 407 major macrovascular (myocardial infarction, stroke, or cardiovascular death) and 476 microvascular (new or worsening nephropathy or retinopathy) were observed. The association of major macrovascular and microvascular with variability was continuous even after adjustment for mean and other confounding factors (both P<5 for trend). Hazard ratios (95% confidence intervals) for the highest tenth of variability were 4 (0.99 2.39) for macrovascular and 1.84 (1.19 2.84) for microvascular in comparison with the lowest tenth. For maximum, hazard ratios (95% confidence intervals) for the highest tenth were 3.64 (1.73 7.66) and 2.18 (4 4.58), respectively. Conclusion Visit-to-visit variability in and maximum were independent risk factors for macrovascular and microvascular complications in type 2 diabetes mellitus. Clinical Trial Registration URL: http://www.clinicaltrials.gov. Unique Identifier: NCT00145925. (Circulation. 2013;128:1325-1334.) Key Words: blood pressure diabetes mellitus, type 2 diabetic nephropathies myocardial infarction stroke Among patients with diabetes mellitus, elevated blood pressure (BP) is one of the major modifiable risk factors for macrovascular diseases such as coronary heart disease and stroke, and is also associated with progression of microvascular complications such as nephropathy and retinopathy. 1 As a result, the current guidelines for the management of BP in patients with diabetes mellitus recommend diagnostic and therapeutic approaches based on a usual BP value (defined as mean of BP over a period of time), and other cardiovascular risk factors and target organ damage, as well. 2 4 Clinical Perspective on p 1334 A recent study of several cohorts of patients with previous transient ischemic attack or stroke and of the Anglo- Scandinavian Cardiac Outcomes Trial Blood Pressure Lowering Arm (ASCOT-BPLA) reported that visit-to-visit Received March 19, 2013; accepted July 30, 2013. From The George Institute for Global Health, University of Sydney, Sydney, Australia (J.H., H.A., M.W., S.Z., C.A., A.P., B.N., S.M., J.C.); University of Oxford, Oxford, United Kingdom (P.M.R., S.M.); School of Public Health, Monash University, Clayton, Australia (S.Z.); Bond University, Gold Coast, Australia (P.G.); Université de Montréal, Montreal, Canada (P.H.); University of Milan-Bicocca, Milan, Italy (G.M.); Imperial College, London, United Kingdom (N.P.); and University College London (UCL) and the National Institute for Health Research UCL Hospitals Biomedical Research Centre, London, United Kingdom (B.W.). The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/circulationaha. 1102717/-/DC1. Correspondence to John Chalmers, MD, PhD, The George Institute for Global Health, University of Sydney, Level 10, King George Building, Royal Prince Alfred Hospital, Missenden Rd, New South Wales, 2050 Australia. E-mail chalmers@georgeinstitute.org.au 2013 American Heart Association, Inc. Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.1102717 1325

1326 Circulation September 17, 2013 variability in systolic blood pressure () and maximum were significantly associated with future risks of stroke and other cardiovascular independent of mean BP and other cardiovascular risk factors. 5 It was suggested that, although long-term mean BP is important in the long-term development of vascular disease, episodic peaks in BP (associated with increased variability and high maximum) might be important in the short-term triggering of vascular. 6 Increased variability of BP may be partially explained by arterial stiffness 7 and abnormal autonomic function, 8 both of which are common complications of diabetes mellitus and have been associated with cardiovascular. 9,10 However, there has been little evidence of the effects of visit-to-visit BP variability and maximum on the risks of macrovascular and microvascular complications in patients with type 2 diabetes mellitus. The aim of the present study was to assess the effects of visit-to-visit variability and maximum on the risks of macrovascular and microvascular outcomes by using the data from the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) trial. 11 13 Methods Study Design ADVANCE was a factorial randomized controlled trial of BP lowering and intensive blood glucose control in patients with type 2 diabetes mellitus. The detailed design has been described previously. 11 13 In brief, a total of 11140 patients with type 2 diabetes mellitus aged 55 years who had a history of major macrovascular or microvascular disease, or at least 1 other risk factor for vascular disease, were recruited from 215 collaborating centers in 20 countries from Asia, Australasia, Europe, and North America. These patients entered a 6-week prerandomization run-in period during which they received a fixed combination of angiotensin-converting enzyme inhibitor, perindopril (2 mg), and thiazide diuretic, indapamide (0.625 mg). After the run-in period, participants were randomly assigned, in a factorial design, to a fixed combination of perindopril and indapamide (2 mg/0.625 mg for the first 3 months and 4 mg/1.25 mg thereafter) or matching placebo, and to either an intensive glucose control strategy (target hemoglobin A1c of 6.5%) or a standard glucose control strategy based on local guidelines. The use of concomitant treatments, 11140 patients randomized BP measurement period Exclusion: 994 patients with major macroor microvascular or death in first 24 months 1335 patients with missing BP value at any of 6 occasions BP lowering intervention (perindopril + indapamide vs. placebo) BP measurement (6 occasions) 8811 patients analyzed Follow up period (median 2.4 years) including BP-lowering therapy, remained at the discretion of the responsible physician with 2 exceptions: the use of thiazide diuretics was not allowed, and open-label perindopril (maximum 4 mg/d) was the only angiotensin-converting enzyme inhibitor allowed. For the present analysis, 8811 patients who had not experienced major macrovascular or microvascular or death during the first 24 months and had BP measurements at all of 6 occasions (3, 4, 6, 12, 18, and 24 months after randomization) were studied (Figure 1). Approval for the trial was obtained from each center s institutional review board, and all participants provided written informed consent. BP Measurements and Visit-to-Visit Variability BP was measured in duplicate, with an interval of at least 1 minute, after 5 minutes of rest in the seated position, by using a standardized automated sphygmomanometer (Omron HEM-705CP, Tokyo, Japan). BP recordings were made at registration, at randomization, at 3, 4, and 6 months after randomization, and at every 6 months thereafter; the mean of the 2 measurements was used as the BP value at each visit. To minimize any misclassification of BP variability due to randomized BP-lowering treatment, BP values from 3 to 24 months were used for the estimation of the mean, visitto-visit variability of, and maximum (Figure 1). Mean was defined as an average of values measured on the 6 occasions. Visit-to-visit variability in was defined by using standard deviation (SD) and coefficient of variation (CV, defined as SD/mean). was defined as a highest value among the 6 occasions. Follow-Up and Study Outcomes Participants were followed up from the 24-month visit to the end of the study (Figure 1). Primary outcomes were major macrovascular and microvascular (both considered jointly and separately). Major macrovascular were defined as myocardial infarction (both nonfatal and fatal), stroke (both nonfatal and fatal), or cardiovascular death. Major microvascular were defined as new or worsening nephropathy (development of macroalbuminuria, doubling of the serum creatinine to a level of at least 200 μmol/l, need for renal replacement therapy, or death due to renal disease) or retinopathy (development of proliferative retinopathy, macular edema, diabetes mellitus related blindness, or retinal photocoagulation therapy). Secondary outcomes in the present study were all-cause mortality and components of primary outcomes (myocardial infarction, stroke, cardiovascular death, new or worsening nephropathy, and new or worsening retinopathy). For patients with multiple, the first applicable event was used in each analysis. An independent end point advisory committee adjudicated all of these outcomes. Figure 1. Flow diagram for study participants. Blood pressure (BP) measured at 6 occasions (3, 4, 6, 12, 18, and 24 months after randomization) was used to determine the mean, visit-tovisit variability, and maximum of systolic BP. After excluding patients who had experienced major macrovascular or microvascular or death within 24 months, and excluding patients with missing BP values at any of 6 occasions, 8811 patients were eligible for the present study. 0 34 6 12 18 24 36 48 Months after randomization

Hata et al Blood Pressure Variability in Type 2 Diabetes Mellitus 1327 Statistical Analyses Associations between baseline characteristics at registration and SD or maximum during the measurement period were evaluated by using univariate and multivariate linear regression analysis. The effects of parameters (ie, mean, SD, CV, and maximum ) on study outcomes were estimated by using a Cox proportional hazards model with adjustment for age, sex, randomized BP-lowering intervention, and randomized glucose control intervention (model 1), or with additional adjustment for region of residence, duration of diabetes mellitus, current smoking, current alcohol drinking, heart rate, total cholesterol, triglycerides (log-transformed), body mass index, use of β-blockers, and use of calcium-channel blockers (model 2), or with additional adjustment for mean during the measurement period (model 3). Each of SD and maximum was categorized into 10 groups according to the tenths within each BP-lowering treatment group (the placebo and the active). Mean, SD, CV, and maximum were also analyzed as continuous variables with hazard ratios (HRs) and 95% confidence intervals (CIs) shown per increase of 1 SD for each parameter. In addition, we performed 8 sets of sensitivity analyses: (1) Patients who discontinued the randomized BP-lowering treatment before the 24-month visit were excluded (n=8230). (2) Patients with a history of macrovascular or microvascular disease at registration were excluded (n=5519). (3) parameters were determined at 4 occasions from 6 to 24 months after randomization (n=9128). (4) parameters were determined at 5 occasions from 3 to 18 months (n=9276). (5) parameters were determined at 7 occasions from 3 to 30 months (n=8364). (6) A study cohort included patients who had measurements at 2 of 6 occasions from 3 to 24 months (n=9966). (7) A study cohort was the same as the original one (n=8811), but the groups of SD and maximum were determined on the basis of the tenths within all patients, pooling those on active treatment and placebo. (8) A study cohort was the same (n=8811), but SD and maximum were categorized into 5 groups according to the fifths. We also performed a subgroup analysis for patients with and without changes in BP-lowering medication regimen during the BP measurement period (n=4448 and 4363, respectively). All statistical analyses were performed with the use of SAS version 9.3 (SAS Institute). A 2-sided P<5 was considered to be statistically significant. Results The characteristics of the patients included and excluded for the present study are summarized in Table 1. Among the 8811 patients studied, the mean age was 66 years, 42% were female, and 37% were registered in Asia. Average BP at registration was 145/81 mm Hg and 69% had a history of hypertension. As shown in multivariate analysis in Tables I and II (in the onlineonly Data Supplement), age, currently treated hypertension, a history of microvascular disease, at registration, and hemoglobin A1c at registration were positively associated with SD and maximum during the measurement period. Female sex and a history of macrovascular disease were positively associated and baseline heart rate was inversely associated with SD. Both BP-lowering and blood glucose control interventions reduced SD and maximum. Among the parameters, maximum was strongly correlated with mean (Pearson correlation coefficient, r=0.91), and CV was strongly correlated with SD (r=0.96) (Table III in the online-only Data Supplement). Figure I in the online-only Data Supplement shows the proportion of patients on 2 or more additional (open-label) BP-lowering drugs at the 24-month visit according to the tenths of SD and maximum. The patients with