Visual Abstract
Abstract
Background Although hypertension is a well known risk factor for CKD, few studies have evaluated the association between temporal trends of systolic BP and kidney function decline in persons without hypertension.
Methods We studied whether changes in systolic BP over time could influence incident CKD development in 4643 individuals without CKD and hypertension participating in the Korean Genome and Epidemiology Study, a prospective community-based cohort study. Using group-based trajectory modeling, we categorized three distinct systolic BP trajectories: decreasing, stable, and increasing. The primary outcome was incident CKD development, defined as two consecutive eGFR measurements <60 ml/min per 1.73 m2.
Results Among participants with an increasing systolic BP trajectory, systolic BP increased from 105 to 124 mm Hg. During 31,936 person-years of follow-up (median 7.7 years), 339 participants developed incident CKD. CKD incidence rates were 8.9, 9.6, and 17.8 cases per 1000 person-years in participants with decreasing, stable, and increasing systolic BP trajectories, respectively. In multivariable cause-specific Cox analysis, after adjustment of baseline eGFR, systolic BP, and other confounders, increasing systolic BP trajectory associated with a 1.57-fold higher risk of incident CKD (95% confidence interval, 1.20 to 2.06) compared with a stable trajectory. There was a significant effect modification of baseline systolic BP on the association between systolic BP trajectories and CKD risk (P value for interaction =0.02), and this association was particularly evident in participants with baseline systolic BP <120 mm Hg. In addition, increasing systolic BP trajectory versus a stable trajectory was associated with higher risk of new development of albuminuria.
Conclusions Increasing systolic BP over time without reaching the hypertension threshold is associated with a significantly increased risk of incident CKD in healthy adults.
CKD is a global public health problem, and it affects 8%–16% of the worldwide population.1–⇓3 Because CKD is inevitably accompanied by increased mortality, many comorbidities, and high medical cost,2,4,5 currently available measures, including low-salt diet, regular exercise, smoking cessation, BP control, lipid management, and strict blood glycemic control in persons with diabetes, should be encouraged to prevent CKD.6 Hypertension is an established risk factor of CKD and the second most common cause of ESKD following diabetes.3,7–⇓9 Not only hypertension but also, elevated systolic BP (SBP) ranging from 120 to 139 mm Hg, known as high normal or prehypertension, are associated with a high risk of CKD.10–⇓⇓13 However, this association was on the basis of the findings of prospective or cross-sectional population-based studies using baseline BP readings alone. Because these studies did not include BP data during the follow-up period, they could not reflect the effect of dynamic changes in BP.
Recently, several randomized, controlled trials (RCTs) have examined the effects of intensive BP control with a target SBP of <120 mm Hg on cardiovascular events and mortality compared with a standard target of <140 mm Hg among patients with a high cardiovascular risk or type 2 diabetes.14,15 Although intensive BP control reduced cardiovascular events and mortality in one study, this lower target resulted in more adverse kidney outcomes, raising concerns about intensive BP lowering causing kidney function decline. However, these RCTs included patients with a substantial cardiovascular risk and baseline SBP ≥130 mm Hg, and it is unknown whether elevated BP without a diagnosis of hypertension should be lowered considering the risk of CKD in adults with a low cardiovascular risk. Individuals with prehypertension are generally not treated with antihypertensive drugs, and they are advised to adopt lifestyle modifications instead. However, many of these individuals do not take the potential hazard of mildly elevated BP into serious consideration and do not place much emphasis on the importance of lifestyle modifications at this stage. In addition, as the magnitude of BP change is relatively small in this population, slight elevations of BP are often ignored, although health care providers recommend dietary changes and exercise to prevent hypertension.
Therefore, using a prospective longitudinal cohort database, we determined the SBP trajectories during the early follow-up period and analyzed the association of these SBP trends with the subsequent development of CKD among Korean middle-aged adults without hypertension.
Methods
Ethics Statement
All participants were enrolled in the study voluntarily, and informed consent was obtained for all participants. This study was carried out in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Yonsei University Health System Clinical Trial Center (4–2019–1108).
