2024 年 31 巻 7 号 p. 1072-1086
Aims: Weight changes from a young age are known to be associated with poor life outcomes in the general population. However, little is known about the association between weight change from a young age and life expectancy in patients with chronic kidney disease (CKD).
Methods: Data of 2,806 nondialysis CKD patients who participated in the Fukuoka Kidney Disease Registry (FKR) Study, a multicenter observational study, were analyzed. The primary outcome was all-cause death, whereas the secondary outcome was cardiovascular mortality. The covariate of interest was weight change, defined as the difference between body weight at study enrollment and at 20 years old. Cox proportional-hazards models were used to estimate the risks of mortality for participants with weight changes of ≥ 5 or ＜5 kg compared with those with stable weights.
Results: During the 5-year observation period, 243 participants died from all causes and 62 from cardiovascular disease. The risk of all-cause mortality in the weight-loss group was significantly higher than that in the stable-weight group (multivariable-adjusted hazard ratio, 2.11; 95% confidence interval [CI], 1.52–2.93). Conversely, the risk of cardiovascular mortality in the weight-loss group was significantly higher than that in the stable-weight group (multivariable-adjusted hazard ratio, 2.48; 95% CI, 1.32–4.64). However, no significant association was observed between weight gain and the risks of all-cause and cardiovascular mortalities.
Conclusion: Weight loss from 20 years of age was found to be associated with higher risks of all-cause and cardiovascular mortalities in patients with CKD.
Chronic kidney disease (CKD) is known to be associated with higher risks of all-cause and cardiovascular mortalities1). The global prevalence of CKD and incidence of related mortality is increasing, presenting an important global health challenge2). Cardiovascular disease (CVD) is the leading cause of death among patients with advanced CKD3, 4), and various risk factors for all-cause or cardiovascular mortality in patients with CKD have been identified, including advanced age5), low estimated glomerular filtration rate (eGFR)6), substantial albuminuria6), hypertension7), and smoking8). However, other modifiable risk factors need to be identified and addressed.
Long-term weight change in adulthood has been demonstrated to have numerous health consequences. For instance, weight gain from young adulthood is associated with the onset of diabetes9), onset of CKD10), cardiovascular mortality11), and all-cause mortality12, 13). Similarly, long-term weight loss in adulthood has been associated with cognitive impairment14) and cardiovascular mortality15). However, the association between weight change from a young age and all-cause and cardiovascular mortalities in nondialysis CKD patients has rarely been investigated.
Characterization of the association between weight change and future mortality in patients with CKD could help clinicians identify those who are at a high risk of mortality and establish effective treatment plans. Furthermore, identification of such patients enables appropriate recruitment of participants for clinical trials to evaluate the efficacy of novel therapies in those who are most likely to benefit. Therefore, the present study aimed to investigate the effects of weight change from a young age on all-cause and cardiovascular mortalities using data from a long-term longitudinal cohort of patients with CKD.
Japanese patients with CKD who were ≥ 16 years old, were under the care of a nephrologist, and were not dependent on dialysis from multiple centers in the Fukuoka and Saga Prefectures were prospectively enrolled in the Fukuoka Kidney Disease Registry (FKR) Study between January 2013 and March 2017 16-18). All the patients were followed up for 5 years until the death the development of end-stage kidney disease. A total of 4,476 patients from 12 centers were enrolled in the registry, of whom 2 withdrew their consent, 22 either did not consent or did not attend outpatient clinics, and 84 were lost to follow-up. Furthermore, 1,171 individuals for whom body weight data at the time of study enrollment or at 20 years of age were unavailable and 382 for whom baseline data were missing were excluded. Individuals who were below 30 years old (n=63) were also excluded, as previously described15), to evaluate the effects of weight change for at least 10 years from the age of 20. Data for the remaining 2,806 patients were analyzed in this study.
The study protocol was approved by the Clinical Research Ethics Committee of the Institutional Review Board of Kyushu University (approval number: 469-09) and the ethics committees of all the participating institutions. The study was registered in the UMIN Clinical Trials Registry (UMIN000007988) and conducted in accordance with the principles of the Declaration of Helsinki and its amendments. Informed consent was obtained from all study participants and/or their legal guardians.
Outcomes and ExposuresThe primary outcome was all-cause mortality, whereas the secondary outcome was cardiovascular mortality. Details of these events were obtained from the medical records of the participants. Cardiovascular mortality was defined as a combination of mortality resulting from cardiovascular or cerebrovascular events and sudden death. Cardiovascular events included heart failure, coronary artery disease, peripheral artery disease, aortic dissection, and ruptured aortic aneurysm. The cerebrovascular events included cerebral infarction, cerebral hemorrhage, and subarachnoid hemorrhage. Sudden death was defined as unexpected death occurring within 24 h of symptom onset. Weight change, the primary exposure used in this study, was defined as the difference between body weight at the time of study enrollment and that at 20 years old. The weight of the participants was measured at the time of study enrollment, and they reported their weight at 20 years old in the questionnaire that they completed at baseline.
CovariatesAt the time of study enrollment, the participants’ demographic and clinical data were collected. The blood pressure values of the participants were extracted from their medical records. The physicians or medical personnel in the clinics measured the blood pressures of patients in the sitting position using appropriately sized cuffs placed on their upper arms. Nonfasting serum and urine biochemical parameters, including the serum concentrations of creatinine, albumin, total cholesterol, and high-sensitivity C-reactive protein (hs-CRP), were measured at a central laboratory. eGFR was calculated using the appropriate equation for Japanese patients with CKD aged ≥ 18 years: eGFR (mL/min/1.73 m2)=194×creatinine−1.094×age−0.287 (×0.739 for women)19). Other biochemical data, including hemoglobin A1c values, were extracted from the participants’ medical records. Body mass index (BMI) was calculated using the formula BMI (kg/m2)=weight (kg) / (height (m))2. The participants completed the questionnaires regarding their smoking and alcohol consumption habits.
Statistical AnalysisThe participants were divided into three groups according to their weight change using previously reported criteria15). Group 1 (weight-loss group) consisted of individuals who experienced a weight loss of ＞5 kg. Group 2 (stable-weight group) included individuals whose weight had changed by ＜5 kg since the age of 20. Group 3 (weight-gain group) consisted of individuals who had gained ≥ 5 kg. The baseline characteristics of the participants, classified according to their weight change, were expressed as number and percentage for categorical data and as median and interquartile range for continuous data. The trends across the weight-change groups were evaluated using the Cochran–Armitage test for categorical data and the Jonckheere–Terpstra test for continuous data. Furthermore, Kaplan–Meier curves and log-rank tests were employed to compare the all-cause and cardiovascular mortality rates of the participants in the three weight-change groups.
Unadjusted, age- and sex-adjusted, and multivariable-adjusted Cox proportional-hazards models were used to calculate the hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause and cardiovascular mortalities in the participants. The multivariable-adjusted models were adjusted for age, sex, the presence of diabetes, a history of CVD, a history of cancer, systolic blood pressure, BMI, blood hemoglobin concentration, serum albumin concentration, serum hs-CRP concentration, eGFR, serum total cholesterol concentration, and smoking habits. These variables were included based on an a priori clinical judgment and published findings. The assumptions of proportional-hazards analysis were validated using log cumulative hazard plots for all-cause mortality, stratified according to weight change and tested using an analysis of Schoenfeld residuals.
We conducted two distinct sensitivity analyses to evaluate the robustness of our findings. In this study, weight change classification was based on previous studies in the general population, and there was a concern that it might not be appropriate for CKD patients. Thus, the first sensitivity analysis was additionally conducted to test the robustness of the primary findings using the tertile points of weight change as the cutoff values. Group details by tertile of weight change were T1 (n=971), ＜−0.1 kg; T2 (n=936), −0.1 to 8.0 kg; and T3 (n=899), ≥ 8 kg. T2 was used as the reference group, and the HRs for T1 and T3 were calculated. As the second sensitivity analysis, all-cause mortality was compared in all groups using propensity score (PS) matching analysis to minimize potential confounding and selection biases in this observational study. PS for weight change was calculated for G1 relative to G2 and G3 relative to G2 using logistic models with the same covariates as the potential confounders previously described to calculate PS20, 21). Matched pairs were generated using the greedy matching algorithm with a caliper width of 0.2 standard deviations of the logit of the PS at a 1:1 ratio without replacement. As the FKR study was a cohort study of nondialysis CKD patients and follow-up was terminated when end-stage kidney disease developed, all-cause mortality was assessed using the Fine–Gray proportional-hazards model, with end-stage kidney disease considered as a competing risk22). In addition, cardiovascular mortality was evaluated using the Fine–Gray proportional-hazards model, in which noncardiovascular mortality and end-stage kidney disease were treated as competing risks.
Subgroup analysis was also conducted to test the heterogeneity in the associations between subgroups by adding multiplicative interaction terms to the relevant Cox models and using the likelihood ratio test, which compares models with and without interaction terms. Furthermore, multivariable-adjusted restricted cubic spline models were used to identify any nonlinear association between weight change and the risks of all-cause and cardiovascular mortalities. Restricted cubic spline curves were plotted using four knots located at the 5th, 35th, 65th, and 95th percentiles of weight change, with the median weight change (4.1 kg) used as reference for each spline plot23).
A two-tailed p-value＜0.05 was considered statistically significant in all the analyses. Statistical analyses were conducted using R version 4.2.2 (The R Foundation for Statistical Computing, Vienna, Austria).
A histogram showing the distribution of weight change from 20 years of age to the time of the study is presented in Fig.1. This was found to be approximately normally distributed.
Histogram of weight change from 20 years of age
Data was stratified according to the weight change of the participants from 20 years of age, and the groups were compared with respect to each parameter (Table 1). Height, body weight at enrollment in the present study, BMI, systolic blood pressure, diastolic blood pressure, and blood hemoglobin, serum total protein, serum albumin, serum hs-CRP, serum total cholesterol, and serum calcium concentrations all increased with increasing weight change. Contrarily, the prevalence of a history of CVD, ischemic heart disease, congestive heart failure, stroke, peripheral arterial disease, bone fracture, ischemic heart disease, and cancer decreased with increasing body weight.
| Characteristic |
Overall, N = 2,806 |
Weight change from 20 years of age (kg) | P for trend | ||
|---|---|---|---|---|---|
|
Group 1 ＜−5 N = 469 |
Group 2 −5 to 5 N = 996 |
Group 3 ≥ 5 N = 1,341 |
|||
| Demographics and comorbidities | |||||
| Age, years | 67.0 (58.0, 75.0) | 72.0 (64.0, 79.0) | 67.0 (55.8, 75.0) | 66.0 (57.0, 73.0) | ＜0.001 |
| Male sex, n | 1,584 (56.5%) | 275 (58.6%) | 510 (51.2%) | 799 (59.6%) | 0.14 |
| Diabetes mellitus, n | 773 (27.5%) | 156 (33.3%) | 241 (24.2%) | 376 (28.0%) | 0.25 |
| Height, cm | 160.0 (154.0, 167.0) | 160.0 (153.0, 166.0) | 160.0 (153.0, 165.0) | 161.3 (155.0, 168.0) | ＜0.001 |
| Body weight, kg | 59.5 (51.8, 67.7) | 52.0 (45.0, 58.5) | 54.0 (48.0, 60.5) | 66.0 (59.7, 73.5) | ＜0.001 |
| Body weight at 20 years old, kg | 55.0 (50.0, 62.0) | 61.0 (55.0, 70.0) | 54.0 (48.0, 60.0) | 55.0 (49.0, 60.0) | ＜0.001 |
| Body mass index, kg/m2 | 22.9 (20.7, 25.6) | 20.4 (18.5, 22.6) | 21.3 (19.7, 22.9) | 25.3 (23.3, 27.5) | ＜0.001 |
| Smoking habit, n | 1,493 (53.2%) | 265 (56.5%) | 466 (46.8%) | 762 (56.8%) | 0.12 |
| Systolic blood pressure, mmHg | 130.0 (120.0, 142.0) | 130.0 (118.0, 141.0) | 129.0 (118.5, 142.0) | 131.0 (121.0, 142.0) | ＜0.001 |
| Diastolic blood pressure, mmHg | 75.0 (68.0, 82.0) | 71.5 (64.0, 79.0) | 74.0 (67.0, 82.0) | 76.0 (70.0, 83.0) | ＜0.001 |
| History of cardiovascular disease, n | 637 (22.7%) | 159 (33.9%) | 213 (21.4%) | 265 (19.8%) | ＜0.001 |
| History of ischemic heart disease, n | 320 (11.4%) | 79 (16.8%) | 103 (10.3%) | 138 (10.3%) | ＜0.001 |
| History of congestive heart failure, n | 77 (2.7%) | 29 (6.2%) | 22 (2.2%) | 26 (1.9%) | ＜0.001 |
| History of stroke, n | 262 (9.3%) | 63 (13.4%) | 87 (8.7%) | 112 (8.4%) | 0.003 |
| History of peripheral artery disease, n | 84 (3.0%) | 27 (5.8%) | 29 (2.9%) | 28 (2.1%) | ＜0.001 |
| History of bone fracture, n | 164 (5.8%) | 51 (10.9%) | 49 (4.9%) | 64 (4.8%) | ＜0.001 |
| History of cancer, n | 369 (13.2%) | 83 (17.7%) | 116 (11.6%) | 170 (12.7%) | 0.04 |
| Laboratory tests | |||||
| Blood hemoglobin, g/dL | 12.8 (11.5, 14.1) | 11.6 (10.6, 12.9) | 12.6 (11.4, 13.9) | 13.3 (12.1, 14.6) | ＜0.001 |
| Serum total protein, g/dL | 6.9 (6.6, 7.3) | 6.9 (6.4, 7.2) | 6.9 (6.5, 7.2) | 7.0 (6.6, 7.3) | ＜0.001 |
| Serum albumin, g/dL | 4.1 (3.8, 4.3) | 3.9 (3.7, 4.2) | 4.1 (3.8, 4.3) | 4.1 (3.9, 4.3) | ＜0.001 |
| Serum high-sensitivity C-reactive protein, mg/dL | 0.05 (0.02, 0.12) | 0.05 (0.02, 0.11) | 0.04 (0.02, 0.10) | 0.07 (0.03, 0.15) | ＜0.001 |
| Serum total cholesterol, mg/dL | 192 (169, 218) | 184 (160, 212) | 195 (173, 220) | 192 (169, 216) | 0.09 |
| Serum urea nitrogen, mg/dL | 22.5 (16.0, 34.4) | 30.4 (20.5, 45.5) | 21.8 (16.0, 33.0) | 20.8 (15.4, 31.0) | ＜0.001 |
| Serum creatinine, mg/dL | 1.3 (0.9, 2.0) | 1.6 (1.2, 2.7) | 1.2 (0.8, 2.0) | 1.2 (0.9, 1.8) | ＜0.001 |
| eGFR, mL/min/1.73 m2 | 40.7 (24.2, 57.3) | 27.6 (17.3, 44.0) | 42.3 (24.4, 59.1) | 43.5 (27.9, 59.5) | ＜0.001 |
| Serum phosphate, mg/dL | 3.4 (3.0, 3.8) | 3.5 (3.1, 4.0) | 3.5 (3.0, 3.8) | 3.3 (2.9, 3.8) | ＜0.001 |
| Serum calcium, mg/dL | 9.4 (9.0, 9.7) | 9.2 (8.9, 9.5) | 9.4 (9.0, 9.6) | 9.4 (9.1, 9.7) | ＜0.001 |
| Serum PTH (intact assay), pg/mL | 67 (47, 108) | 80 (51, 140) | 65 (45, 105) | 65 (48, 102) | ＜0.001 |
| Urinary albumin-to-creatinine ratio, mg/gCr | 196.6 (33.7, 736.7) | 249.4 (41.0, 941.3) | 170.4 (28.1, 701.0) | 200.4 (37.3, 713.7) | ＜0.001 |
Baseline data are expressed as median (interquartile range) or number (percentage). Conversion factors for units: hemoglobin in g/dL to g/L, ×10; protein in g/dL to g/L, ×10; albumin in g/dL to g/L, ×10; high-sensitivity C-reactive protein in mg/dL to nmol/L, ×9.524; cholesterol in mg/dL to mmol/L,×0.02586; urea nitrogen in mg/dL to mmol/L, ×0.357; creatinine in mg/dL to mmol/L, ×88.4; phosphate in mg/dL to mmol/L, ×0.3229; calcium in mg/dL to mmol/L. Abbreviations: eGFR, estimated glomerular filtration rate; PTH, parathyroid hormone.
During the 5-year observation period (mean duration, 1,568 days; standard deviation, 488 days), 243 participants (8.7%) died from all causes, including 93, 66, and 84 in Groups 1, 2, and 3, respectively. Unadjusted Kaplan–Meier curves exhibited a significant difference in the event-free survival probability with respect to all-cause mortality among the weight-change groups (log-rank test, P＜0.05; Fig.2). The Cox proportional-hazards risk models indicated that Group 1 had a significantly higher HR than Group 2, which was the reference. This finding remained significant even after adjustment for age and sex and for multiple confounding factors (Table 2). The HRs (95% Cis) for the weight-change groups were 2.14 (1.55–2.96) for Group 1, 1.00 (reference) for Group 2, and 1.02 (0.70–1.47) for Group 3 using the multivariable-adjusted model. The Fine–Gray proportional-hazards model was used to allow for the influence of competing risks on the association between weight change from 20 years of age and the risk of all-cause mortality. When end-stage kidney disease was included as a competing risk, the risk of all-cause mortality in Group 1 was significantly higher (P＜0.01) than that in Group 2 (HR [95% CI]): Group 1, 2.06 [1.47–2.88]; Group 2, 1.00 [reference]; and Group 3, 1.05 [0.74–1.5]] (Table 2). In addition, the multivariable-adjusted restricted cubic spline curve showed that the HR for all-cause mortality gradually increased with the degree of weight loss since 20 years of age, suggesting a nonlinear association (P for nonlinearity ＜0.05; Fig.3). Heterogeneity was not observed among the groups regarding the association between weight change from 20 years of age and the risk of all-cause mortality, except for blood hemoglobin concentration (Table 3). In the first sensitivity analysis, the participants were divided into three groups according to the tertiles of weight change, and the groups were compared with respect to each parameter (Supplemental Table 1). The numbers of deaths from all causes during the observation period were 129, 56, and 58 for T1, T2, and T3. In the Cox proportional-hazards risk models, T1 had a significantly (P＜0.05) higher HR than T2, the reference, even after adjustment for multiple confounding factors. In the multivariable-adjusted model, the HRs (95% CIs) were 1.74 (1.74–2.43) for T1, 1.00 (reference) for T2, and 1.19 (0.79–1.78) for T3 (Supplemental Table 2). In the second sensitivity analysis, all-cause mortality was compared in all groups using PS matching analysis. In the PS-matched cohort, Group 1 had a significantly higher HR than Group 2 (reference). The HRs (95% CIs) for each weight-change group were 1, 1.72 (1.18–2.51) for Group 1, 1.00 (reference) for Group 2, and 0.80 (0.50–1.31) for Group 3 (Supplemental Table 3).
Survival rates of the participants with respect to all-cause mortality, categorized according to weight change, during the 5-year follow-up period
| Weight change since 20 years of age (kg) | |||
|---|---|---|---|
|
Group 1 ＜−5 |
Group 2 −5 to 5 |
Group 3 ≥ 5 |
|
| All-cause mortality, n (%) | 93 (20) | 66 (6.6) | 84 (6.3) |
| Unadjusted model | 3.66 (2.68–5.00) | 1.00 (reference) | 0.91 (0.66–1.26) |
| Age- and sex-adjusted model | 2.46 (1.80–3.38) | 1.00 (reference) | 1.06 (0.77–1.46) |
| Multivariable-adjusted model | 2.11 (1.52–2.93) | 1.00 (reference) | 1.02 (0.71–1.47) |
| Fine & Gray model | 2.05 (1.46–2.88) | 1.00 (reference) | 1.05 (0.73–1.50) |
Values are presented as HRs (95% CIs). The multivariable-adjusted models and Fine & Gray models were adjusted for age; sex; smoking habits; body mass index; systolic blood pressure; history of cardiovascular disease; history of cancer; presence of diabetes; the blood hemoglobin, serum albumin, serum high-sensitivity C-reactive protein, and serum total cholesterol concentrations; eGFR; and urinary albumin-to-creatinine ratio. Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio.
Association between percentages of weight change from 20 years of age and all-cause mortality in patients with CKD, allowing for nonlinear effects, with 95% confidence intervals (CIs). The x-axis represents the weight change as a percentage, calculated by dividing the weight change by the weight at age 20 and multiplying by 100. The solid line denotes the HR; gray area, the 95% CI; and dotted line, an HR of 1.0. The model was fitted with four knots and was adjusted for age; sex; the presence of diabetes; a history of cardiovascular disease; a history of cancer; systolic blood pressure; body mass index; blood hemoglobin, serum albumin, serum hs-CRP, and serum total cholesterol concentrations; eGFR; and smoking habits. The curves show the HRs compared with the reference weight change of 4.1 kg. Abbreviations: CKD, chronic kidney disease; CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio.
| Subgroups | Weight change since 20 years of age (kg) | P for heterogeneity | ||
|---|---|---|---|---|
|
Group 1 ＜−5 |
Group 2 −5 to 5 |
Group 3 ≥ 5 |
||
| Age, years | 0.57 | |||
| ＜65 | 4.86 (0.83–28.4) | 1.00 (reference) | 1.16 (0.20–6.73) | |
| ≥ 65 | 1.93 (1.38–2.70)＊ | 1.00 (reference) | 0.88 (0.60–1.28) | |
| Sex | 0.30 | |||
| Female | 1.81 (0.98–3.35) | 1.00 (reference) | 0.68 (0.31–1.49) | |
| Male | 2.19 (1.47–3.26)＊ | 1.00 (reference) | 1.15 (0.75–1.77) | |
| Body mass index, kg/m2 | 0.96 | |||
| ＜22.9 | 2.00 (1.36–2.94)＊ | 1.00 (reference) | 1.09 (0.60–1.99) | |
| ≥ 22.9 | 2.29 (1.17–4.46)＊ | 1.00 (reference) | 0.98 (0.59–1.63) | |
| Smoking habit | 0.44 | |||
| Absence | 1.71 (0.95–3.08) | 1.00 (reference) | 0.67 (0.33–1.38) | |
| Presence | 2.37 (1.58–3.56)＊ | 1.00 (reference) | 1.19 (0.77–1.85) | |
| Diabetes | 0.36 | |||
| Absence | 2.07 (1.36–3.14)＊ | 1.00 (reference) | 1.54 (0.97–2.45) | |
| Presence | 1.87 (1.09–3.24)＊ | 1.00 (reference) | 0.58 (0.31–1.09) | |
| Systolic blood pressure, mmHg | 0.58 | |||
| ＜130 | 2.64 (1.61–4.32)＊ | 1.00 (reference) | 0.86 (0.49–1.52) | |
| ≥ 130 | 1.89 (1.21–2.96)＊ | 1.00 (reference) | 1.21 (0.75–1.97) | |
| History of cardiovascular events | 0.88 | |||
| Absence | 2.35 (1.46–3.77)＊ | 1.00 (reference) | 1.13 (0.67–1.91) | |
| Presence | 1.90 (1.19–3.02)＊ | 1.00 (reference) | 0.90 (0.53–1.53) | |
| History of cancer | 0.11 | |||
| Absence | 2.07 (1.41–3.04)＊ | 1.00 (reference) | 0.92 (0.59–1.42) | |
| Presence | 2.32 (1.20–4.51)＊ | 1.00 (reference) | 1.35 (0.66–2.75) | |
| Blood hemoglobin level, g/dL | 0.04 | |||
| ＜12.9 | 1.78 (1.24–2.57)＊ | 1.00 (reference) | 0.80 (0.51–1.26) | |
| ≥ 12.9 | 4.05 (1.88–8.72)＊ | 1.00 (reference) | 1.74 (0.85–3.53) | |
| Serum albumin, g/dL | 0.14 | |||
| ＜4.1 | 2.00 (1.36–2.96)＊ | 1.00 (reference) | 1.06 (0.68–1.65) | |
| ≥ 4.1 | 2.47 (1.31–4.66)＊ | 1.00 (reference) | 1.04 (0.54–2.00) | |
| Serum high-sensitivity C-reactive protein, mg/dL | 0.14 | |||
| ＜0.05 | 2.33 (1.32–4.14)＊ | 1.00 (reference) | 0.99 (0.48–2.04) | |
| ≥ 0.05 | 2.39 (1.54–3.73)＊ | 1.00 (reference) | 1.08 (0.66–1.78) | |
| Serum total cholesterol, mg/dL | 0.14 | |||
| ＜192 | 2.04 (1.36–3.05)＊ | 1.00 (reference) | 0.86 (0.53–1.41) | |
| ≥ 192 | 2.20 (1.22–3.98)＊ | 1.00 (reference) | 1.33 (0.75–2.35) | |
| eGFR, mL/min/1.73 m2 | 0.19 | |||
| ＜40.6 | 1.84 (1.27–2.65)＊ | 1.00 (reference) | 0.98 (0.64–1.47) | |
| ≥ 40.6 | 3.47 (1.57–7.64)＊ | 1.00 (reference) | 1.19 (0.51–2.76) | |
| Urinary albumin-to-creatinine ratio, mg/gCr | 0.53 | |||
| ＜0.197 | 1.85 (1.11–3.07)＊ | 1.00 (reference) | 0.89 (0.51–1.55) | |
| ≥ 0.197 | 2.40 (1.54–3.73)＊ | 1.00 (reference) | 1.08 (0.66–1.77) | |
＊P＜0.05 vs. Group 2 for each subgroup.
Models were adjusted for age; sex; body mass index; smoking habits; presence of diabetes; systolic blood pressure; history of cardiovascular events; history of cancer; the presence of diabetes; the blood hemoglobin, serum albumin, serum high-sensitivity C-reactive protein, and serum total cholesterol concentration; estimated glomerular filtration rate; and urinary albumin-to-creatinine ratio. The variable relevant to the subgroup was excluded from each model. Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio.
| Characteristic | Weight change from 20 years of age (kg) | P for trend | |||
|---|---|---|---|---|---|
|
T1 ＜-0.1 |
T2 −0.1 to 8.0 |
T3 ≥ 8.0 |
|||
| Overall, N = 2,806 | N = 935 | N = 972 | N = 899 | ||
| Demographics and comorbidities | |||||
| Age, years | 67.0 (58.0, 75.0) | 70.0 (60.0, 78.0) | 66.0 (56.0, 74.0) | 66.0 (57.0, 73.0) | ＜0.001 |
| Male sex, n | 1,584.0 (56.5%) | 495.0 (52.9%) | 546.0 (56.2%) | 543.0 (60.4%) | 0.005 |
| Diabetes mellitus, n | 773.0 (27.5%) | 262.0 (28.0%) | 251.0 (25.8%) | 260.0 (28.9%) | 0.3 |
| Height, cm | 160.0 (154.0, 167.0) | 159.2 (153.0, 165.5) | 160.0 (154.0, 167.0) | 162.0 (155.0, 168.4) | ＜0.001 |
| Body weight, kg | 59.5 (51.8, 67.7) | 52.0 (45.5, 58.5) | 58.2 (52.4, 64.0) | 68.5 (62.9, 76.0) | ＜0.001 |
| Body weight at 20 years old, kg | 55.0 (50.0, 62.0) | 58.0 (51.0, 65.0) | 54.0 (48.0, 60.0) | 55.0 (49.0, 60.0) | ＜0.001 |
| Body mass index, kg/m2 | 22.9 (20.7, 25.6) | 20.5 (18.8, 22.5) | 22.6 (21.0, 24.2) | 26.2 (24.4, 28.4) | ＜0.001 |
| Smoking habit, n | 1,493.0 (53.2%) | 474.0 (50.7%) | 498.0 (51.2%) | 521.0 (58.0%) | 0.002 |
| Systolic blood pressure, mmHg | 130.0 (120.0, 142.0) | 129.0 (118.0, 140.5) | 130.0 (120.0, 142.0) | 132.0 (122.0, 142.5) | ＜0.001 |
