Abstract
Changes over time have been shown in renal manifestations in individuals with diabetes in the United States; however, whether the trends are shared across ethnicities is unknown. We conducted this single-center serial cross-sectional study to determine temporal changes in albuminuria and reduced kidney function in Japanese patients with type 2 diabetes. This study included adult Japanese patients with type 2 diabetes who first visited our institute between 2004 and 2013. Temporal changes during the 10 years in the frequency of albuminuria ( ≥ 30 mg/g creatinine) and reduced eGFR ( < 60 mL/min/1.73 m2) were analyzed using the univariate and multivariate logistic regression analyses and Granger causality test. 5331 Japanese patients with type 2 diabetes, 1892 women and 3439 men, with the mean age of 56 ± 13 years, were studied. There was no change in the prevalence of albuminuria in the univariate analysis; however, a significantly decreasing trend was observed after adjustment for several covariates. On the other hand, patients with reduced eGFR significantly increased over time, although the statistical significance disappeared after adjustment for the covariates, including levels of serum uric acid and hemoglobin and use of renin–angiotensin inhibitors. The Granger causality test showed that time series for use of RAS inhibitors and BMI had a causative role in time series for reduced eGFR. In conclusion, prevalence of albuminuria decreased and that of reduced eGFR remained stable after adjustment for clinical characteristics in Japanese patients with type 2 diabetes during the last decade.
Keywords: Albuminuria, eGFR, Type 2 diabetes, Granger causality
Introduction
Nephropathy is one of the most serious complications of both type 1 and type 2 diabetes. Previous studies have shown that the natural course of nephropathy differs between type 1 and type 2 diabetes mellitus [1–10]. In type 1 diabetes, urinary albumin excretion first increases and subsequently glomerular filtration rate (GFR) starts to decrease [5]. In type 2 diabetes, on the other hand, a considerable number of patients already exhibit reduced GFR at the diagnosis of type 2 diabetes [6]. Longitudinal and cross-sectional studies [6–10], including ours, [4, 9] have found that reduction of GFR precedes the onset of albuminuria in patients with type 2 diabetes.
Recently, the National Health and Nutrition Examination Survey in the United States (NHANES) has shown an increasing trend in the prevalence of reduced GFR and a decreasing trend in the prevalence of albuminuria among overall adult patients with type 2 diabetes in the United States during the last 26 years [11, 12]. However, significant heterogeneities were observed among ethnicities, remaining unclear whether the temporal changes in the phenotype of renal parameters are shared beyond ethnicities [12, 13]. In addition, there was no change in the prevalence of albuminuria among elderly people [12].We, therefore, conducted this single-center serial cross-sectional study to explore whether the renal phenotype of Japanese patients with type 2 diabetes has similarly changed. We also aimed to determine clinical characteristics and patient background associated with the temporal changes.
Patients and methods
Subjects
This was a single-center serial cross-sectional study, including Japanese patients who (1) had type 2 diabetes (2) were ≥ 20 years of age (3) first visited the Diabetes Center, Tokyo Women’s Medical University, Tokyo Japan, during the period between 2004 and 2013 (4) had serum creatinine and urinary albumin levels measured within 3 months after the initial presentation (5) exhibited an estimated GFR (eGFR) ≥ 15 mL/min/1.73 m2, and (6) underwent an assessment of the presence and stage of diabetic retinopathy by ophthalmologists. The period from the diagnosis of diabetes and the first visit to the hospital was not restricted. Patients who were treated with dialysis, had undergone kidney transplantation, or were pregnant were excluded.
The present study was conducted as a part of the Prospective Cohort Study Elucidating Factors Associated with the Pathogenesis, Prognosis, and Prognosis of Diabetic Nephropathy conducted in Diabetes Center, Tokyo Women’s Medical University, the protocol of which was approved on April 25, 2009 by the Ethics Committee of Tokyo Women’s Medical University School of Medicine (Approval No. 1584). Since this was an observational but not a prospective intervention study, the Ethics Committee provided a waiver of informed consent.
Measurements
Urinary albumin was measured in the first morning urine with the immunoturbidimetry method, serum and urinary creatinine levels were measured using the enzymatic method, LDL and HDL cholesterol were measured by the homogenous assays, triglycerides were measured by the enzymatic colorimetric method, uric acid was measured by the uricase method; these measurements were performed using an autoanalyzer (Hitachi Labospect 008, Hitachi, Tokyo). Glycated hemoglobin (HbA1c) was measured by the HPLC using a fully automated HbA1c analyzer (ADAMS A1c HA-8180, Arkray, Inc., Kyoto). Urinary albumin was divided by urinary creatinine to calculate the urinary albumin-to-creatinine ratio (UACR).GFR was estimated (eGFR) using the formula proposed by the Japanese Society for Nephrology [14]. Clinical findings and laboratory data were collected from the hospital medical records.
The geometric mean levels of urinary albumin, arithmetic mean of eGFR, and frequencies of patients with albuminuria, defined as UACR ≥ 30 mg/g, and reduced kidney function, defined as eGFR < 60 mL/min/1.73 m2 were calculated for each year.
Statistical analysis
Data were expressed as mean ± standard deviation (SD), geometric mean, or frequency. Changes over time in the frequency of albuminuria, reduced kidney function, and other binary data were examined using (1) the single regression analysis with the first-visiting calendar year treated as an independent continuous variable, (2) the Cochran–Armitage trend test, and (3) univariate logistic regression analysis. Changes over time in continuous variables were examined using the Jonckheere–Terpstra trend test.