Study Population
Data were retrieved from the Korean Genome and Epidemiology Study (KoGES), a prospective community-based study. The detailed cohort profiles and methods with respect to the development of the KoGES have been published elsewhere.16 Briefly, the participants of the KoGES consisted of middle-aged adults (age 40–69 years) in Ansan (urban area) or Ansung (rural area), located in the south of Seoul, Republic of Korea. A total of 10,030 participants were enrolled from 2001 to 2002 and followed up until 2014. The participants underwent anthropometric examinations and laboratory tests, completed health-related lifestyle questionnaires at baseline, and repeated the tests biennially up to 14 years from the baseline visit. For this study, we excluded participants who were treated with antihypertensive drugs or those with SBP of ≥140 mm Hg or diastolic BP (DBP) of ≥90 mm Hg at baseline BP measurement. Furthermore, individuals with known kidney disease, positive dipstick proteinuria (trace or greater), or eGFR<60 ml/min per 1.73 m2 were also excluded. The remaining 5822 persons were the subjects for the analysis of time-updated SBP. To determine SBP trajectories during the first 4 years of follow-up, participants who did not undergo more than one BP measurement or those who developed CKD during this period were further excluded. Thus, a total of 4643 participants were finally included in the analysis of BP trajectories (Supplemental Figure 1).
Data Collection and Measurements
Demographic and socioeconomic data, including age, sex, education, income, marital status, smoking history, alcohol consumption history, daily physical activity, food frequency questionnaire, and medical history, were collected at the baseline study. Income status was categorized into three groups: low, <$850 per month; middle, ≥$850 to <$1700 per month; and high, ≥$1700 per month. Education level was also divided into three groups: lower than middle school, middle school, and higher than middle school. Physical activity levels were surveyed using semiquantitative questionnaires and calculated as daily estimated metabolic equivalents of task (hours per day). Single-day dietary data for sodium (grams per day), potassium (grams per day), and total calorie intake (kilocalories per day) were estimated by semiquantitative 24-hour dietary recall food frequency questionnaire that was validated previously.17,18 Anthropometric measurements were performed by trained nurses using validated and standardized protocols and calibrated instruments. BP was measured by trained nurses using a mercury sphygmomanometer in the sitting position after at least a 5-minutes rest. BP was measured at least twice with >1-minute intervals between measurements and repeated until the difference between the last two measurements was <5 mm Hg. The mean of these two BP readings was recorded as the BP value at each visit. Hypertension was defined as SBP≥140 mm Hg, DBP≥90 mm Hg, or treatment with antihypertensive medications according to the Korean Society of Hypertension.19 Participants who were taking antidiabetic drugs, had glycated hemoglobin levels ≥6.5%, had fasting plasma glucose levels ≥126 mg/dl after ≥8 hours of fasting, or had a positive 75-g oral glucose tolerance test were considered as having diabetes. Cardiovascular disease was defined as having a history of myocardial infarction, coronary artery disease, congestive heart failure, peripheral artery disease, or cerebrovascular disease. Body mass index (BMI) was calculated as weight divided by the square of height in meters (kilograms per meter2). The homeostatic model assessment for insulin resistance (HOMA-IR) score was calculated using the fasting glucose and insulin levels.20
Blood samples obtained after an 8-hour fast were transported to Seoul Clinical Laboratories (Seoul, Republic of Korea) within 24 hours of sampling and analyzed for the following biochemical parameters: serum creatinine, BUN, glucose, glycated hemoglobin, albumin, total protein, HDL cholesterol (HDL-C), and C-reactive protein levels. Serum creatinine was measured by the Jaffe method during entire period. Thus, we converted nonisotope dilution-mass spectrometry creatinine to isotope dilution-mass spectrometry creatinine using the equation as previously suggested.21,22 Then, eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation.23 Urine samples were collected in the morning after the first void. Urine test was performed in fresh urine samples with a urine reagent strip using a semiautomatic urine analyzer that was regularly calibrated (URiSCAN Pro II; YD Diagnostics Corp., Seoul, Korea). The urine albumin amounts were determined as absent, trace, 1+, 2+, or 3+, which approximately correspond with urine albumin levels of <10, 10–20, >30, >100, and >500 mg/dl, respectively.