higher levels of SD or maximum were more likely to receive multiple drugs. During median follow-up of 2.4 years from the 24-month visit after randomization, 407 major macrovascular and 476 microvascular were observed. The risks of primary outcomes (major macrovascular and microvascular ) and all-cause mortality according to participant groups defined by the tenths of SD and maximum are shown in Figure 2. The risks of major macrovascular and microvascular and all-cause mortality rose progressively with higher SD levels of, even after adjustments for mean and other cardiovascular risk factors (all P<5 for trend). Multivariable-adjusted HRs (95% CIs) in the highest SD group in comparison with the lowest group were 1.69 (1.24 2.31) for combined macro- and microvascular, 4 (0.99 2.39) for major macrovascular, 1.84 (1.19 2.84) for major microvascular, and 8 (1.30 3.31) for all-cause mortality. Higher levels of maximum were significantly associated with the risks of the combined primary outcomes, major macrovascular and all-cause mortality (all P for trend <5). A similar but statistically nonsignificant association was observed between maximum and major microvascular (P=8 for trend). Multivariable-adjusted HRs (95% CIs) in the highest maximum group were 2.68 (7 4.59) for the combined primary outcomes, 3.64 (1.73 7.66) for macrovascular, 2.18 (4 4.58) for microvascular, and 2.44 (1.14 5.23) for all-cause mortality. In addition, we performed 8 sets of sensitivity analyses for Figure 2 (Figures II to IX in the online-only Data Supplement) and confirmed that the main results were essentially similar to the original analysis, although some probability values did not reach a level of significance. During the 24-month measurement period, the BP-lowering treatment regimen was changed for 4448 patients (50%). However, the effects of SD and maximum on the risk of combined macro- and microvascular were similar between the subgroups of patients with and without changes in BP-lowering medication (Figure X in the online-only Data Supplement). The risks for the components of the major macrovascular and microvascular outcomes are shown in Figures 3 and 4, respectively. Of note, each higher SD level of was significantly associated with higher risks of myocardial infarction, cardiovascular death, and new or worsening nephropathy (all P<5 for trend). A similar pattern was observed for the outcome new or worsening retinopathy, although the association did not reach statistical significance (P=6 for trend). In contrast, SD levels of did not show any clear associations with stroke. A higher maximum was significantly associated with an elevated risk of myocardial infarction (P=4 for trend). Similar but nonsignificant trends were observed for cardiovascular death and new or worsening retinopathy (both P<0.1 for trend). The risk of stroke was clearly elevated in the highest tenth in comparison with the lowest tenth (HR, 3.27; 95% CI, 6 15; P=4), although the association did not show a clear dose response relationship (P=0.26 for trend). On the other hand, maximum did not show any association with new or worsening nephropathy. The effects of SD and maximum, and the mean, as well, considered as continuous variables are shown in Table 2. In the analysis with adjustment for age, sex, and randomized treatments (model 1), all parameters were

1328 Circulation September 17, 2013 Table 1. Characteristics of ADVANCE Participants Analyzed and Excluded in the Present Study Variables Total (n=11 140) Analyzed (n=8811) Excluded (n=2329) Demographic factors at registration Age, y 66 (6) 66 (6) 67 (7)* Female 4735 (43) 3735 (42) 1000 (43) Residence in Asian countries (China, India, 4136 (37) 3263 (37) 873 (37) Malaysia, and Philippines) Medical and lifestyle history at registration Currently treated hypertension 7655 (69) 6046 (69) 1609 (69) Duration of diabetes mellitus, y 7.9 (6.4) 7.7 (6.3) 8.7 (6.7)* History of major macrovascular disease 3590 (32) 2772 (31) 818 (35)* History of major microvascular disease 1152 (10) 775 (9) 377 (16)* Current smoking 1682 (15) 1325 (15) 357 (15) Current alcohol drinking 3396 (30) 2730 (31) 666 (29)* BP at registration, mm Hg 145 (22) 145 (21) 146 (22)* DBP, mm Hg 81 (11) 81 (11) 80 (11)* parameters during the measurement period Mean, mm Hg 138 (16) 137 (15) 140 (17)* SD, mm Hg 11 (5) 11 (5) 12 (6)* CV, % 8.0 (3.6) 7.9 (3.4) 8.3 (4.2)*, mm Hg 153 (19) 152 (19) 154 (21)* Risk factors at the 24-mo visit Heart rate, bpm 74 (12) 73 (12) 74 (12)* Hemoglobin A1c, % 7.0 (1.3) 7.0 (1.2) 7.3 (1.4)* Total cholesterol, mmol/l 4.9 (1.1) 4.9 (1.1) (1.2)* Triglycerides, mmol/l 1.6 (1.1 2.2) 1.6 (1.1 2.2) 1.6 (1.1 2.2) Body mass index, kg/m 2 28.2 (5.2) 28.3 (5.2) 27.9 (5.3)* Randomized treatments Active BP-lowering treatment with 5569 (50) 4418 (50) 1151 (49) perindopril-indapamide Intensive blood glucose control 5571 (50) 4479 (51) 1092 (47)* Additional BP-lowering treatments at the 24-mo visit β-blockers 3336 (30) 2619 (30) 717 (31) Calcium-channel blockers 3797 (34) 2993 (34) 804 (35) Diuretics 1793 (16) 1263 (14) 530 (23)* Angiotensin-converting enzyme inhibitors 5640 (51) 4553 (52) 1087 (47)* Angiotensin II receptor blockers 804 (7) 610 (7) 194 (8)* Other antihypertensive agents 1161 (10) 910 (10) 251 (11) Any BP-lowering agents 8543 (77) 6752 (77) 1791 (77) Blood glucose treatments at the 24-mo visit Oral hypoglycemic agents (gliclazide and others) 10 453 (94) 8348 (95) 2105 (90)* Insulin 1944 (17) 1520 (17) 424 (18) Values are mean (SD) for continuous variables (except for triglycerides), median (interquartile range) for triglycerides, and number (%) for categorical variables. BP indicates blood pressure; CV, coefficient of variation; DBP, diastolic blood pressure;, systolic blood pressure; and SD, standard deviation. *P<5 vs analyzed patients. History of myocardial infarction or stroke. Macroalbuminuria at baseline or a history of microvascular eye disease (proliferative retinopathy, macular edema, diabetes mellitus related blindness, or retinal photocoagulation therapy). Defined using values at 3, 4, 6, 12, 18, and 24 months after randomization. When data at the 24-month visit were missing, the latest data before the 24-month visit were applied. Randomized BP-lowering treatment with perindopril-indapamide was not included.