| Diastolic blood pressure, mmHg | 75.0 (68.0, 82.0) | 72.5 (65.0, 80.0) | 74.5 (68.0, 81.5) | 77.0 (70.0, 84.0) | ＜0.001 |
| Histrory of cardiovascular disease, n | 637 (22.7%) | 260 (27.8%) | 200 (20.6%) | 177 (19.7%) | ＜0.001 |
| Histrory of ischemic heart disease, n | 320 (11.4%) | 124 (13.3%) | 102 (10.5%) | 94 (10.5%) | 0.091 |
| History of congestive heart failure, n | 77 (2.7%) | 42 (4.5%) | 17 (1.7%) | 18 (2%) | ＜0.001 |
| Histrory of stroke, n | 262 (9.3%) | 105 (11.2%) | 86 (8.8%) | 71 (7.9%) | 0.04 |
| History of peripheral artery disease, n | 84 (3.0%) | 43 (4.6%) | 19 (2.0%) | 22 (2.4%) | 0.002 |
| History of bone fracture, n | 164 (5.8%) | 78 (8.3%) | 46 (4.7%) | 40 (4.5%) | ＜0.001 |
| History of cancer, n | 369 (13.2%) | 129 (13.8%) | 119 (12.2%) | 121 (13.5%) | 0.6 |
| Laboratory tests | |||||
| Blood hemoglobin, g/dL | 12.8 (11.5, 14.1) | 12.0 (10.9, 13.2) | 13.1 (11.8, 14.2) | 13.4 (12.2, 14.7) | ＜0.001 |
| Serum total protein, g/dL | 6.9 (6.6, 7.3) | 6.9 (6.5, 7.2) | 6.9 (6.6, 7.3) | 7.0 (6.7, 7.3) | ＜0.001 |
| Serum albumin, g/dL | 4.1 (3.8, 4.3) | 4.0 (3.7, 4.2) | 4.1 (3.9, 4.4) | 4.1 (3.9, 4.3) | ＜0.001 |
| Serum high-sensitivity C-reactive protein, mg/dL | 0.1 (0.0, 0.1) | 0.0 (0.0, 0.1) | 0.0 (0.0, 0.1) | 0.1 (0.0, 0.2) | ＜0.001 |
| Serum total cholesterol, mg/dL | 192.0 (169.0, 218.0) | 191.0 (165.5, 218.0) | 194.0 (173.0, 217.3) | 191.0 (167.5, 217.0) | 0.27 |
| Serum urea nitrogen, mg/dL | 22.5 (16.0, 34.4) | 26.4 (17.9, 41.0) | 20.8 (15.1, 32.0) | 21.0 (15.8, 30.6) | ＜0.001 |
| Serum creatinine, mg/dL | 1.3 (0.9, 2.0) | 1.5 (1.0, 2.4) | 1.2 (0.9, 1.9) | 1.2 (0.9, 1.7) | ＜0.001 |
| eGFR, ml/min/m2 | 40.7 (24.2, 57.3) | 32.4 (19.3, 51.6) | 44.0 (25.3, 60.1) | 43.1 (28.6, 59.1) | ＜0.001 |
| Serum phosphate, mg/dL | 3.4 (3.0, 3.8) | 3.5 (3.1, 4.0) | 3.4 (3.0, 3.8) | 3.3 (2.9, 3.7) | ＜0.001 |
| Serum calcium, mg/dL | 9.4 (9.0, 9.7) | 9.3 (8.9, 9.6) | 9.4 (9.1, 9.7) | 9.4 (9.1, 9.7) | ＜0.001 |
| Serum PTH(intact assay), pg/mL | 67.0 (47.0, 108.0) | 72.0 (47.0, 119.3) | 63.0 (46.0, 102.0) | 66.0 (48.0, 102.0) | 0.06 |
| Urinary albumin to creatinine ratio, mg/gCr | 196.6 (33.7, 736.7) | 218.8 (37.0, 802.0) | 164.6 (24.5, 729.6) | 204.4 (46.7, 657.0) | 0.38 |
Baseline data are expressed as median (interquartile range) or number (percentage). The trends across the weight change groups were evaluated using the Cochran–Armitage test for categorical data and the Jonckheere–Terpstra test for continuous data. Conversion factors for units: hemoglobin in g/dL to g/L, ×10; protein in g/dL to g/L, ×10; albumin in g/dL to g/L, ×10; high-sensitivity C-reactive protein in mg/dL to nmol/L, ×9.524; cholesterol in mg/dL to mmol/L, ×0.02586; urea nitrogen in mg/dL to mmol/L, ×0.357; creatinine in mg/dL to mmol/L, × 88.4; phosphate in mg/dL to mmol/L, ×0.3229; calcium in mg/dL to mmol/L. Abbreviations: eGFR, estimated glomerular filtration rate; PTH, parathyroid hormone.
| Weight change since 20 years of age (kg) | |||
|---|---|---|---|
|
T1 ＜−0.1 |
T2 −0.1 to 8.0 |
T3 ≥ 8.0 |
|
| All-cause mortality, n (%) | 129 (13.8) | 56 (6.0) | 58 (6.2) |
| Unadjusted model | 2.65 (1.94–3.63) | 1.00 (reference) | 1.08 (0.75–1.56) |
| Age- and sex-adjusted model | 1.93 (1.40–2.65) | 1.00 (reference) | 1.12 (0.78–1.62) |
| Multivariable-adjusted model | 1.74 (1.24–2.43) | 1.00 (reference) | 1.19 (0.79–1.78) |
Values are presented as HR (95% CIs). The multivariable-adjusted model was adjusted for age; sex; smoking habits; body mass index; systolic blood pressure; history of cardiovascular disease; history of cancer; presence of diabetes; the blood hemoglobin, serum albumin, serum high-sensitivity C-reactive protein, and serum total cholesterol concentrations; eGFR; and urinary albumin-to-creatinine ratio. Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio; T, tertile according to weight change from 20 years of age.
| Weight change since 20 years of age (kg) | |||
|---|---|---|---|
|
Group 1 ＜-5 |
Group 2 -5 to 5 |
Group 3 ≥ 5 |
|
| All-cause mortality, n (%) | 93 (20) | 66 (6.6) | 84 (6.3) |
| PS matching model | 2.05 (1.46–2.88) | 1.00 (reference) | 1.05 (0.73–1.50) |
Values are presented as HRs (95% CIs). The PS for weight change was calculated using a logistic model with the covariates as age, sex, smoking habits, body mass index, systolic blood pressure, history of cardiovascular disease, history of cancer, presence of diabetes, blood hemoglobin, serum albumin, serum high sensitive C-reactive protein, serum total cholesterol, eGFR, and urinary albumin to creatinine ratio. Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazards ratio; PS, propensity score.
In total, 62 participants (2.2%) died from CVD during the observation period, including 26, 18, and 18 in Groups 1, 2, and 3, respectively. Unadjusted Kaplan–Meier curves showed significant differences in the event-free survival probability with respect to cardiovascular mortality among the weight-change groups (log-rank test, P＜0.05; Fig.4).
Survival rates with respect to cardiovascular mortality in the participants, categorized according to weight change, during the 5-year follow-up period
In the Cox proportional-hazards risk models, Group 1 showed a significantly higher HR than Group 2 (reference), even after adjustment for age and sex or for multiple potential confounding factors (Table 4). The HRs (95% CI) were 2.52 (1.35–4.70) for Group 1, 1.00 (reference) for Group 2, and 0.70 (0.33–1.46) for Group 3 in the multivariable-adjusted model. The Fine–Gray proportional-hazards model was used to allow for the influences of competing risks on the association between weight change from 20 years of age and the risk of cardiovascular mortality. When end-stage kidney disease and noncardiovascular mortality were included as competing risks, the risk of cardiovascular mortality in Group 1 was significantly higher (P＜0.01) than that in Group 2 (HR (95% CI): Group 1, 2.25 (1.21–4.19); Group 2, 1.00 (reference); and Group 3, 0.64 (0.33–1.24) (Table 4). Furthermore, the multivariable-adjusted restricted cubic spline curve showed that the HR for cardiovascular mortality gradually increased with the weight loss from 20 years of age, suggesting a nonlinear association (P for nonlinearity ＜0.05; Fig.5).
| Weight change since 20 years of age (kg) | |||
|---|---|---|---|
|
Group 1 ＜−5 |
Group 2 −5 to 5 |
Group 3 ≥ 5 |
|
| Cardiovascular mortality, n (%) | 25 (5.5) | 18 (1.8) | 18 (1.3) |
| Unadjusted model | 3.83 (2.07–7.05) | 1.00 (reference) | 0.76 (0.39–1.47) |
| Age- and sex-adjusted model | 2.56 (1.09–3.31) | 1.00 (reference) | 0.87 (0.45–1.70) |
| Multivariable-adjusted model | 2.48 (1.32–4.64) | 1.00 (reference) | 0.69 (0.33–1.45) |
| Fine & Gray model | 2.16 (1.16–4.03) | 1.00 (reference) | 0.64 (0.33–1.26) |
Values are presented as HRs (95% CIs). The multivariable-adjusted models and Fine & Gray models were adjusted for age; sex; body mass index; smoking habits; the presence of diabetes; systolic blood pressure; a history of cardiovascular events; a history of cancer; the blood hemoglobin, serum albumin, serum high sensitivity C-reactive protein, and serum total cholesterol concentrations; eGFR; and urinary albumin-to-creatinine ratio. Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio.
Association between percentages of weight change from 20 years of age and cardiovascular mortality in patients with CKD, allowing for nonlinear effects, with 95% CIs. The x-axis represents the weight change as a percentage, calculated by dividing the weight change by the weight at age 20 and multiplying by 100. The solid line denotes the HR; gray area, the 95% CI; and the dotted line, an HR of 1.0. The model was fitted with four knots and adjusted for age; sex; the presence of diabetes; a history of cardiovascular disease; a history of cancer; systolic blood pressure; body mass index; blood hemoglobin, serum albumin, serum hs-CRP, and serum total cholesterol concentrations; eGFR; and smoking habits. Curves show the HR compared with the reference weight change of 4.1 kg. Abbreviations: CKD, chronic kidney disease; CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio.
This study characterized the association between long-term weight change and all-cause and cardiovascular mortalities in Japanese patients with CKD. The primary finding of the study was that weight loss from 20 years of age is associated with higher risks of all-cause and cardiovascular mortalities compared with stable weight. The Fine–Gray proportional model showed similar associations between weight loss and the risks of all-cause and cardiovascular mortality. Similar results were obtained in a PS matching analysis conducted as a second sensitivity analysis to minimize possible confounding and selection biases. Furthermore, the association between weight change after the age of 20 and all-cause mortality was stratified according to age; sex; BMI; smoking habits; the presence of diabetes; systolic blood pressure; a history of cardiovascular events and cancer; blood hemoglobin, serum albumin, serum hs-CRP, and serum total cholesterol concentrations; eGFR; and urinary albumin-to-creatinine ratio. No heterogeneity was observed among the subgroups, except for blood hemoglobin concentration. In addition, the restricted cubic spline analysis revealed that the association between weight change from 20 years of age and all-cause and cardiovascular mortalities was nonlinear, with a deteriorating prognosis observed with increasing weight loss. To the best of our knowledge, this is the first study to evaluate the association between long-term weight change and all-cause and cardiovascular mortalities in nondialysis patients with CKD.
Several previous studies have demonstrated that weight loss from young adulthood is also associated with poor prognosis in the general population11, 15, 24); however, the difference in the prognosis of the participants that experienced weight loss and those that maintained their weight in the present study is consistent with the previous findings in the general population, indicating that this trend also exists for patients with CKD. This could be explained by the fact that individuals with CKD are commonly malnourished owing to low food consumption, mainly as a result of anorexia25-27). Low energy intake results in greater catabolism28) and worsening nutritional status. Furthermore, conditions such as sarcopenia, frailty, and protein energy wasting frequently occur in patients with CKD, and their effects are exacerbated by low nutrient intake29, 30). Previous studies have reported that frailty and sarcopenia are associated with cardiovascular31) and all-cause32) mortalities. These findings suggest that inadequate nutritional status, caused by lower dietary intake, in patients with CKD is associated with weight loss and higher risks of cardiovascular and all-cause mortalities.
Moreover, several studies have shown that substantial weight gain from young adulthood is associated with adverse health outcomes11-13, 33-36). However, no significant differences were observed in the all-cause or cardiovascular mortalities between the weight-gain group and the stable-weight group. One possible reason for this discrepancy is the presence of protective nutritional effects of weight gain in patients with CKD37). In these patients, obesity is associated with a superior prognosis, which is referred to as the “obesity paradox”37-40). In addition, the adipose tissue may resist the effects of diseases characterized by catabolism41-44), such as CKD. These protective effects of obesity might antagonize detrimental metabolic effects typically associated with obesity in the general population45-47). As a result, the prognosis of patients with CKD who gained weight from a young age may not significantly differ from that of patients who maintained their weight.
The subgroup analysis revealed that the association between weight change and all-cause mortality was closer in participants with high concentrations of blood hemoglobin. In the baseline characteristics, Group 1 was characterized by older age, more previous CVD, and poorer kidney function than the other groups, but subgroup analyses by median age, previous CVD, and median eGFR revealed no heterogeneity in the association between weight change and all-cause mortality. Patients with CKD are more likely to be anemic due to lower erythropoietin synthesis48) and iron deficiency49), and low hemoglobin levels have been reported to be associated with CKD progression and higher all-cause mortality50). It is speculated that because anemia in itself exerts a substantial influence on life expectancy in patients with CKD, weight change may have a comparatively negligible effect on the prognosis of patients with CKD and anemia, whereas the effect of weight change on prognosis may be more significant in a subgroup of patients with nearly normal hemoglobin levels. It is also possible that patients with low Hb have poorer nutritional status at baseline than those with high Hb, and residual confounding may have affected the results.
The present study had several strengths. First, it was a large prospective cohort study of patients with CKD of a wide range of severities. Second, the follow-up rate in the study was very high, indicating that the results are highly valid and generalizable. Third, the clinical event adjudication committee made outcome judgments based on specific definitions using all the available clinical data obtained during the study using a standardized protocol16). This enhances the accuracy, reliability, and validity of the outcome assessments, thereby improving the generalizability of the findings, which means that they can be used to inform clinical practice.
The study also had several limitations. First, the weight data for the participants at 20 years of age was based on recall, rather than on their medical records; therefore, these data may have been inaccurate and could have resulted in misclassification of the exposure. Second, the baseline data were obtained at a single time point, which may have also led to misclassification of the study participants. Third, we have treated end-stage kidney disease as a censoring event. It is important to acknowledge that patients who transition to end-stage kidney disease remain susceptible to mortality even after this transition. Ideally, post-end-stage kidney disease deaths should have been integrated into our analysis, but, unfortunately, we could not do so. In a future study, we intend to explore continuous follow-up until death for patients with end-stage kidney disease, rather than applying censoring as the approach. Finally, the main exposure was weight change from the age of 20, but due to the differing ages of the participants at the time of enrollment, the time elapsed varied. Nevertheless, the subgroup analysis revealed no apparent heterogeneity with respect to the association between age and weight change. Therefore, this study provides new insights into the association between weight change from a young age and all-cause and cardiovascular mortalities in nondialysis CKD patients.
In conclusion, we have demonstrated that long-term weight loss from the age of 20 is associated with higher risks of all-cause and cardiovascular mortalities in Japanese patients with CKD. Further studies are warranted to determine the underlying mechanisms and to develop interventions to prevent weight loss in this population.
None declared.
This work was supported by a Grant-in-Aid for Scientific Research [No. 15H06800] from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
We would like to express our sincere thanks to the participants in the FKR Study, the members of the FKR Study Group, and all the personnel at participating institutions that were involved in the study. The following personnel (institutions) participated in the study: Satoru Fujimi (Fukuoka Renal Clinic), Hideki Hirakata (Fukuoka Renal Clinic), Tadashi Hirano (Hakujyuji Hospital), Tetsuhiko Yoshida (Hamanomachi Hospital), Takashi Deguchi (Hamanomachi Hospital), Hideki Yotsueda (Harasanshin Hospital), Kiichiro Fujisaki (Iizuka Hospital), Keita Takae (Japanese Red Cross Fukuoka Hospital), Koji Mitsuiki (Hara Sanshin Hospital), Akinori Nagashima (Japanese Red Cross Karatsu Hospital), Ritsuko Katafuchi (Kano Hospital), Hidetoshi Kanai (Kokura Memorial Hospital), Kenji Harada (Kokura Memorial Hospital), Tohru Mizumasa (Kyushu Central Hospital), Takanari Kitazono (Kyushu University), Toshiaki Nakano (Kyushu University), Toshiharu Ninomiya (Kyushu University), Kumiko Torisu (Kyushu University), Akihiro Tsuchimoto (Kyushu University), Shunsuke Yamada (Kyushu University), Hiroto Hiyamuta (Kyushu University), Shigeru Tanaka (Kyushu University), Dai Matsuo (Munakata Medical Association Hospital), Yusuke Kuroki (National Fukuoka-Higashi Medical Centre), Hiroshi Nagae (National Fukuoka-Higashi Medical Centre), Masaru Nakayama (National Kyushu Medical Centre), Kazuhiko Tsuruya (Nara Medical University), Masaharu Nagata (Shin-eikai Hospital), Taihei Yanagida (Steel Memorial Yawata Hospital), and Shotaro Ohnaka (Tagawa Municipal Hospital). We also thank Mark Cleasby, PhD, from Edanz Group (https://jp.edanz.com/ac) for editing a draft of this manuscript.