Exposure and Outcome
The primary exposure of interest was SBP trajectories. For the analysis, temporal SBP trends were determined by trajectory modeling with BP readings at baseline visit and subsequent biennial visits until the second follow-up visit (“SBP trend assessment window,” which was up to 4 years after the baseline visit) (Figure 1A). The primary outcome was de novo CKD development, which was defined as two consecutive eGFR values of <60 ml/min per 1.73 m2, the first of which was designated as the study end point. The secondary outcome included new development of albuminuria, which was defined as dipstick test for albumin trace or ≥1+ and the occurrence of death.
Study design and approach. We used the SBP trajectory model for the primary analysis (A). Additional two approaches with the SBP trajectory time-updated (B), and the time-updated SBP models were used for sensitivity analyses (C). In time-updated models, antihypertensive medication use, SBP, BMI, HDL-C, and HOMA-IR score were considered as time-varying covariates. The outcome assessment started from year 4, and people who developed CKD during the first 4 years were excluded in (A) the SBP trajectory model and (B) the SBP trajectory time-updated model. (C) In the time-updated SBP model, the outcome assessment started from year 2. Cumulative SBPs were determined by averaging the mean of SBP readings at any given visit and those from all prior visits.
Statistical Analyses
Statistical analyses were conducted using STATA version 16.1 (Stata Corporation, College Station, TX) and SAS version 9.4 (SAS Institute Inc., Cary, NC). To classify the trend of SBP over time, we used a group-based trajectory modeling, the method that has previously been used.24–⇓⇓⇓⇓29 This modeling assumes that the participants are made up of multiple trajectory groups that can simultaneously estimate probabilities for multiple trajectories. According to these assumptions, time-dependent covariates explain variation in the average trajectory within each group. We constructed models with different numbers of trajectory groups and different forms of potential trajectories (linear, quadratic, or cubic) using SAS Proc Traj procedure. The heterogeneity of the slopes of the SBP trajectory groups was tested using the “TRAJTEST” macro. Model fit was assessed with the following methods: (1) the Bayesian Information Criterion, (2) the number of participants in each trajectory (>5% of overall population), and (3) average probability of final group membership across the trajectory groups to assess discrimination power of individual model. We started a model with three trajectories and then, constructed the trajectory models with four and five. We evaluated different functional forms to enhance model fit, and we calculated the posterior predicted probabilities for each individual of being a member of a given trajectory group to assess model’s discrimination power. As the number of trajectory group increased, the Bayesian Information Criterion also decreased, and the numbers of participants in several trajectory groups were too small, resulting in less discrimination power across the trajectory groups in models with four or more trajectories (Supplemental Figure 2, Supplemental Table 1). As a result, three distinct SBP trends were identified using SBP readings from baseline to year 4 and were named according to their visual appearance and clinically meaningful BP trends: decreasing (n=1152), stable (n=2802), and increasing (n=689) patterns (Figure 2). The average probability of final group membership was 94%.
Predicted and observed SBP changes according to SBP trajectory. Three distinct SBP patterns by group-based trajectory modeling were concordant to observed SBP changes.
There were 492 participants with two BP readings only during the first 4 years, and we did not impute missing data for BP because this imputation method cannot capture a real trend of BP change and group-based trajectory modeling allows missing values. The median and interquartile range of BP measurement during entire period were seven (five to eight) times, and 74.6% of the participants undertook BP measurements seven times or more (Supplemental Table 2). We further repeated analysis with complete BP data until year 4 in the sensitivity analysis. There were missing values for other measures, such as metabolic equivalents of task, total energy consumption, sodium-potassium consumption rate, and HOMA-IR score, which were <5% of all measurements (Supplemental Table 3). We assumed missing data as missing at random. The multiple imputation by chained equations method was used to impute ten independent copies of the data for the multivariable Cox regression analyses. Hazard ratios (HRs), 95% confidence intervals (95% CIs), and P values for each model were obtained by fitting the model on each imputation dataset and combining the results by the Rubin rule.30
Continuous variables are expressed as either mean ± SD or median (interquartile rage), and categorical variables are expressed as number (percentage). The distribution of variables was ascertained using the Kolmogorov–Smirnov test. To compare differences between categorized groups, ANOVA and chi-squared tests were used for continuous and categorical variables, as appropriate. The Kruskal–Wallis test was used for variables with a skewed distribution. To compare the primary outcome of different SBP metrics, multivariable cause-specific Cox proportional hazard models were constructed. Model 1 represents unadjusted HRs. Model 2 included demographic characteristics, such as age, sex, BMI, baseline SBP, education status, income status, alcohol drinking, and smoking history. In model 3, further covariates were added to model 2, including HDL-C, serum albumin, C-reactive protein, and eGFR. Finally, in model 4, we included metabolic covariates in model 3, such as HOMA-IR score, physical activity level, dietary sodium/potassium consumption rate, and total energy consumption. Deaths that occurred before reaching the primary outcome were treated as a competing risk and censored. Patients who were lost to follow-up were censored at the date of the last examination. Furthermore, we examined the effect modification of SBP trajectories for CKD development in prespecified subgroups by age (<60 or ≥60 years), sex (man or woman), BMI (≤25 or >25 kg/m2), baseline SBP (<120 or ≥120 mm Hg), and presence of diabetes.