Hata et al Blood Pressure Variability in Type 2 Diabetes Mellitus 1329 SD Combined macro- and microvascular P for trend <01 Major macrovascular (407 ) P for trend = 02 Major microvascular (476 ) P for trend = 2 All-cause mortality (846 ) (374 ) 3.5 P for trend <01 Max 4.5 3.5 P for trend = 05 9.0 P for trend = 05 P for trend = 8 8.0 4.5 7.0 3.5 Tenths of maximum Tenths of maximum Tenths of maximum positively associated with the risks of major macrovascular and microvascular (P<01 for all). These associations remained significant even after adjustments for other confounding factors and mean (models 2 and 3), with the exception of maximum, which failed to reach statistical significance for the outcome microvascular in model 3. SD and maximum were positively associated with all-cause mortality (both P<01 in all models), whereas mean did not show a clear association with all-cause mortality. When major macrovascular were divided into their components, greater SD and maximum were significantly associated with elevated risks of myocardial infarction and cardiovascular death in all models. However, no significant associations were observed for stroke in model 3. With regard to the components of major microvascular, the associations of SD and maximum with nephropathy and retinopathy did not reach a level of significance in model 3. These associations were essentially unchanged when we determined visit-to-visit variability using CV instead of SD. Discussion The present large-scale observational analysis of 8811 patients with type 2 diabetes mellitus from the ADVANCE trial has demonstrated that the levels of visit-to-visit variability P for trend = 3 Tenths of maximum Figure 2. Hazard ratios and 95% confidence intervals (CIs) for major macrovascular and microvascular and death according to tenths of standard deviation (SD) or maximum (Max) of systolic blood pressure (). values measured at 6 occasions (from 3 24 months after randomization) were used to determine SD and maximum. Each parameter was categorized into 10 groups according to the tenths within each BP-lowering treatment group. The ranges of SD were 0.6 to 5.2, 5.3 to 6.8, 6.9 to 8.1, 8.2 to 9.3, 9.4 to 1, 10.6 to 11.7, 11.8 to 13.2, 13.3 to 15.2, 15.3 to 18.0, and 18.1 to 47.3 mm Hg in the placebo group; 0.4 to, 5.1 to 6.6, 6.7 to 7.7, 7.8 to 8.8, 8.9 to 9.9, 1 to 1, 11.1 to 1, 12.6 to 14.2, 14.3 to 16.7, and 16.8 to 33.5 mm Hg in the active group. The ranges of maximum were 97.5 to 133.5, 13 to 14, 14 to 144.5, 14 to 149.5, 15 to 153.5, 15 to 158.5, 159.0 to 16, 164.5 to 17, 17 to 18, and 18 to 26 mm Hg in the placebo group; 9 to 127.0, 127.5 to 133.5, 13 to 138.0, 138.5 to 14, 14 to 146.5, 147.0 to 15, 15 to 15, 156.5 to 16, 16 to 17, and 17 to 247.0 mm Hg in the active group. Hazard ratios were adjusted for age, sex, randomized blood pressure lowering intervention, randomized glucose control intervention, region of residence, duration of diabetes mellitus, current smoking, current alcohol drinking, heart rate, total cholesterol, log of triglycerides, body mass index, use of β-blockers, use of calcium-channel blockers, and mean during the measurement period. defined by using SD and CV are positively associated with the risks of major macrovascular, microvascular, and death. These associations persisted after the adjustment for mean, other cardiovascular risk factors, and randomized treatments. Similar associations were also observed for maximum. Previous studies using the data from ambulatory BP monitoring reported that short-term or circadian variability of BP was an important prognostic factor of cardiovascular outcomes. 14 17 Similarly, a number of observational studies have investigated the impact of long-term or visit-to-visit BP variability on the risks of cardiovascular outcomes. 5,18 21 Rothwell and colleagues 5 recently reported that visit-to-visit variability was a strong predictor of cardiovascular disease among patients with transient ischemic attack or stroke and among hypertensive patients in the ASCOT-BPLA trial. The Ohasama Study also reported that day-by-day variability was associated with cardiovascular and all-cause death in a general population from Japan. 18 However, there has been little investigation of the effects of BP variability on cardiovascular among patients with type 2 diabetes mellitus. 19 The analyses reported here expand the findings of previous studies and suggest that the association of visit-to-visit variability with cardiovascular is also observed in patients with type 2 diabetes mellitus.

1330 Circulation September 17, 2013 Myocardial infarction Stroke Cardiovascular death (125 ) (176 ) (170 ) SD 3.5 P for trend = 3 P for trend = 8 P for trend = 02 Max 14 1 2 3 4 5 6 7 8 910 P for trend = 4 12 P for trend = 0.26 12 P for trend = 7 Figure 3. Hazard ratios and 95% confidence intervals (CIs) for components of macrovascular outcomes according to tenths of standard deviation (SD) or maximum (Max) of systolic blood pressure (). The definition and categorization of each parameter and the covariates in the multivariable model were the same as in Figure 2. 12 10 8 6 4 2 10 8 6 4 2 10 8 6 4 2 0 Tenths of maximum 0 Tenths of maximum 0 1 2 3 4 5 6 7 8 910 Tenths of maximum In the present analysis of the ADVANCE trial, visitto-visit variability was clearly associated with myocardial infarction and cardiovascular death, but not with stroke. In contrast, Rothwell and colleagues 5 reported that visit-to-visit variability was predictive of both stroke SD Max 7.0 New or worsening nephropathy (230 ) P for trend = 4 1 2 3 4 5 6 7 8 910 P for trend = 0.24 1 2 3 4 5 6 7 8 910 Tenths of maximum 3.5 8.0 7.0 New or worsening retinopathy (266 ) P for trend = 6 1 2 3 4 5 6 7 8 910 P for trend = 7 1 2 3 4 5 6 7 8 910 Tenths of maximum Figure 4. Hazard ratios and 95% confidence intervals (CIs) for components of microvascular outcomes according to tenths of standard deviation (SD) or maximum (Max) of systolic blood pressure (). The definition and categorization of each parameter and the covariates in the multivariable model were the same as in Figure 2. and coronary heart disease in patients with previous transient ischemic attack or stroke and in ASCOT-BPLA. The Ohasama Study showed that day-by-day variability was associated with stroke mortality but not with cardiac mortality in a general Japanese population. 18 In a hospitalbased case-control study of Japanese patients, the variability of office was higher in patients with stroke than in controls, 20 but not in patients with myocardial infarction. 21 The reasons for the differences among ADVANCE and previous studies are not clear but may be associated with underlying differences in the study populations, in the definitions of BP variability, in the length of follow-up, and in the covariates included in the multivariable analyses, or the differences may be due to chance associations. The impact of BP variability on stroke may be stronger in Asian populations or nondiabetic populations than in non-asian or diabetic populations. In addition, the Ohasama Study 18 was based on home BP in which white coat effect was absent, but other studies, including ADVANCE, used BP measured mainly at the clinic or hospital. Another new and important finding of the present analysis was that visit-to-visit variability of clearly predicted the future development of major microvascular complications among patients with type 2 diabetes mellitus. Although there has been limited evidence of the association of BP variability with the microvascular complications of diabetes mellitus, a cross-sectional study of 422 Japanese patients with type 2 diabetes mellitus has shown a close association between variability and the prevalence of albuminuria. 22 The Diabetes Control and Complications Trial also reported that SD in and diastolic blood pressure were predictive of future development of nephropathy among patients with type 1 diabetes mellitus. 23 These findings support the hypothesis that BP variability is an important predictor of microvascular complications among patients with type 2 diabetes mellitus.

Hata et al Blood Pressure Variability in Type 2 Diabetes Mellitus 1331 Table 2. Effects of 1-SD Increments in Mean, SD, CV, and on Primary and Secondary Outcomes No. of Model 1* Model 2 Model 3 HR (95% CI) P HR (95% CI) P HR (95% CI) P Combined 846 macro- and microvascular Mean 1.28 (1.20 1.37) <01 1.25 (1.17 1.34) <01 1.25 (1.17 1.34) <01 SD 1.22 (1.15 1.30) <01 1.20 (1.12 1.28) <01 1.14 (6 1.22) <01 CV 1.16 (9 1.24) <01 1.14 (7 1.22) <01 1.14 (6 1.21) <01 1.30 (1.22 1.39) <01 1.28 (1.20 1.37) <01 1.27 (9 1.48) 02 Major 407 macrovascular Mean 1.24 (1.12 1.36) <01 1.20 (9 1.33) <01 1.20 (9 1.33) <01 SD 1.25 (1.14 1.37) <01 1.23 (1.12 1.34) <01 1.18 (7 1.30) <01 CV 1.20 (1.10 1.32) <01 1.19 (8 1.30) <01 1.18 (7 1.29) <01 1.28 (1.17 1.41) <01 1.25 (1.14 1.38) <01 1.37 (1.10 1.70) 05 Major 476 microvascular Mean 1.33 (1.22 1.45) <01 1.30 (1.18 1.42) <01 1.30 (1.18 1.42) <01 SD 1.20 (1.10 1.31) <01 1.17 (8 1.28) <01 1.10 (1 1.21) 3 CV 1.13 (4 1.23) 06 1.11 (2 1.21) 2 1.10 (1 1.20) 3 1.33 (1.23 1.45) <01 1.31 (1.20 1.43) <01 1.22 (0.99 1.49) 6 All-cause 374 mortality Mean 9 (0.99 1.21) 9 6 (0.96 1.18) 0.26 6 (0.96 1.18) 0.26 SD 1.31 (1.19 1.43) <01 1.29 (1.17 1.41) <01 1.29 (1.17 1.43) <01 CV 1.30 (1.18 1.42) <01 1.28 (1.16 1.40) <01 1.28 (1.16 1.40) <01 1.18 (7 1.30) 01 1.16 (4 1.28) 06 7 (1.26 1.96) <01 Myocardial 125 infarction Mean 1.30 (1.10 4) 02 1.26 (5 0) 1 1.26 (5 0) 1 SD 1.39 (1.19 1.63) <01 1.37 (1.17 1.60) <01 1.32 (1.11 5) 01 CV 1.32 (1.12 5) <01 1.30 (1.11 3) 01 1.29 (1.10 2) 02 1.38 (1.17 1.62) <01 1.34 (1.13 9) <01 6 (7 2.26) 2 Stroke 176 Mean 1.35 (1.17 5) <01 1.35 (1.17 6) <01 1.35 (1.17 6) <01 SD 1.18 (3 1.36) 2 1.17 (1 1.35) 3 8 (0.93 1.25) 0.30 CV 1.10 (0.95 1.27) 0.19 9 (0.95 1.26) 0.23 8 (0.93 1.25) 0.30 1.36 (1.19 6) <01 1.37 (1.19 9) <01 1.28 (0.92 1.79) 0.15 Cardiovascular 170 death Mean 1.15 (0.99 1.34) 6 1.10 (0.94 1.28) 0.25 1.10 (0.94 1.28) 0.25 SD 1.35 (1.18 4) <01 1.30 (1.14 1.49) <01 1.30 (1.13 0) <01 CV 1.33 (1.16 2) <01 1.29 (1.13 1.48) <01 1.29 (1.12 1.48) <01 1.24 (7 1.43) 04 1.19 (2 1.38) 3 5 (1.12 2.15) 09 (Continued)

1332 Circulation September 17, 2013 Table 2. Continued. No. of Model 1* Model 2 Model 3 HR (95% CI) P HR (95% CI) P HR (95% CI) P New or 230 worsening nephropathy Mean 1.49 (1.32 1.68) <01 1.37 (1.21 6) <01 1.37 (1.21 6) <01 SD 1.27 (1.12 1.43) <01 1.19 (6 1.35) 05 1.11 (0.97 1.25) 0.12 CV 1.16 (2 1.31) 2 1.11 (0.98 1.26) 0.10 1.10 (0.97 1.24) 0.15 1.48 (1.31 1.66) <01 1.36 (1.20 5) <01 1.16 (0.87 5) 0.31 New or 266 worsening retinopathy Mean 1.24 (1.10 1.39) <01 1.26 (1.12 1.42) <01 1.26 (1.12 1.42) <01 SD 1.18 (6 1.32) 04 1.18 (5 1.33) 04 1.12 (0.99 1.26) 6 CV 1.14 (1 1.28) 3 1.13 (0 1.27) 4 1.12 (0 1.26) 6 1.26 (1.13 1.42) <01 1.29 (1.15 1.46) <01 1.29 (0.98 1.70) 7 HR (95% CI) per increase of 1 SD for each parameter was shown. SD values in the placebo group were 14.6 mm Hg for mean, 5.3 mm Hg for SD, 3.5% for CV, and 18.9 mm Hg for maximum. SD values in the active group were 14.8 mm Hg for mean, 4.8 mm Hg for SD, 3.4% for CV, and 18.1 mm Hg for maximum. CI indicates confidence interval; CV, coefficient of variation; HR, hazard ratio;, systolic blood pressure; and SD, standard deviation. *Model 1 was adjusted for age, sex, randomized blood pressure lowering intervention, and randomized glucose control intervention. Model 2 was adjusted for age, sex, randomized blood pressure lowering intervention, randomized glucose control intervention, region of residence, duration of diabetes mellitus, current smoking, current alcohol drinking, heart rate, total cholesterol, log of triglycerides, body mass index, use of β-blockers, and use of calciumchannel blockers. Model 3 was adjusted for all variables in model 2 and mean. The impact of variability on combined major macrovascular and microvascular seemed to be weaker than that of mean (Table 2). However, it does not mean that variability is less important. Because the effects of variability were independent of mean, higher variability may be an important factor for the prediction of future vascular and a potential therapeutic target, even in patients without hypertension. In addition, Rothwell and colleagues 5 observed positive associations of maximum with risks of cardiovascular. In the present study, we confirmed similar findings in patients with type 2 diabetes mellitus. The impacts of maximum were strong and independent of a wide set of potential confounding variables. They also appeared to be independent of mean, although the strong correlation between mean and maximum (r=0.91, Table III in the online-only Data Supplement), and the relatively wide CIs for the HRs comparing the extreme tenths of maximum in Figures 2 to 4, suggests that caution should be applied to our estimates and 95% CIs of effects of maximum adjusted for mean. There are several possible mechanisms to explain the link between visit-to-visit BP variability or maximum and macrovascular and microvascular. First, BP variability may be a marker of arterial stiffness and thus associated with vascular outcomes. The elevation in variability of BP has been associated with reduced compliance of large elastic arteries, 7 which is an intermediate stage of vascular disease and a strong predictor of the future vascular. 9 Second, instability of BP may reflect abnormal autonomic function. 8 Increased sympathetic activity may directly promote the development of arterial and organ damage 8 and increase the risk of vascular. 10 However, in the present study, the effects of variability and maximum of remained significant even after the adjustment for heart rate, a marker of autonomic function. Third, animal studies have suggested that elevated BP variability causes direct endothelial damage, renin-angiotensin system activation, inflammation and cardiomyocyte apoptosis augmentation, and thus cardiovascular remodeling and end-organ damage. 24 It is reasonably considered that maximum is also a marker of arterial stiffness or abnormal autonomic function, or a direct cause of vascular dysfunction. Fourth, poor adherence with BP-lowering medications and inadequate BP control might have played a role in the link between increased BP variability or maximum and poor clinical outcomes, although risk estimates were essentially unchanged even in the sensitivity analysis that excluded patients who discontinued BP-lowering intervention during the measurement period (Figure II in the online-only Data Supplement). In the previous analysis of the ASCOT-BPLA trial, risk associations were actually strengthened after the exclusion of poorly compliant patients. 5 Fifth, changes in BP-lowering treatment regimen during the measurement period might affect the association between variability and outcomes. In the ADVANCE protocol, the dose of active perindopril/indapamide was doubled at the 3-month visit, and the use of other BP-lowering drugs was at the discretion of the physician. However, the sensitivity analysis that excluded BP measurements at 3- and 4-month visits (Figure IV in the online-only Data Supplement) was similar to the original analyses, and the impacts of variability were comparable between the subgroups of patients with and

Hata et al Blood Pressure Variability in Type 2 Diabetes Mellitus 1333 without changes in BP-lowering treatment (Figure X in the online-only Data Supplement). The patients with higher variability were more likely to end up on multiple drugs than patients with less variability (Figure I in the online-only Data Supplement). Therefore, we consider that variability predicted an increased risk of vascular independent of the changes in treatment, even though the patients with higher variability were treated more intensively than average. The method to calculate visit-to-visit variability may be debatable. was measured twice at each occasion, and the mean value was used in the present study. In the ASCOT- BPLA trial, 5 analyses based on the mean of 3 measurements and the mean of 2 (the second and the third) measurements revealed no material difference. Accordingly, we believe that our data based on 2 measurements is satisfactory. Next, the calculation of variability was based on 6 visits and their intervals were not equal (ie, 3, 4, 6, 12, 18, and 24 months).the sensitivity analyses using BP values from equally spaced visits (ie, 6, 12, 18, and 24 months; Figure IV in the online-only Data Supplement) or different number of visits (Figures V and VI in the online-only Data Supplement) made no difference. There are substantial gaps in our understanding of the best ways to apply this knowledge to the practical management of BP in patients with type 2 diabetes mellitus, and more work is needed to determine the best and simplest way to evaluate visit-to-visit BP variability in routine clinical practice. In addition, it remains uncertain how best to treat patients with type 2 diabetes mellitus who have high BP variability or maximum. Recent analyses of individual patient data from 2 trials 25 and a systematic review 26 suggested that calcium-channel blockers and nonloop diuretic drugs were particularly effective for reducing BP variability, but no studies have reported on optimal therapeutic ranges of BP variability. Although this is the first and the largest prospective study to have demonstrated the clear effects of visit-to-visit variability and maximum on macrovascular and microvascular complications in type 2 diabetes mellitus, the study still had relatively limited power to show corresponding definitive evidence about the effects of variability on each component of the microvascular outcome (nephropathy and retinopathy). However, the HRs of SD and CV of for nephropathy and for retinopathy were similar to those for composite microvascular (Table 2). The adjustment for covariates in multivariable analysis might be substantively incomplete, and the degree of independence of variability or maximum might be overestimated as a consequence. Moreover, the present evaluation was performed in the context of a clinical trial and may not be generalizable to an unselected population with type 2 diabetes mellitus. Nevertheless, the exclusion of subjects with macrovascular or microvascular disease before the registration or within the first 24 months from the present analyses makes the findings more representative of a general population with type 2 diabetes mellitus (Table III in the online-only Data Supplement). In conclusion, visit-to-visit variability in and maximum were predictive of macrovascular complications and of mortality in a population with type 2 diabetes mellitus from 20 countries worldwide. In addition we report that this variability is predictive of microvascular. The prognostic values of BP variability and maximum were independent of mean BP and of the other major cardiovascular risk factors. Sources of Funding The ADVANCE trial was funded by grants from the National Health and Medical Research Council of Australia and Institut de Recherches Internationales Servier. Disclosures Dr Hata holds an International Society of Hypertension Visiting Postdoctoral Award from the Foundation for High Blood Pressure Research of Australia and the Banyu Fellowship Program from the Banyu Life Science Foundation International. Dr Arima holds an Australian Research Council Future Fellowship. Dr Rothwell is funded by a National Institutes of Health Research (NIHR, the United Kingdom) Senior Investigator Award, a Wellcome Trust Senior Investigator Award, and the NIHR Biomedical Research Center, Oxford. Dr Woodward holds a Senior Research Fellowship from the National Health and Medical Research Council of Australia and has recently received consulting fees from Roche and lecturing fees from Servier. Dr Zoungas holds a National Health and Medical Research Council of Australia Health Professional Research Fellowship and reports being a member of an advisory board for Merck and receiving lecture fees from Servier, GlaxoSmithKline, and Sanofi-Aventis and grant support from Novo Nordisk and Pfizer. Dr Anderson holds a Senior Principal Research Fellowship from the National Health and Medical Research Council of Australia and has received consulting and lecture fees from Astra Zeneca, Boehringer Ingelheim, Novo Nordisk, Sanofi-Aventis and Servier. Dr Patel holds a National Heart Foundation of Australia Career Development award and reports receiving lecture fees from Servier, Pfizer, and Abbott and grant support from Servier, Pfizer, and Sanofi-Aventis. Dr Neal holds an Australian Research Council Future Fellowship. Dr Williams is an NIHR Senior Investigator and his research is supported by the NIHR University College London Hospitals Biomedical Research Centre and has received lecture fees from Novartis, Boehringer Ingelheim, and Servier. Dr MacMahon reports being a member of advisory boards for Servier, Pfizer, and Novartis and receiving lecture fees from Servier and Pfizer and grant support from Servier, Pfizer, and Novartis. Dr Chalmers reports receiving research grants and lecture fees from Servier. References 1. Adler AI, Stratton IM, Neil HA, Yudkin JS, Matthews DR, Cull CA, Wright AD, Turner RC, Holman RR. Association of systolic blood pressure with macrovascular and microvascular complications of type 2 diabetes (UKPDS 36): prospective observational study. BMJ. 2000;321:412 419. 2. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ; Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. National Heart, Lung, and Blood Institute; National High Blood Pressure Education Program Coordinating Committee. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension. 2003;42:1206 1252. 3. Mancia G, De Backer G, Dominiczak A, Cifkova R, Fagard R, Germano G, Grassi G, Heagerty AM, Kjeldsen SE, Laurent S, Narkiewicz K, Ruilope L, Rynkiewicz A, Schmieder RE, Boudier HA, Zanchetti A, Vahanian A, Camm J, De Caterina R, Dean V, Dickstein K, Filippatos G, Funck-Brentano C, Hellemans I, Kristensen SD, McGregor K, Sechtem U, Silber S, Tendera M, Widimsky P, Zamorano JL, Erdine S, Kiowski W, Agabiti-Rosei E, Ambrosioni E, Lindholm LH, Viigimaa M, Adamopoulos S, Agabiti-Rosei E, Ambrosioni E, Bertomeu V, Clement D, Erdine S, Farsang C, Gaita D, Lip G, Mallion JM, Manolis AJ, Nilsson PM, O Brien E, Ponikowski P, Redon J, Ruschitzka F, Tamargo J, van Zwieten P, Waeber B, Williams B; Management of Arterial Hypertension of the European Society of Hypertension; European Society of Cardiology. 2007 Guidelines for the Management of Arterial Hypertension: The Task Force for the Management of Arterial Hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). J Hypertens. 2007;25:1105 1187.