Trends in the frequency of the renal parameters were also examined using the multivariate logistic regression analysis, after adjustment for age, sex, diabetes duration, smoking history, presence of retinopathy, history of coronary artery disease and stroke, use of renin–angiotensin system (RAS) inhibitors, body mass index (BMI), systolic blood pressure, diastolic blood pressure, hemoglobin, HbA1c, serum levels of low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, logarithmically transformed triglycerides, and uric acid, without variable-selection procedures. Continuous variables were standardized by converting the data to yield mean of 0 and SD of 1. Strengths of contribution among the variables in the multivariate logistic regression analysis were compared by the Wald Chi-square test.
As a sensitivity analysis, logistic regression analysis after adjustment for propensity score was also performed. The propensity score in each patient was calculated using the above 17 covariates and incorporated in the logistic model as a solitary covariate.
Causality between time-series data was examined using the Granger causality test, which was originally developed to analyze econometric data as a method for detecting causal correlations between time-series data [15]. This test has been recently used widely in medical research of the field of neurology, particularly neural networks [16]. In the present study, the null hypothesis verified by the Granger causality test was that the time-series data for the frequency of albuminuria or reduced eGFR (X) was influenced by X itself, and not by other time-series data (Y) such as changes in age. Rejection of this null hypothesis suggests that changes in X over time are affected by the time-series data on Y and that including Y time-series data may enable more accurate predictions of how the frequency X will change over time.
SAS/STAT 14.3 (SAS Institute, Inc., Cary, NC) was used for the statistical analyses.
Results
We studied a total of 5331 Japanese patients with type 2 diabetes, 1892 women and 3439 men, with the mean age of 56 ± 13 years, ranging 20–90 years, who met the above criteria. As shown in Table 1, no significant trend was observed in age or sex over the 10-year period. The number of patients taking anti-hypertensive drugs and RAS inhibitors significantly increased over time, with the rate of annual increase of 1.3% and 1.8%, respectively. No significant trends were observed in the frequency of patients with a history of coronary artery disease or stroke. There was a significantly increasing trend in patients having retinopathy, with the increase rate of 1.0% per year. Significantly increasing trends were also observed for the mean levels of BMI, uric acid, LDL cholesterol, and HDL cholesterol. A significantly decreasing trend was observed for diastolic blood pressure.
Table 1.
Temporal trends in clinical characteristics and laboratory data during 10 years
| First-visit year | P value | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2004 (N = 632) | 2005 (N = 664) | 2006 (N = 609) | 2007 (N = 508) | 2008 (N = 556) | 2009 (N = 490) | 2010 (N = 520) | 2011 (N = 485) | 2012 (N = 430) | 2013 (N = 437) | ||
| Age (years) | 55.4 | 56.3 | 57.0 | 55.5 | 55.3 | 55.6 | 57.0 | 56.1 | 57.3 | 55.8 | 0.394 |
| Women (%) | 36.9 | 36.5 | 35.1 | 35.2 | 32.9 | 37.8 | 34.2 | 34.4 | 34.9 | 36.8 | 0.642 |
| Diabetes duration (years) | 6.0 | 6.4 | 6.7 | 5.5 | 6.8 | 7.2 | 6.9 | 7.7 | 6.7 | 7.0 | 0.051 |
| Current smoker (%) | 54.9 | 53.5 | 54.6 | 56.0 | 57.8 | 52.0 | 54.3 | 54.0 | 48.6 | 52.9 | 0.154 |
| Hypertension (%) | 59.0 | 64.6 | 62.2 | 74.1 | 66.2 | 63.9 | 64.6 | 61.4 | 63.1 | 65.6 | 0.438 |
| Use of anti-hypertensive agents (%) | 30.1 | 30.5 | 30.1 | 31.3 | 36.2 | 36.3 | 41.5 | 37.5 | 40.2 | 38.5 | < 0.001 |
| Use of RA system blockers (%) | 16.2 | 18.6 | 17.5 | 18.5 | 25.7 | 25.8 | 29.9 | 30.2 | 27.4 | 31.7 | < 0.001 |
| History of cardiovascular diseases (%) | 8.5 | 12.6 | 15.1 | 9.8 | 12.5 | 10.6 | 10.1 | 8.7 | 13.5 | 11.4 | 0.917 |
| History of coronary heart diseases (%) | 5.7 | 6.4 | 9.5 | 6.2 | 7.0 | 5.0 | 6.7 | 3.2 | 8.3 | 6.9 | 0.619 |
| History of stroke (%) | 4.1 | 6.6 | 6.1 | 4.0 | 6.8 | 6.5 | 3.8 | 6.1 | 6.0 | 5.5 | 0.795 |
| Retinopathy (%) | 28.8 | 31.6 | 31.4 | 30.9 | 39.2 | 37.1 | 41.0 | 40.6 | 33.5 | 38.2 | < 0.001 |
| Systolic blood pressure (mmHg) | 144 | 142 | 142 | 142 | 143 | 142 | 141 | 143 | 142 | 142 | 0.328 |
| Diastolic blood pressure (mmHg) | 82 | 83 | 82 | 82 | 83 | 82 | 82 | 80 | 81 | 81 | < 0.001 |
| Body mass index (kg/m2) | 24.9 | 25.2 | 25.4 | 25.5 | 25.4 | 25.8 | 25.6 | 25.9 | 25.7 | 25.7 | < 0.001 |
| HbA1c (%) | 8.6 | 8.7 | 8.6 | 8.4 | 8.6 | 8.5 | 8.4 | 8.4 | 8.6 | 8.7 | 0.487 |
| HbA1c (mmol/mol) | 70 | 71 | 70 | 68 | 70 | 69 | 68 | 68 | 70 | 71 | 0.487 |
| