Association of Blood Pressure Trajectories in Early Life with Subclinical Renal Damage in Middle Age

Abstract

Background

Although high BP is one of the most important factors affecting renal function, whether longitudinal BP trajectories in early life course are associated with renal function damage in later life is unclear.

Methods

To investigate the correlation between BP trajectories from childhood to adulthood and renal function in middle age, we used group-based trajectory models to identify BP trajectories in 2430 individuals (aged 6–15 years old at baseline) participating in the ongoing Hanzhong Adolescent Hypertension Cohort. We tested the association between these trajectories and subclinical renal damage in middle age, adjusting for several covariates.

Results

We identified four distinct systolic BP trajectories among 2430 subjects: low stable, moderate stable, high stable, and moderate increasing on the basis of systolic BP levels at baseline and during the 30-year follow-up period. The urinary albumin-to-creatinine ratio (uACR) was higher in moderate stable, high stable, and moderate increasing groups compared with the low stable group. A total of 228 individuals had subclinical renal disease by 2017. Compared with the low stable trajectory group, the other groups had increasingly greater odds of experiencing subclinical renal disease in middle age. These associations were not altered after adjustment for other covariates, except for in the moderate stable group. Analyzed results were similar for the mean arterial pressure and diastolic BP trajectory groups.

Conclusions

Higher BP trajectories were correlated with higher of uACR levels and risk of subclinical renal disease in middle age. Identifying long-term BP trajectories from early age may assist in predicting individuals’ renal function in later life.

CKD is now one of the most significant global health problems. The population prevalence of CKD is >10%, and over 50% of the population are at high risk for this disease.¹ Other studies reported that the prevalence of CKD among adults was 10.8% in China² and 13.0% in the United States.³ The 2016 European Guidelines on cardiovascular disease prevention reported that high-risk and very high–risk cardiovascular disease categories correspond to subjects with moderate CKD (30≤eGFR<59 ml/min per 1.73 m²) and severe CKD (eGFR<30 ml/min per 1.73 m²), respectively.⁴ In addition, subjects with the highest risk benefit most from preventive efforts. Therefore, early renal damage detection or screening and intervention are of utmost importance.

Hypertension is one of the most important CKD risk factors.^5,6 Elevated BP may lead to peripheral arterial disease, which in turn, contributes to the development of CKD. The relationship between hypertension and progression of CKD is direct and progressive.⁷ In addition, antihypertensive therapy has an additional protective effect on renal function.⁸

Population-based cohorts support the view that high BP in later life originates in childhood.^9–11 However, available evidence on the association between BP and CKD mainly concerns adult or elderly subjects with or without hypertension.¹² Few take into account the potential effect of BP levels experienced earlier in life or changes in BP levels over time. Long-term BP trajectory patterns from childhood to middle age and their effect on CKD risk remain unknown. Consequently, the aims of this study are to identify subgroups of population with similar trajectories in BP development from childhood to middle age and reveal the relationship of BP trajectories with urinary albumin-to-creatinine ratio (uACR) and the presence of subclinical renal damage (SRD).

Methods

Study Population

This study was conducted within the Hanzhong Adolescent Hypertension Cohort, an ongoing prospective study focusing on evaluating the development of cardiovascular risk factors originating from children and young adults. The study began in 1987 when 4623 schoolchildren were enrolled from 26 rural sites of three towns (Qili, Laojun, and Shayan) in Hanzhong, Shaanxi, China.¹³ During baseline evaluation, the inclusion criteria were as follows: aged 6–15 years old in 1987, no chronic disease in the child’s medical history, and can communicate normally in Mandarin. Later information collection was conducted in 1989, 1992, 1995, 2005, 2013, and 2017, resulting in a maximum follow-up time of 30 years. Among those follow-up activities, we randomly selected several subjects to visit in 2005 and obtained BP and other data from 436 individuals. Except for the visit in 2005, other follow-ups were large in scale and aimed to visit each individual who was enrolled in 1987. The response rate was 77.7% (n=3592) in 1989, 84.8% (n=3918) in 1992, 82.1% (n=3794) in 1995, 65.3% (n=3018) in 2013, and 60.1% (n=2780) in 2017. Reasons for loss of follow-up mainly included death, mental illness, military service, and migration.

The study was approved by the Academic Committee of the First Affiliated Hospital of Xi’an Jiaotong University (XJTU1AF2015LSL-047) and clinically registered (NCT02734472). All participants in this study signed informed consent for each visit, and for those <18 years of age at baseline, consent of a parent/guardian was obtained. These analyses included individuals with BP values available at three or more examinations and excluded subjects with three or more BP measurements only from the initial four visits (1987, 1989, 1992, and 1995) or the last three follow-ups (2005, 2013, and 2017).

Anthropometric Measurements

Personal information, including demographic characteristics, personal/family medical history (hypertension, diabetes, hyperlipidemia, and stroke), cigarette smoking, and drinking history, was obtained through a questionnaire. Height, body weight, hip/waist circumference, and bust size were measured with underwear through appropriate scales by trained staff. Replicate measurements of these indices were made, and the mean values were used for analysis. Body mass index (BMI) was calculated as weight in kilograms divided by the square of the height (kilograms per meter squared); waist-to-hip ratio (WHR) was calculated as waist circumference divided by the hip circumference, both in centimeters.