1334 Circulation September 17, 2013 4. Diabetes Australia Guideline Development Consortium. National Evidence Based Guidelines for the Management of Type 2 Diabetes Mellitus: Blood Pressure Control in Type 2 Diabetes. National Health and Medical Research Council (Australia). http://www.nhmrc.gov.au/guidelines/publications/subject/diabetes Accessed on August 3, 2012. 5. Rothwell PM, Howard SC, Dolan E, O Brien E, Dobson JE, Dahlöf B, Sever PS, Poulter NR. Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet. 2010;375:895 905. 6. Rothwell PM. Limitations of the usual blood-pressure hypothesis and importance of variability, instability, and episodic hypertension. Lancet. 2010;375:938 948. 7. Imai Y, Aihara A, Ohkubo T, Nagai K, Tsuji I, Minami N, Satoh H, Hisamichi S. Factors that affect blood pressure variability. A communitybased study in Ohasama, Japan. Am J Hypertens. 1997;10:1281 1289. 8. Palatini P, Julius S. The role of cardiac autonomic function in hypertension and cardiovascular disease. Curr Hypertens Rep. 2009;11:199 205. 9. Stehouwer CD, Henry RM, Ferreira I. Arterial stiffness in diabetes and the metabolic syndrome: a pathway to cardiovascular disease. Diabetologia. 2008;51:527 539. 10. Vinik AI, Maser RE, Mitchell BD, Freeman R. Diabetic autonomic neuropathy. Diabetes Care. 2003;26:1553 1579. 11. ADVANCE Management Committee. Study rationale and design of ADVANCE: Action in Diabetes and Vascular Disease-Preterax and Diamicron MR Controlled Evaluation. Diabetologia. 2001;44:1118 1120. 12. ADVANCE Collaborative Group. Effects of a fixed combination of perindopril and indapamide on macrovascular and microvascular outcomes in patients with type 2 diabetes mellitus (the ADVANCE trial): a randomised controlled trial. Lancet. 2007;370:829 840. 13. ADVANCE Collaborative Group. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med. 2008;358:2560 2572. 14. Verdecchia P, Porcellati C, Schillaci G, Borgioni C, Ciucci A, Battistelli M, Guerrieri M, Gatteschi C, Zampi I, Santucci A, Santucci C, Reboldi G. Ambulatory blood pressure. An independent predictor of prognosis in essential hypertension. Hypertension. 1994;24:793 801. 15. Pringle E, Phillips C, Thijs L, Davidson C, Staessen JA, de Leeuw PW, Jaaskivi M, Nachev C, Parati G, O Brien ET, Tuomilehto J, Webster J, Bulpitt CJ, Fagard RH; Syst-Eur investigators. Systolic blood pressure variability as a risk factor for stroke and cardiovascular mortality in the elderly hypertensive population. J Hypertens. 2003;21:2251 2257. 16. Tamura K, Tsurumi Y, Sakai M, Tanaka Y, Okano Y, Yamauchi J, Ishigami T, Kihara M, Hirawa N, Toya Y, Yabana M, Tokita Y, Ohnishi T, Umemura S. A possible relationship of nocturnal blood pressure variability with coronary artery disease in diabetic nephropathy. Clin Exp Hypertens. 2007;29:31 42. 17. Pierdomenico SD, Di Nicola M, Esposito AL, Di Mascio R, Ballone E, Lapenna D, Cuccurullo F. Prognostic value of different indices of blood pressure variability in hypertensive patients. Am J Hypertens. 2009;22:842 847. 18. Kikuya M, Ohkubo T, Metoki H, Asayama K, Hara A, Obara T, Inoue R, Hoshi H, Hashimoto J, Totsune K, Satoh H, Imai Y. Day-by-day variability of blood pressure and heart rate at home as a novel predictor of prognosis: the Ohasama Study. Hypertension. 2008;52:1045 1050. 19. Hsieh YT, Tu ST, Cho TJ, Chang SJ, Chen JF, Hsieh MC. Visit-to-visit variability in blood pressure strongly predicts all-cause mortality in patients with type 2 diabetes: a 5 5-year prospective analysis. Eur J Clin Invest. 2012;42:245 253. 20. Hata Y, Kimura Y, Muratani H, Fukiyama K, Kawano Y, Ashida T, Yokouchi M, Imai Y, Ozawa T, Fujii J, Omae T. Office blood pressure variability as a predictor of brain infarction in elderly hypertensive patients. Hypertens Res. 2000;23:553 560. 21. Hata Y, Muratani H, Kimura Y, Fukiyama K, Kawano Y, Ashida T, Yokouchi M, Imai Y, Ozawa T, Fujii J, Omae T. Office blood pressure variability as a predictor of acute myocardial infarction in elderly patients receiving antihypertensive therapy. J Hum Hypertens. 2002;16:141 146. 22. Okada H, Fukui M, Tanaka M, Inada S, Mineoka Y, Nakanishi N, Senmaru T, Sakabe K, Ushigome E, Asano M, Yamazaki M, Hasegawa G, Nakamura N. Visit-to-visit variability in systolic blood pressure is correlated with diabetic nephropathy and atherosclerosis in patients with type 2 diabetes. Atherosclerosis. 2012;220:155 159. 23. Kilpatrick ES, Rigby AS, Atkin SL. The role of blood pressure variability in the development of nephropathy in type 1 diabetes. Diabetes Care. 2010;33:2442 2447. 24. Su DF. Treatment of hypertension based on measurement of blood pressure variability: lessons from animal studies. Curr Opin Cardiol. 2006;21:486 491. 25. Rothwell PM, Howard SC, Dolan E, O Brien E, Dobson JE, Dahlöf B, Poulter NR, Sever PS; ASCOT-BPLA and MRC Trial Investigators. Effects of beta blockers and calcium-channel blockers on withinindividual variability in blood pressure and risk of stroke. Lancet Neurol. 2010;9:469 480. 26. Webb AJ, Fischer U, Mehta Z, Rothwell PM. Effects of antihypertensivedrug class on interindividual variation in blood pressure and risk of stroke: a systematic review and meta-analysis. Lancet. 2010;375:906 915. Clinical Perspective Elevated blood pressure (BP) is a modifiable risk factor for macrovascular diseases such as coronary heart disease and stroke, and microvascular complications of diabetes mellitus such as nephropathy and retinopathy. In the present analyses, we proved that visit-to-visit variability of systolic blood pressure (), defined as the standard deviation or coefficient of variation during 24 months, was a significant risk factor for major macrovascular and microvascular complications of diabetes mellitus, and for all-cause mortality, as well, by using data from a large-scale randomized controlled trial of patients with type 2 diabetes mellitus, Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE). These associations were independent of usual BP, defined as the mean during the same period, and other cardiovascular risk factors and BP-lowering treatment. Similar associations were found for maximum during the same period. Our findings suggest that the higher variability or maximum of may be an important factor for the prediction of future vascular and a potential therapeutic target, even in patients without hypertension. However, there are substantial gaps in our understanding of the best ways to apply this knowledge to the practical management of BP in patients with type 2 diabetes mellitus, and it remains uncertain how best to treat patients with type 2 diabetes mellitus who have high BP variability or maximum. More work is needed to determine how to evaluate and treat the instability of BP in routine clinical practice.