We conducted several sensitivity analyses to assess robustness our findings. Because there were differences in SBP levels between SBP trajectories during the follow-up period, we constructed two time-varying models with SBP trajectory and time-updated SBP. In the first model with SBP trajectory, three SBP trends as described above were used, and the subsequent SBP readings after SBP trend assessment window and other variables, such as antihypertensive drug use, BMI, HDL-C, and HOMA-IR score, that were repeatedly measured at all visits were treated as time-varying exposures (Figure 1B). In the second model, we used conventional time-updated SBP, which was determined by averaging the mean of SBP readings at any given visit and those from all prior visits (Figure 1C). In this analysis, time-updated SBP was analyzed as a continuous variable per 10-mm Hg increase and as a categorical variable by 10-mm Hg increments (<100, 100–120, 120–129, 130–139, and ≥140 mm Hg). Furthermore, we conducted additional sensitivity analyses by varying SBP exposure windows (from 2 to 6 years) and with fixed outcome assessment period of the following 4 years (Supplemental Figure 3). There may be adaptation effects that make first measurements inaccurate, which is also applicable to first-visit measurements. Thus, we further created models with 2- and 4-year SBP trajectories, in which the SBP reading at year 2 was considered a baseline (Supplemental Figure 3, D and E). To evaluate the changes of parameters, including eGFR and BP, over time according to SBP trajectories, we used a linear mixed model. Two-sided P<0.05 was considered significant.
Results
Baseline Characteristics
Table 1 shows the baseline characteristics according to three SBP trajectory groups. The mean age was 50.2 years, and 2243 (48.3%) were men. The mean SBP was 112.3 mm Hg, and 9.9% had diabetes. Participants in the increasing SBP trajectory group had lower BP. They were older, had less education, and had lower income than other groups. In addition, physical activity level was higher in the increasing SBP group, but there was no difference in BMI among three groups. When the temporal trends of BP were calculated using the linear mixed model according to SBP trajectories, the observed trends of BP were similar with the trends determined by trajectory analysis. Interestingly, SBP increased from 105 to 124 mm Hg in the increasing trajectory group, whereas it decreased from 120 to 106 mm Hg in the decreasing trajectory group within 2 years of the initial visit. The differences in BP between groups slightly decreased thereafter but remained significant throughout the entire follow-up period (Supplemental Figure 4). The proportions of participants with ≥120 and ≥140 mm Hg were consistently higher in the increasing SBP trajectory group than in the other groups throughout the observation period (Supplemental Figure 5).
Baseline characteristics of the study cohort according to SBP trajectory groups
Association of SBP Trends with CKD Development
During a median follow-up of 7.7 (6.1–7.8) years and 31,936 person-years, CKD occurred in 339 participants, with an overall incidence rate of 10.6 per 1000 person-years (Table 2). There were 71 (6.2%), 185 (6.6%), and 83 (12.0%) CKD events in the decreasing, stable, and increasing SBP trajectory groups, respectively, with corresponding incidence rates of 8.9 (95% CI, 7.1 to 11.2), 9.6 (95% CI, 8.3 to 11.1), and 17.8 (95% CI, 14.4 to 22.1) per 1000 person-years, respectively. The cumulative incidence curves also showed that the CKD event-free survival rate was significantly lower in the increasing SBP trajectory group than in the other two groups (log-rank test, P<0.001) (Supplemental Figure 6).