BP Measurements

Participants were required to avoid coffee/tea, alcohol, cigarette smoking, and strenuous exercise for at least 30 minutes before BP measurement. BP was measured three times in a seated position on the right upper arm after a 5-minute rest, with a 2-minute interval between measurements. The average of three BP values was used in the analyses. A Hawksley random zero sphygmomanometer was used¹⁴ for the first six visits, and an Omron M6 (Omron, Kyoto, Japan) device was used¹⁵ in 2017 for BP measurements by trained and certified staff. An appropriate cuff size was used, and the procedure was performed in a quiet and comfortable environment. Mean arterial pressure (MAP) was calculated by the formula: 1/3 systolic BP (SBP) +2/3 diastolic BP (DBP; millimeters of Hg).

Biochemical Parameter Measurements

Fasting venous blood samples were obtained by experienced nurses in the morning after fasting for 8–10 hours. Blood samples were immediately centrifuged at 3000×g for 10 minutes and stored at −80°C until analysis. Urine samples were collected for the first time in the morning. The samples were kept frozen at −40°C until analysis. Certain biomarkers, including total bilirubin, total cholesterol (TC), triglyceride (TG), LDL cholesterol (LDL-C), HDL cholesterol (HDL-C), serum creatinine, fasting plasma blood glucose (GLU), urinary uric acid (UA), creatinine, and albumin levels, were evaluated with an automatic biochemical analyzer (model 7600; Hitachi Ltd., Tokyo, Japan). Serum creatinine was used to evaluate eGFR, and urine microalbumin and creatinine were used to evaluate uACR.

Assessment of Renal Function

Renal function was assessed with eGFR and uACR. eGFR was calculated using the formula adapted from the Modification of Diet in Renal Disease equation on the basis of data from Chinese subjects with CKD.¹⁶ The specific formula is as follows: eGFR=175× serum creatinine^−1.234 × age^−0.179 (×0.79 for girls/women), where serum creatinine concentration is in milligrams per deciliter and age is in years.² uACR was calculated as urine albumin in milligrams divided by the urine creatinine in millimole (milligrams per millimole). Presence of SRD was defined as eGFR between 30 and 60 ml/min per 1.73 m²¹⁷ or elevated uACR of at least 2.5 mg/mmol in men and 3.5 mg/mmol in women as previously described.¹⁸

Definitions

Participants who reported continuous or cumulative smoking for 6 months or more during their lifetime were defined as cigarette smokers.¹³ Alcohol consumption was defined as subjects who claimed that they consumed alcohol (liquor, beer, or wine) every day and that consumption lasted for >6 months.¹⁹ The bust size is defined as the chest size of the annular double-nipple level.²⁰ Hypertension was defined as an average measured SBP of at least 140 mm Hg and/or DBP at least 90 mm Hg or currently receiving antihypertension treatment. Diabetes was defined as GLU≥7.0 mmol/L or a physician diagnosis of diabetes that was self-reported by the individuals.²¹ Hyperlipidemia was defined as the occurrence of any one of the following four situations: hypertriglyceridemia (TG≥2.26 mmol/L), hypercholesterolemia (TC≥6.22 mmol/L), high levels of LDL-C (≥4.14 mmol/L), or low levels of HDL-C (<1.04 mmol/L).²²

Statistical Analyses

During the 30-year follow-up period, SBP, DBP, and MAP trajectories were identified by using latent mixture modeling within the SAS Proc Traj.^23,24 BP trajectories were modeled among 2430 subjects with three or more BP examinations, excluding those with only the initial four visits data (1987, 1989, 1992, and 1995) or only BP measurements for the last three visits (2005, 2013, and 2017). The “latent mixture modeling” estimated the model parameters through the maximum likelihood method and assigned each individual to a corresponding group with the greatest posteriori probability. We first constructed trajectory shape with a cubic specification and then, quadratic or linear if necessary.²⁵ The best-fitting trajectory models were evaluated using the Bayesian information criterion, and the appropriate average posterior probability (>0.7) with a minimum sample size in each group accounted for >5.0%.

Continuous data were reported as means±SDs if normally distributed; otherwise, they were shown as medians (25th, 75th percentile ranges). Categorical data were presented as frequency and percentage. Differences between continuous variables were analyzed by t test for two-group comparisons and one-way ANOVA for three or more groups when the distribution and variance met the conditions; otherwise, Mann–Whitney U test and Kruskal–Wallis test were used. Categorical variables were analyzed by chi-squared tests. Correlation analysis was determined with the Pearson correlation coefficient or the Spearman correlation coefficient when appropriate.

Logistic regression analysis was used to determine the association between SRD in 2017 and different BP trajectory groups. Strengths of associations were determined by estimating the odds ratios (ORs) and their 95% confidence intervals (95% CIs). To detect changes in associations between outcome and main exposures, several models were fitted on unadjusted analysis as follows: model 1 = sex, race, BMI, heart rate (HR), and WHR in 2017; model 2 = model 1 + hypertension, diabetes, hyperlipidemia, smoking, and drinking in 2017; and model 3 = model 2 + GLU (millimoles per liter), serum UA (micromoles per liter), TC (millimoles per liter), TG (millimoles per liter), LDL-C (millimoles per liter), and HDL-C (millimoles per liter) in 2017. Furthermore, because of the fact that use of several medications may affect urinary protein and renal function, a sensitivity analysis was conducted by excluding individuals who had diabetes or received antihypertensives therapy (n=101).

All statistical tests were two sided, and statistical significance was considered at P<0.05. Statistical analyses were performed using SPSS software (SPSS Inc., Chicago, IL).

Results

The Hanzhong Adolescent Hypertension Cohort enrolled 4623 schoolchildren in 1987. To fit BP trajectories, we included subjects with three or more BP measurements, except for those with data only derived from the initial four visits or the last three visits during the follow-up period. As a result, a total of 2430 individuals were included in the BP trajectory analysis. In this population, 1732 individuals with eGFR values and 1647 participants with uACR data available in 2017 were included in verifying the association between BP trajectories and eGFR, uACR, or SRD.

Characteristics of the Trajectory Groups

Four separate trajectories in SBP from childhood to middle age were identified (Figure 1). We refer to these trajectories as “low stable,” “moderate stable,” “high stable,” and “moderate increasing” groups. The low stable group (n=724; 29.8%) was characterized by maintaining relatively lower SBP levels. The moderate stable group (n=1283; 52.8%) was characterized by sharing moderate SBP levels. The high stable group (n=296; 12.2%) was characterized by enduring relatively higher SBP levels. The moderate increasing group (n=127; 5.2%) was characterized by experiencing a rapid increase in SBP, which started at moderate level and overtook the high stable group at about 30 years old (Figure 1, Table 1). Similar to SBP trajectories, four isolated trajectory groups were identified in MAP (Figure 2A). The numbers and proportions of participants were 905 (37.2%), 1155 (47.5%), 214 (8.8%), and 156 (6.4%) in “low stable,” “moderate stable,” “high stable,” and “moderate increasing” groups of MAP trajectories, respectively (Figure 2A, Supplemental Table 1).

Figure 1.

Trajectory modeling identified four distinct systolic BP trajectory groups from childhood to middle age in the Hanzhong Adolescent Hypertension Cohort.

Table 1.

Average systolic and diastolic BP (millimeters of Hg) by age periods in systolic BP trajectory groups

Trajectory Groups	SBP, mm Hg				DBP, mm Hg
	Low Stable	Moderate Stable	High Stable	Moderate Increasing	Low Stable	Moderate Stable	High Stable	Moderate Increasing
N (%)	724 (29.8)	1283 (52.8)	296 (12.2)	127 (5.2)	724 (29.8)	1283 (52.8)	296 (12.2)	127 (5.2)
6–10, yr	96.5±8.0	101.1±8.3	105.4±8.5	107.7±10.6	61.2±7.8	63.9±8.4	66.9±8.0	70.6±8.2
11–15, yr	98.6±8.7	104.9±9.5	112.2±10.2	106.9±9.0	62.4±8.1	65.6±8.4	69.7±8.9	67.8±9.0
16–20, yr	107.5±9.9	117.1±10.2	127.1±11.8	117.9±11.2	67.1±7.8	71.7±8.9	76.3±9.2	73.8±8.6
21–25, yr	110.2±10.9	119.7±9.8	132.3±11.8	124.6±12.7	70.5±8.3	74.2±8.2	79.4±10.0	77.4±7.8
26–30, yr	109.4±7.2	118.8±8.2	133.3±7.7	135.5±10.6	72.0±6.7	74.6±8.8	82.3±10.0	87.5±7.7
31–35, yr	105.8±9.5	121.0±9.6	133.1±11.1	147.2±10.0	71.2±8.1	80.3±9.1	88.6±11.4	100.8±10.3
36–40, yr	109.0±9.8	121.5±10.3	135.6±11.6	157.3±19.1	71.4±8.2	79.0±9.6	88.7±10.1	103.2±12.9
41–45, yr	111.1±10.4	123.4±11.5	135.7±11.2	157.2±19.6	70.6±7.9	78.4±9.7	87.6±10.2	100.6±12.5

SBP, systolic BP; DBP, diastolic BP.

Figure 2.

Trajectory modeling identified four discrete (A) mean arterial pressure and (B) diastolic BP trajectory groups from childhood to middle age in the Hanzhong Adolescent Hypertension Cohort.

In addition, four discrete trajectory groups were identified in DBP (Figure 2B). The “low stable” group (1165; 47.9%) holds a low DBP level during the 30-year follow-up period. The “moderate stable” group (628; 25.8%) kept moderate DBP levels. The “low increasing” group (465; 19.1%) started at the lowest DBP levels and shared a rapid increase in DBP, which surpassed low stable and moderate stable groups subsequently. The “moderate increasing” group (172; 7.1%) originated in moderate DBP levels and experienced a rapid increase in DBP at about 20 years old (Supplemental Table 2).

The BP trajectory curve was not smooth, and it fluctuated over time. Generally, the SBP, MAP, and DBP trajectories showed larger BP increase during the children and adolescents period (6–19 years old) than after adulthood. Specially, the BP of the low increasing and moderate increasing groups increased with age, even after adulthood (Figures 1 and 2). Each individual was not always assigned to the same level group identified by SBP, DBP, or MAP trajectories. For example, individuals assigned to the low stable group in DBP or MAP trajectory had partial overlap with the low stable group in SBP trajectory; however, they were not identical, because individuals belonging to the low stable group in the SBP trajectory could also be assigned to low increasing (n=53) or moderate stable (n=34) groups in DBP trajectories (Supplemental Figures 1 and 2). Rarely, 14 subjects in the high stable group by SBP trajectories were assigned to the low stable group by DBP trajectories (Supplemental Figure 1).

Cardiovascular Risk Factors by BP Trajectory Groups

We analyzed partial anthropometry (age and bust in 1987, WHR in 2017, and HR and BMI in both 1987 and 2017) and biochemical indicator tests in 2017 in each group. The median age of the 2430 analyzed subjects was 12 (9, 14) years old at baseline. Among them, 1378 participants (56.7%) were boys, and 1052 (43.3%) were girls. Differences in the proportion of boys, highest level of education, smoking, drinking, and anthropometric measurements (including age, bust at baseline, WHR and HR in 2017, and BMI in both 1987 and 2017) were statistically significant (P<0.05) (Table 2). Biochemical parameters (GLU, TC, TG, LDL-C, HDL-C, serum creatinine, serum UA, urine UA, and urine albumin) tested in 2017 were also significantly different (P<0.05) (Table 2) among the four SBP trajectory groups. In addition, the incidence of hypertension and hyperlipidemia in 2017 increased with the trajectories of SBP (P<0.05) (Table 2).

Table 2.

Demographic characteristics and cardiovascular risk factors by the systolic BP trajectory group

Parameter	Total, N	Low Stable	Moderate Stable	High Stable	Moderate Increasing	P Value
Boys (%)	1378	283 (39.09)	785 (61.18)	227 (76.69)	83 (65.35)	<0.001
Age in 1987, yr	2430	12 (9, 14)	12 (9, 14)	12 (10, 14)	12 (10, 14)	0.03
Bust in 1987, cm	2421	60.0 (55.5, 66.0)	61.0 (57.0, 67.0)	63.0 (59.0, 70.0)	64.0 (60.0, 70.0)	<0.001
HR in 1987, times per min	2425	78 (72, 84)	78 (72, 84)	80 (72, 84)	78 (72, 84)	0.09
HR in 2017, times per min	1907	73 (66, 80)	73 (67, 79)	75 (67, 82)	77 (68, 85)	<0.01
BMI in 1987, kg/m2	2419	15.6 (14.4, 17.2)	15.9 (14.7, 17.4)	16.1 (15.0, 17.8)	16.3 (15.1, 18.1)	<0.001
BMI in 2017, kg/m2	1904	22.9 (21.1, 24.6)	23.8 (22.0, 26.0)	25.1 (23.2, 26.9)	26.6 (23.9, 29.2)	<0.001
WHR in 2017, m	1901
Men	1097	0.925±0.057	0.940±0.063	0.958±0.053	0.987±0.061	<0.001
Women	804	0.879±0.060	0.886±0.066	0.889±0.051	0.924±0.062	0.003
Occupation (%)	2170					0.20
Farmer	871	256 (39.81)	457 (39.91)	109 (41.60)	49 (40.83)
Worker	424	105 (16.33)	227 (19.83)	61 (23.28)	31 (25.83)
Businessman	173	56 (8.71)	91 (7.95)	19 (7.25)	7 (5.83)
Governor	71	19 (2.95)	37 (3.23)	10 (3.82)	5 (4.17)
Other	631	207 (32.19)	333 (29.08)	63 (24.05)	28 (23.33)
Education (%)	2206					0.01
Primary school or less	179	51 (7.74)	96 (8.25)	27 (10.07)	5 (4.31)
Middle school	1328	375 (56.90)	719 (61.82)	150 (55.97)	84 (72.41)
High school	504	159 (24.13)	251 (21.58)	71 (26.49)	23 (19.83)
College or more	195	74 (11.23)	97 (8.34)	20 (7.46)	4 (3.45)
Marital status (%)	2255					0.20
Married	2162	648 (96.57)	1131 (95.12)	267 (97.45)	116 (95.87)
Divorced	46	15 (2.24)	26 (2.19)	4 (1.46)	1 (0.83)
Unmarried or other	47	8 (1.19)	32 (2.69)	3 (1.09)	4 (3.31)
Smoking (%)	1908	185 (32.23)	470 (47.33)	148 (60.66)	57 (58.16)	<0.001
Drinking (%)	1908	118 (20.59)	326 (32.83)	97 (39.75)	34 (34.69)	<0.001
Hypertension (%)	1908	9 (1.57)	63 (6.34)	78 (31.97)	64 (65.31)	<0.001
Diabetes (%)	1908	9 (1.57)	34 (3.42)	4 (1.64)	7 (7.22)	<0.01
Hyperlipidemia (%)	1908	26 (4.54)	95 (9.57)	29 (11.89)	19 (19.39)	<0.001
GLU, mmol/L	1730	4.52 (4.25, 4.80)	4.55 (4.27, 4.93)	4.69 (4.32, 5.02)	4.74 (4.39, 5.19)	<0.001
TC, mmol/L	1732	4.43 (3.98, 4.90)	4.49 (4.04, 5.01)	4.69 (4.14, 5.10)	4.52 (4.08, 5.17)	0.02
Triglycerides, mmol/L	1732	1.22 (0.89, 1.77)	1.35 (0.97, 1.94)	1.50 (1.01, 2.04)	1.69 (1.17, 2.52)	<0.001
LDL-C, mmol/L	1732	2.44 (2.07, 2.79)	2.51 (2.15, 2.91)	2.60 (2.18, 3.05)	2.55 (2.18, 3.00)	<0.001
HDL-C, mmol/L	1732	1.19 (1.03, 1.38)	1.14 (0.97, 1.32)	1.11 (0.98, 1.27)	1.04 (0.90, 1.19)	<0.001
Serum uric acid, μmol/L	1732	254.6 (211.5, 312.5)	283.1 (229.6, 334.4)	310.7 (263.6, 358.2)	329.6 (262.4, 379.8)	<0.001
Urine uric acid, μmol/L	1671	1251 (876, 1855)	1315 (959, 2083)	1403 (991, 2243)	1522 (948, 2206)	<0.01
Serum creatinine, μmol/L	1669	71.60 (63.60, 83.08)	76.70 (68.00, 86.80)	78.85 (71.53, 89.73)	77.45 (68.70, 88.10)	<0.001
Urine albumin, mg/L	1671	6.20 (3.20, 11.20)	7.70 (4.20, 13.73)	10.35 (5.40, 21.58)	14.40 (7.00, 45.50)	<0.001
eGFR, ml/min per 1.73 m2	1732	98.47 (88.39, 112.26)	97.09 (87.80, 110.33)	96.21 (85.19, 109.20)	95.35 (85.62, 111.56)	0.34
uACR, mg/mmol	1669	0.878 (0.593, 1.439)	0.932 (0.619, 1.631)	1.182 (0.738, 2.367)	2.574 (1.020, 6.392)	<0.001

Continuous variables were shown as mean±SD if normally distributed or median (quartile 1, quartile 3) if non-normally distributed. Categorical variables were expressed as numbers and percentages of subjects. Statistical ANOVA was performed by one-way ANOVA when normally distributed; otherwise, the Kruskal–Wallis test was used. Differences between groups of categorical variables were compared with chi-squared tests. HR, heart rate; BMI, body mass index; WHR, waist-to-hip ratio; GLU, fasting plasma blood glucose; TC, total cholesterol; LDL-C, LDL cholesterol; HDL-C, HDL cholesterol; uACR, urinary albumin-to-creatinine ratio.

Individuals in the low stable group were more likely to be girls; have lower BMI, WHR, GLU, TC, TG, LDL-C, serum UA, and serum creatinine; and have higher HDL-C than those in other groups. Boys with high BMI or WHR were more likely to suffer from high SBP throughout or rapid increase in SBP during the follow-up. In addition, high-risk SBP trajectory groups were associated with high rates of smoking and drinking (Table 2). Findings were similar for the isolated DBP (Supplemental Table 3) and MAP trajectories (Supplemental Table 4). Interestingly, values of BMI and WHR in 2017, the percentage of hypertension, GLU, TC, serum UA, and urine albumin increased with the trajectories of SBP, DBP, and MAP, whereas eGFR values decreased (Supplemental Tables 3 and 4, Table 2). However, no significant differences existed in the percentages of employment type and marital status among SBP, DBP, and MAP trajectories (Supplemental Tables 3 and 4, Table 2).

We further classified BMI, SBP, and DBP according to age periods and analyzed the relationship between BMI and BP. BMI had group differences in prepuberty, puberty, and middle-aged adulthood by SBP trajectories (Supplemental Table 5) (P<0.05). SBP and DBP showed significant group differences from prepuberty to middle-aged adulthood (Supplemental Table 5) (P<0.05). Results were similar in DBP and MAP trajectories (Supplemental Tables 6 and 7). Additional analyses showed that the SBP was positively correlated with BMI in prepuberty (r=0.239; P<0.01), puberty (r=0.350; P<0.01), young adulthood (r=0.226; P<0.01) and middle-aged adulthood (r=0.336; P<0.01). In addition, DBP and MAP were also significantly positively correlated with BMI (Supplemental Table 8).

Association between BP Trajectories and eGFR or uACR

As shown in Supplemental Table 9, eGFR≥90 ml/min per 1.73 m² accounts for the majority of the proportions (>60%) in each SBP trajectory followed by 60≤eGFR<90 ml/min per 1.73 m² (>25%). Because subjects included in this study were relatively young (the oldest subjects were 45 years old in 2017), individuals with eGFR<60 ml/min per 1.73 m² were limited. The median of eGFR was higher in the low stable group in comparison with other trajectories of SBP, DBP, and MAP (Supplemental Tables 3 and 4, Table 2).We used eGFR values as continuous variables to analyze the association between discrete BP trajectories and eGFR level. Compared with the low stable group, eGFR values were significantly lower in the moderate stable group of MAP trajectories (P<0.01) (Figure 3). However, there were no differences between eGFR and other SBP, DBP, or MAP trajectories (Figure 3).

Figure 3.

No differences in eGFR among groups in (A) systolic BP, (B) diastolic BP, and (C) mean arterial pressure trajectories except for the moderate stable group of MAP trajectories in comparison with the low stable group (C).

Then, we analyzed the relationship between uACR levels and isolated BP trajectories. Interestingly, the uACR levels were significantly higher in high stable (1.182 [0.738, 2.367] mg/mmol) and moderate increasing (2.574 [1.020, 6.392] mg/mmol) groups compared with the low stable group (0.878 [0.593, 1.439] mg/mmol) in SBP trajectories (P<0.05) (Figure 4A). In the DBP trajectory groups, the uACR levels were significantly higher in moderate stable (0.990 [0.621, 1.653] mg/mmol), low increasing (1.134 [0.696, 2.134] mg/mmol), and moderate increasing (2.233 [0.930, 4.802] mg/mmol) groups in comparison with the low stable group (0.872 [0.583, 1.463] mg/mmol; P<0.05) (Figure 4B). In the MAP trajectories, compared with the low stable group (0.868 [0.583, 1.398] mg/mmol), the uACR levels were significantly higher in moderate stable (0.990 [0.635, 1.670] mg/mmol), high stable (1.195 [0.716, 2.572] mg/mmol), and moderate increasing (2.323 [0.988, 4.197] mg/mmol) groups (P<0.05) (Figure 4C). Moreover, the uACR levels between moderate increasing, high stable (low increasing), and moderate stable groups in SBP, MAP, and DBP trajectories were also significantly different, respectively (P<0.05) (Figure 4).

Figure 4.

The urinary albumin-to-creatinine ratio (uACR) values shows significant differences among groups in (A) systolic BP, (B) diastolic BP, and (C) mean arterial pressure trajectories. ^#P<0.05 compared with the low stable group; ^&P<0.05 compared with the moderate stable group; ^$P<0.05 compared with the high stable group.

Risk of Subclinical Renal Damage for the Discrete Trajectories

We evaluated renal function by eGFR and uACR jointly. Among the 2430 subjects who were included to fit the BP trajectories, 1732 individuals had eGFR values and 1647 had uACR data in 2017. As a result, a total of 1738 participates obtained either eGFR or uACR data. Among them, 228 individuals developed SRD by the most recent follow-up in April 2017 (Supplemental Table 10). The prevalence of SRD varied from 7.85% in the low stable group up to 43.62% in the moderate increasing group in the SBP trajectories (Table 3). To examine the association between BP trajectory groups and SRD, the groups were included as independent variables in a logistic regression model estimating the predictors of the presence of SRD in the most recent visit, and the low stable group was considered as the control group. Compared with the low stable group, other trajectory groups had increasingly greater odds of existing SRD in 2017. In unadjusted analyses, ORs (95% CIs) were 1.45 (95% CI, 0.99 to 2.12) for moderate stable, 3.25 (95% CI, 2.07 to 5.10) for high stable, and 9.08 (95% CI, 5.42 to 15.20) for moderate increasing in comparison with low stable. Adjustment for subjects’ demographic characteristics (sex and race) and anthropometric parameters in 2017 (BMI, HR, and WHR) significantly attenuated the ORs (Table 3, model 1). Additional adjustment for cardiovascular risk factors in 2017 (hypertension, diabetes, hyperlipidemia, smoking, and drinking) led to slightly attenuated ORs for the high-risk groups (Table 3, model 2). In the final model, additional adjustment was performed with individuals’ biochemical indicators in 2017 (GLU [millimoles per liter], serum UA [micromoles per liter], TC [millimoles per liter], TG [millimoles per liter], LDL-C [millimoles per liter], and HDL-C [millimoles per liter]) (Table 4, model 3), resulting in additional attenuations of the ORs for the high-risk groups.

Table 3.

Adjusted odds ratios and 95% confidence intervals of the association of BP trajectory groups with subclinical kidney damage

Trajectory Groups	No. of Subjects with SRD in 2017 (%)	Unadjusted	Model 1	Model 2	Model 3
SBP trajectory group
Low stable	42 (7.85)	1.00	1.00	1.00	1.00
Moderate stable	98 (10.99)	1.45 (0.99 to 2.12)	1.19 (0.79 to 1.78)	1.16 (0.77 to 1.74)	1.10 (0.73 to 1.67)
High stable	47 (21.66)	3.25 (2.07 to 5.10)	2.12 (1.29 to 3.49)	1.76 (1.03 to 2.99)	1.63 (0.95 to 2.81)
Moderate increasing	41 (43.62)	9.08 (5.42 to 15.20)	5.03 (2.86 to 8.87)	3.21 (1.70 to 6.08)	3.04 (1.58 to 5.85)
DBP trajectory group
Low stable	67 (7.94)	1.00	1.00	1.00	1.00
Moderate stable	52 (11.71)	1.54 (1.05 to 2.25)	1.30 (0.86 to 1.96)	1.23 (0.81 to 1.87)	1.16 (0.76 to 1.78)
High stable	58 (17.63)	2.48 (1.70 to 3.62)	1.82 (1.19 to 2.78)	1.74 (1.13 to 2.68)	1.63 (1.04 to 2.53)
Moderate increasing	51 (42.15)	8.45 (5.45 to 13.10)	5.27 (3.22 to 8.63)	3.43 (1.93 to 6.12)	3.38 (1.87 to 6.12)
MAP trajectory group
Low stable	47 (6.93)	1.00	1.00	1.00	1.00
Moderate stable	98 (12.41)	1.90 (1.32 to 2.74)	1.47 (0.98 to 2.19)	1.44 (0.96 to 2.16)	1.33 (0.88 to 2.01)
High stable	37 (23.42)	4.11 (2.56 to 6.59)	2.73 (1.62 to 4.60)	2.37 (1.36 to 4.11)	2.29 (1.30 to 4.03)
Moderate increasing	46 (41.07)	9.36 (5.80 to 15.11)	5.42 (3.17 to 9.28)	3.67 (1.96 to 6.87)	3.37 (1.76 to 6.43)

Model 1 = sex, race, body mass index, heart rate, and waist-to-hip ratio in 2017. Model 2 = model 1 + hypertension, diabetes, hyperlipidemia, smoking, and drinking in 2017. Model 3 = model 2 + fasting blood glucose (millimoles per liter), serum uric acid (micromoles per liter), total cholesterol (millimoles per liter), triglycerides (millimoles per liter), LDL (millimoles per liter), and HDL (millimoles per liter) in 2017. SBP, systolic BP; DBP, diastolic BP; MAP, mean arterial pressure.

Table 4.

Adjusted odds ratios and 95% confidence intervals of the association of BP trajectory groups with subclinical kidney damage (sensitivity analysis)

Trajectory Groups	No. of Subjects with SRD in 2017 (%)	Unadjusted	Model 1	Model 2	Model 3
SBP trajectory group
Low stable	47 (8.23)	1.00	1.00	1.00	1.00
Moderate stable	89 (10.48)	1.31 (0.90 to 1.89)	1.12 (0.75 to 1.66)	1.09 (0.73 to 1.63)	1.06 (0.70 to 1.58)
High stable	28 (19.44)	2.69 (1.62 to 4.48)	1.81 (1.03 to 3.18)	1.48 (0.82 to 2.68)	1.51 (0.83 to 2.77)
Moderate increasing	34 (41.98)	8.07 (4.73 to 13.74)	4.93 (2.73 to 8.91)	3.44 (1.80 to 6.55)	3.24 (1.67 to 6.28)
DBP trajectory group
Low stable	60 (7.58)	1.00	1.00	1.00	1.00
Moderate stable	47 (10.54)	1.44 (0.96 to 2.15)	1.20 (0.78 to 1.84)	1.17 (0.76 to 1.80)	1.12 (0.73 to 1.74)
High stable	53 (17.15)	2.53 (1.70 to 3.75)	1.88 (1.21 to 2.94)	1.80 (1.14 to 2.82)	1.74 (1.09 to 2.76)
Moderate increasing	38 (38.78)	7.73 (4.76 to 12.54)	5.11 (2.97 to 8.79)	3.59 (1.94 to 6.64)	3.63 (1.94 to 6.78)
MAP trajectory group
Low stable	43 (6.60)	1.00	1.00	1.00	1.00
Moderate stable	91 (11.86)	1.91 (1.31 to 2.78)	1.54 (1.02 to 2.32)	1.51 (0.99 to 2.28)	1.43 (0.94 to 2.18)
High stable	24 (18.32)	3.18 (1.85 to 5.45)	2.30 (1.28 to 4.12)	2.07 (1.13 to 3.79)	2.05 (1.11 to 3.80)
Moderate increasing	40 (42.11)	10.30 (6.18 to 17.18)	6.15 (3.46 to 10.95)	4.48 (2.33 to 8.62)	4.28 (2.19 to 8.37)

Model 1 = sex, race, body mass index, heart rate, and waist-to-hip ratio in 2017. Model 2 = Model 1 + hypertension, diabetes, hyperlipidemia, smoking, and drinking in 2017. Model 3 = model 2 + fasting blood glucose (millimoles per liter), serum uric acid (micromoles per liter), total cholesterol (millimoles per liter), triglycerides (millimoles per liter), LDL (millimoles per liter), and HDL (millimoles per liter) in 2017. SBP, systolic BP; DBP, diastolic BP; MAP, mean arterial pressure.

Similarly, relative risk was evaluated in discrete DBP and MAP trajectories. ORs (95% CIs) were 1.54 (95% CI, 1.05 to 2.25) for moderate stable, 2.48 (95% CI, 1.70 to 3.62) for low increasing, and 8.45 (95% CI, 5.45 to 13.10) for moderate increasing in the DBP trajectories and 1.90 (95% CI, 1.32 to 2.74) for moderate stable, 4.11 (95% CI, 2.56 to 6.59) for high stable, and 9.36 (95% CI, 5.80 to 15.11) for moderate increasing in the MAP trajectories. The relative risk still existed, although ORs were attenuated after adjusting for confounding factors, except for the moderate stable group in both DBP and MAP trajectories (Table 3).

Sensitivity Analyses

We performed additional analyses after excluding individuals who had diabetes or received antihypertensive drugs (n=101) to eliminate the effect of these factors on the results. Trajectories in BP were identified among the rest of the 2329 participants. There was no significant change found in the characteristics of BP trajectories (Supplemental Figure 3). Among the 2329 subjects, there were 1639 eGFR values and 1560 uACR data available in 2017, leading to a total of 1645 individuals who had either a calculated eGFR or an uACR value. We analyzed the risk of SRD for each trajectory, and the results remained the same (Table 4). Taken together, the results of our study indicated that the trajectories of BP during early life were important for the development of CKD in later life.

Discussion

Four trajectory groups in SBP from childhood to middle age were identified. We illustrated that the SBP trajectories were significantly associated with uACR levels and the incidence of SRD in middle age (that is, the higher the level of SBP in early life course, the higher uACR and risk of SRD in middle age). In addition, rapid increase in SBP throughout the follow-up period predicted the highest uACR and risk of SRD in middle age. The analyzed results were similar for MAP trajectories. In DBP trajectory groups, subjects with DBP starting at either low or moderate levels with rapid increase during the follow-up period had higher uACR and risk of SRD in middle age.

The risk of renal damage is reported to increase with the increase in BP.^26–28 A meta-analysis among 4.2 million adults revealed that elevated SBP is closely related to high risk of peripheral arterial disease, which in turn, contributes to a nearly 30% increased risk of CKD.¹² A randomized, multicenter study determined that SBP is an independent predictor of renal function insufficiency among older subjects with isolated systolic hypertension.²⁹ These studies focusing on the relationship between BP and renal function status were mainly on the basis of BP data from single measurements, multiple measurements in a short time, or 24-hour ambulatory BP measurements. However, BP values were easily influenced by mental status, physical activity, and other factors, which may not adequately reflect individuals’ accurate BP level.

Group-based trajectory modeling takes into account variations in time to distinguish changes in BP over time and heterogeneity within multiple BP measurements. It assigns individuals sharing similar BP development trends to the same subgroup on the basis of multiple measurement data over a long time period. It provides an effective approach to reflect the life course association of certain predictors with its potential effect on target organ. Available evidence showed that continuous and high BP trajectories for years are closely correlated with subclinical atherosclerosis,³⁰ intima media thickness, and left ventricular mass index.³¹

Studies indicated that high BP in later life originates in childhood,⁹ and severe BP abnormality in childhood is responsible for increased risk of target organ damage secondary to hypertension.³² BP trajectories are identifiable in childhood and associated with adverse childhood experiences¹¹ and other modifiable risk factors over time.³³ Similarly, our study identified individuals’ BP (SBP, DBP, and MAP) trajectories throughout childhood, puberty, early adulthood, and middle-aged adulthood.

Interestingly, several studies have reported that group-based trajectory may predict individuals’ renal function status. Data from a large population-based health examination have indicated that continuous high SBP was related to kidney damage in nonhypertensive subjects.³⁴ Unfortunately, the study did not fit DBP or MAP trajectories. In our study, we found that continuous high SBP trajectory (high stable group) was closely associated with increased uACR levels and SRD, so did MAP trajectories. Furthermore, subjects with rapid increase in BP were assigned to moderate increasing or low increasing groups, which had stronger predictive ability for SRD in middle age. Evidence from the Mater-University of Queensland Study of Pregnancy (MUSP) birth cohort showed that continuous high and middle pulse pressures rather than MAP trajectories in early life can predict the risk of lower eGFR by 30 years of age.³⁵ However, our study found that the predictive ability of MAP trajectories was similar to that of SBP trajectories. The difference may be caused by various characteristics of the analyzed population, because the population of the MUSP birth cohort study was relatively young (average age was 30 years old at the last follow-up) with a lower incidence of CKD, which was defined by eGFR using serum creatinine values alone. In our study, subjects were followed up for 30 years, and the renal function was assessed by eGFR and uACR jointly. In addition, we reported that uACR was higher in the moderate stable, high stable, and moderate increasing groups compared with the low stable group, and MAP trajectories were a strong predictor for the risk of SRD in middle age.

In addition to the effect of hypertension on SRD, notable differences in sex were observed in the different BP trajectory groups. Women were more likely to be included in the low stable group, whereas men were more likely to be included in the high stable (low increasing) and moderate increasing groups. This result is consistent with previous reports.^9,35 Cigarette smoking and alcohol consumption are not only risk factors for hypertension but also, independently associated with CKD.^36,37 In our study, the high stable (low increasing) and moderate increasing groups were associated with higher rates of smoking and drinking in SBP and DBP trajectories. Furthermore, higher-level trajectory groups had markedly abnormal metabolic risk factors, because the moderate stable, high stable (low increasing), and moderate increasing groups were related to higher incidence of diabetes; hyperlipidemia; higher BMI, GLU, TC, TG, and LDL-C; and lower HDL-C. BMI in early life played an important role in the development of hypertension³⁸ and CKD.³⁹ Locatelli et al.⁴⁰ reported that patients with metabolic disorder were at higher risk for microalbuminuria or CKD, and the number of metabolic disorders determined the level of risk. Another study showed that GLU levels of 110 mg/dl, reduced HDL-C, and high TG levels were independently associated with the greatest risk for microalbuminuria and low eGFR.⁴¹ Although mechanisms of the inter-relationship between metabolic disorders and renal damage were not fully elucidated, several pathophysiologic factors were involved, including oxidative stress, inflammation, and adipocytokines.⁴² Serum UA proved to be a strong predictor for increased urine albumin excretion and the development of CKD⁴³: the concentration of serum UA increased with the increase of BP trajectories in our study.

Although our study used a large population, had high response rate, was prospective in nature, and used a community-based cohort followed for 30 years, it had several limitations, which should be acknowledged. The participants were recruited from multiple rural areas in northern China, and most of them were Han nationality (98.2%). Therefore, our findings should be verified in other cohorts to determine generalizability to other ethnicities and populations with different backgrounds. In addition, not all subjects had BP data available during visits, especially follow-up in 2005; we only randomly selected about 500 individuals to visit, resulting in a large number of individuals’ data being unavailable between 1995 and 2013. However, the analyzed population had three or more BP measurements, including one or more measurements from visits in 1987, 1989, 1992, and 1995 and at least one measurement in 2005, 2013, and 2017. Indeed, the majority of the 2430 subjects (n=2054; 84.53%) had five or more BP measurements. Generally, variability in BP was more significant before than after adulthood; thus, we visited the population more frequently from 1987 to 1995 (average age was 11.6–19.6 years old), resulting in four visits within 8 years. We visited the population three times in their adulthood (2005, 2013, and 2017). These measures ensured the accuracy of the trajectory construction and preciseness of the conclusion. Finally, participants with mean age of 42 years old in the last follow-up experienced a relatively low prevalence of SRD; in particular, the number of individuals who met the eGFR range was rather small. To our knowledge, this is the first study that has taken into account the variability of SBP, DBP, and MAP in early life course and revealed the association between BP variation trends and SRD among healthy young adults in a community-based cohort. The prospective design of our research provided us an opportunity to achieve further follow-up in determining the future risk of clinical CKD development.

BP trajectories throughout early life are independently associated with uACR level and SRD risk. Higher BP trajectories were correlated with higher level of uACR and risk of SRD in middle age, indicating that identifying long-term trajectories of BP from early age may assist in predicting individuals’ renal function status in later life.

Disclosures

None.

Supplemental Material

Supplemental Table 1. Average systolic and diastolic BP (millimeters of Hg) by age periods in MAP trajectory groups.

Supplemental Table 2. Average systolic and diastolic BP (millimeters of Hg) by age periods in DBP trajectory groups.

Supplemental Table 3. Demographic characteristics and cardiovascular risk factors by DBP trajectory group.

Supplemental Table 4. Demographic characteristics and cardiovascular risk factors by MAP trajectory group.

Supplemental Table 5. Characteristics in SBP, DBP, and BMI by SBP trajectories.

Supplemental Table 6. Characteristics in SBP, DBP, and BMI by DBP trajectories.

Supplemental Table 7. Characteristics in SBP, DBP, and BMI by MAP trajectories.

Supplemental Table 8. Pearson correlation coefficients between BMI and BP.

Supplemental Table 9. Comparison of eGFR (millimeters per minute per 1.73 m²) in the different groups.

Supplemental Table 10. Number and percentage of subjects with subclinical renal damage in BP trajectory groups.

Supplemental Figure 1. Percentage of SBP trajectories by DBP and MAP trajectory groups.

Supplemental Figure 2. (A) The percentage of DBP trajectories by SBP and MAP trajectory groups and (B) the percentage of MAP trajectories by SBP and DBP trajectory groups.

Supplemental Figure 3. Trajectories in (A) SBP, (B) DBP, and (C) MAP in the Hanzhong Adolescent Hypertension Cohort (sensitivity analysis).