Incidence rates of CKD development according to SBP trajectory patterns
Cause-specific Cox proportional hazard models also showed significantly higher risk of CKD development in participants with increasing SBP trajectory than in those with a stable trajectory. After adjustment for confounding factors, increasing SBP trajectory was associated with a 1.6-fold higher risk of CKD development (model 4; HR, 1.57; 95% CI, 1.20 to 2.06; P=0.001) (Table 3). Spline regression analyses also showed a graded association of changes in SBP during SBP exposure period with the risk of incident CKD (Supplemental Figure 7). To evaluate the short-term association of increasing SBP trend during certain time periods, we varied SBP exposure periods (2, 4, and 6 years) and analyzed the association of initial SBP trends during these periods with subsequent risk of CKD during the following 4 years (Supplemental Figure 3). The results yielded similar findings that individual increasing SBP trajectory during each exposure period was significantly associated with the development of incident CKD (Supplemental Table 4).
HRs for CKD development according to SBP trajectory patterns
We further analyzed the association of DBP trajectory with CKD development. Similar to the results of the SBP model, increasing DBP was associated with higher risk of CKD development. However, this did not reach statistical significance (Supplemental Figure 8, Supplemental Table 5).
Subgroup Analyses
We examined the effect modification by an a priori–selected set of baseline characteristics, including age (<60 or ≥60 years), sex (man or woman), baseline SBP (≥120 or <120 mm Hg), BMI (<25 or ≥25 kg/m2), and the presence of diabetes. The significant association of SBP trends with the CKD outcome was consistent across subgroups by age, sex, presence of diabetes, and BMI. However, there was a significant interaction between SBP trajectory and baseline SBP (P value for interaction =0.02) for incident CKD. In the subgroup with SBP<120 mm Hg, participants with increasing SBP trajectory had a 1.38-fold higher risk of CKD than those with a stable trajectory (HR, 1.38; 95% CI, 1.05 to 1.82; P=0.02). In contrast, in the subgroup with SBP≥120 mm Hg, decreasing SBP trajectory was associated with a lower risk of CKD development (HR, 0.72; 95% CI, 0.54 to 0.96; P=0.03) (Figure 3).
Forest plot showing HRs and interaction P values for CKD development. There was a significant effect modification of baseline SBP on the association between SBP trajectory and baseline SBP for CKD development. In subgroup with baseline SBP <120 mmHg, HR for incident CKD was higher in increasing SBP trajectory compared to stable SBP trajectory group. On contrary, lower HR for CKD development showed in decreasing trajectory compared to stable SBP trajectory in subgroup with baseline SBP ≥120 mmHg.
Differences in the Rates of Renal Function Decline among the SBP Trajectory Groups
To further corroborate our findings, we compared the slopes of the decline in eGFR among the SBP trajectory groups (Table 4). There was a significant difference in eGFR changes among the three SBP trajectory groups, and the annual decline rate of eGFR for the increasing trajectory (−1.29 ml/min per 1.73 m2) was higher than that for the stable (−1.20 ml/min per 1.73 m2) and decreasing trajectory (−1.18 ml/min per 1.73 m2). As there was an interaction between baseline SBP and the trends of SBP for decline in renal function (P value for interaction =0.04), we separately calculated the rates of eGFR decline of the three SBP trajectory groups according to baseline SBP <120 or ≥120 mm Hg. In participants with baseline SBP <120 mm Hg, kidney function declined faster in the increasing trajectory group than in the stable trajectory group (−1.25 versus −1.15 ml/min per 1.73 m2 per year; P=0.002). In contrast, among participants with baseline SBP ≥120 mm Hg, the rate of eGFR decline was significantly lower in the decreasing SBP trajectory group than in the stable SBP trajectory group (−1.34 versus −1.41 ml/min per 1.73 m2 per year; P<0.001), and the eGFR decline rates did not differ between the stable and increasing trajectory groups.
Annual rate of renal function decline according to SBP trajectory patterns
Secondary Outcome
In the secondary analysis, we studied the relationship between three SBP trajectories and the development of albuminuria. During 33,239 person-years, albuminuria occurred in 1038 participants, with an overall incidence rate of 31.6 per 1000 person-years. Incidence rate of albuminuria was significantly higher in the increasing SBP trajectory than in the other two groups (log-rank test, P<0.001) (Supplemental Table 6). In cause-specific multivariable Cox models with all adjustment levels, increasing SBP trajectory was associated with a 23% higher risk of incident albuminuria as compared with stable trajectory. The similar relationship was also observed in restricted cubic spline analyses for changes in SBP and albuminuria (Supplemental Figure 9).
During follow-up, only 92 death events occurred. There was no difference in incidence rate of death among SBP categories, and SBP trajectories were not associated with risk of death (Supplemental Table 7).
Sensitivity Analyses
To substantiate the robust association of SBP trends with the subsequent outcome, we performed several sensitivity analyses. First, in the analyses with 2- and 4-year SBP trajectories on the basis of SBP readings starting from year 2 (baseline), we found similar association with SBP trajectory with risk of CKD development (Supplemental Table 8). Second, in the SBP trajectory–based time-varying model that took into account time-dependent changes of follow-up SBP readings after year 4, the results consistently showed that an increasing SBP trend was associated with a 1.46-fold higher risk of CKD development (Supplemental Table 9). Third, the conventional time-varying model clearly showed that time-updated SBP had a graded association with the risk of CKD development. Compared with time-updated SBP of 100–119 mm Hg, the HRs for time-updated SBP of <100, 120–129, 130–139, and ≥140 mm Hg were 0.91 (95% CI, 0.66 to 1.26), 1.19 (95% CI, 0.98 to 1.44), 1.33 (95% CI, 1.04 to 1.70), and 1.90 (95% CI, 1.27 to 2.85), respectively. In the same model with time-updated SBP as a continuous measure, a 10-mm Hg increase in SBP was associated with an 11% higher risk of CKD development (Table 5). Finally, we additionally performed analysis using the complete dataset without missing data for SBP within the first 4 years and found similar results (Supplemental Table 10).
HRs for CKD development according to time-updated SBP
Development of De Novo Hypertension during Follow-Up
During follow-up, 777 people were newly diagnosed with hypertension, which was defined as SBP≥140 mm Hg or DBP≥90 mm Hg. We further examined the association between de novo development of hypertension and risk of CKD. In the multivariable-adjusted logistic regression model, people with de novo incident hypertension had a 1.43-fold (95% CI, 1.07 to 1.90) higher risk of CKD than those with normotension. The similar association was observed between incident hypertension and risk of albuminuria development (Supplemental Table 11).
Discussion
In this prospective cohort study, we investigated the association of SBP change over time with CKD development in nonhypertensive Korean adults. Using trajectory modeling, we identified three distinctive patterns of SBP trends and found that an increasing SBP trajectory was associated with a significantly higher risk of new CKD development. In addition, the annual rate of eGFR decline was higher in participants with an increasing trajectory. This relationship was more robust among participants with baseline SBP <120 mm Hg. Similar findings were observed in the analysis of association between occurrence of albuminuria and SBP trends over time. These findings suggest that an increase in BP from the normal range without reaching the hypertension threshold may impair kidney function. Our study supports the existing literature indicating that prehypertension can be a warning sign of vascular organ damage, including kidney disease.
Many studies have attempted to determine the clinical effect of elevated SBP without reaching hypertension level on various outcomes, including kidney disease.13,31 Recently, a long-term observation of the Atherosclerosis Risk in Communities study showed an inverse graded association between baseline hypertension status and kidney function decline during a 30-year follow-up of a general population cohort.32 Notably, the annual eGFR decline was greater in persons with elevated BP without hypertension than in normotensive individuals. This finding is in agreement with two meta-analysis studies showing that prehypertension is an independent risk factor for CKD development. However, these studies examined only baseline SBP as an exposure of interest and did not take into account the dynamic changes of BP over time. In fact, few studies have addressed whether the transition into prehypertension level from a normotensive status can increase the risk of CKD development. Interestingly, the Multi-ethnic Study of Atherosclerosis examined the correlation between SBP changes and the magnitude of eGFR decline and found that increasing SBP over time was associated with a greater risk for accelerated decline in kidney function even in patients without hypertension.33 In addition, in the Coronary Artery Risk Development in Young Adulthood (CARDIA) study, every 10-mm Hg increase in SBP and DBP during young adulthood was significantly associated with subsequent decline in kidney function.34 Similar to the assessment method of the CARDIA study, we designed our analytical approach to examine the temporal association between exposure and outcome. We determined various SBP trajectories using different SBP assessment windows and analyzed the association of the SBP trends with the subsequent risk of CKD. This approach enabled us to robustly reveal that increasing SBP trajectory was associated with a significantly higher risk of CKD development and kidney function decline.
It should be noted that participants in the increasing trajectory group had lower SBP levels at baseline than those in the other two groups. In addition, their SBP increased from 105 to 124 mm Hg, reaching a plateau at this level, and 38% had SBP≥130 mm Hg within 4 years of follow-up. This pattern persisted throughout the observation period. Thus, it can be interpreted that an increase in BP by ≥20 mm Hg from a normotensive status may significantly increase the glomerular pressure and cause vascular damage, ultimately resulting in meaningful kidney impairment, although the elevated BP level did not reach the hypertension threshold. In addition, our findings can also highlight the importance of elevated SBP before hypertension because CKD events more frequently occurred in participants with increasing SBP trajectory and long-term exposure to elevated BP ≥120 mm Hg. We further substantiated this finding by showing that a time-updated SBP of ≥120 mm Hg was also associated with a higher risk of CKD. Therefore, our data add more evidence to the accumulating literature data on the harmful effects of elevated BP level without reaching hypertension on kidney function decline.
In this observational study, we cannot draw a confirmative conclusion about the BP threshold for CKD prevention, particularly among individuals with elevated SBP between 120 and 140 mm Hg. As mentioned, recent RCTs showed that intensive BP control at <120 mm Hg resulted in more adverse kidney outcomes. However, the findings of RCTs cannot be extrapolated to all situations because RCTs have strictly specified enrollment criteria. In fact, the baseline characteristics of our study participants were very different from those in previous RCTs, which included individuals with a high risk of cardiovascular disease and higher BP levels. However, it should be noted that the risk of CKD started in persons with SBP≥120 mm Hg and that a decreasing SBP trajectory was associated with a lower risk of CKD among those with SBP≥120 mm Hg in subgroup analysis. Furthermore, although several RCTs showed that BP reduction <120 mm Hg among hypertensive people resulted in more CKD events, such lowering BP significantly decreased albuminuria.14,15 In line with these findings, we showed that increasing SBP trajectory was associated with higher risk of new development of albuminuria. As discussed above, it is possible that a rise in SBP in this trajectory group might contribute to damage to glomerular barrier, resulting in albumin leakage. Thus, our results support the new definition of hypertension proposed by the American Heart Association,35 in which the threshold for the treatment of hypertension was lowered to 130 mm Hg.
This study had limitations. First, as with all observational studies, we cannot exclude the possibility of residual bias due to potential unmeasured confounders. In addition, our observational analysis cannot prove a causality between BP and CKD development. In fact, there has been a debate about the bidirectional relationship between these factors. It is possible that BP increases after kidney function loss. To mitigate this effect, we determined the SBP trajectories during the early period of follow-up and analyzed the temporal association of SBP changes with subsequent development of CKD at a later time point. In addition, we excluded participants with kidney function decline during this period, and all participants were normotensive at baseline. Finally, although we excluded antihypertensive drug users at baseline in the main analysis, BP-lowering drugs were given to people who were newly diagnosed with hypertension during follow-up. To minimize the potential effect of these drugs on outcomes, we treated the use of antihypertensive drugs as a time-varying exposure in the time-updated model. Nevertheless, such a bidirectional relationship was likely present in this observational study. Second, although BP increased to >120 mm Hg in participants with increasing SBP trajectory, it is possible that there are other factors affecting kidney function. In other words, BP could be elevated when individuals become more obese and more metabolically unhealthy. Accumulating evidence have suggested that obesity and insulin resistance are risk factors of the development and progression of CKD.36,37 Thus, the significant association of increasing SBP trajectory with the risk of CKD may not be solely explained by BP-induced effects. However, BMI, lipid levels, and insulin resistance did not differ among the three groups, suggesting that these factors contributed little to the association of SBP with the CKD risk. Nevertheless, we rigorously adjusted these factors in multivariable cause-specific Cox models and found that increasing SBP trajectory was associated with the CKD risk independently of BMI and insulin resistance. Furthermore, in a time-updated model to account for the time-dependent effects of these factors, the independent association persisted. Third, BP was measured in a standard manner by trained nurses; however, clinic BP cannot represent the overall BP status and is unlikely to detect diverse patterns, such as white coat hypertension, variability of BP, and reverse dipping pattern.38,39 Fourth, we cannot exclude possibility that elevated BP itself over the follow-up period, not SBP changes during the initial period, contributes to higher risk of CKD because SBP rose during the first 2 years and then was persistently elevated thereafter. Nevertheless, we tested this using various methods and found consistent results. Finally, as our study included middle-aged Korean adults with a low cardiovascular risk, our findings may not be generalizable to other ethnic groups.
In conclusion, this study showed that increasing SBP over time reaching >120 mm Hg from normotension is significantly associated with CKD development. Thus, close and serial monitoring of BP and concomitant assessment of kidney function, particularly in individuals with elevated BP, may be helpful for the early detection and prevention of CKD.
Disclosures
All authors have nothing to disclose.
Funding
None.
Acknowledgments
The authors thank the staff and participants of the Korean Genome and Epidemiology Study (KoGES) for important contributions.
The epidemiologic data used in this study were obtained from KoGES (4851-302) of the National Research Institute of Health, Centers for Disease Control and Prevention, Ministry for Health and Welfare, Republic of Korea.
The funding sources for KoGES had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. Parts of this study were presented at the 39th Annual Meeting of the Korean Society of Nephrology, which was held in Seoul, Korea on May 23–26, 2019.
Supplemental Material
This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2020010084/-/DCSupplemental.
Supplemental Figure 1. Flow chart of analysis.
Supplemental Figure 2. Examples of systolic BP trajectory models.
Supplemental Figure 3. Study design and approach for sensitivity analyses.
Supplemental Figure 4. Observed BP trends according to systolic BP trajectory throughout the follow-up period.
Supplemental Figure 5. Proportions of systolic BP categories according to three trajectory groups during the follow-up period.
Supplemental Figure 6. Cumulative incidence curve for CKD development according to systolic BP trajectory groups.
Supplemental Figure 7. Restricted cubic spline curve for incident CKD.
Supplemental Figure 8. Changes in diastolic BP according to three diastolic BP trajectories.
Supplemental Figure 9. Restricted cubic spline curve for risk of albuminuria development.
Supplemental Table 1. Estimation process for the trajectory groups of systolic BP.
Supplemental Table 2. The frequency of measurements for BP, serum creatinine, and urinalysis during the study period.
Supplemental Table 3. The number of imputed missing data.
Supplemental Table 4. Association between SBP trajectories with 2-, 4-, and 6-year assessment window periods and risk of CKD development during the next 4 years.
Supplemental Table 5. Incidence rates and hazard ratios for CKD development according to diastolic BP trajectory patterns.
Supplemental Table 6. Incidence rates and hazard ratios of albuminuria development according to systolic BP trajectory patterns.
Supplemental Table 7. Mortality rates and hazard ratios for all-cause death according to systolic BP trajectory patterns.
Supplemental Table 8. Hazard ratios for risk of CKD according to 2- and 4-year systolic BP trajectory starting from year 2.
Supplemental Table 9. Time-varying model for risk of CKD according to systolic BP trajectory patterns.
Supplemental Table 10. Hazard ratios for risk of CKD according to systolic BP trajectory with complete BP dataset.
Supplemental Table 11. Odd ratios for the development of CKD and albuminuria in people with de novo development of hypertension during follow-up.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2020 by the American Society of Nephrology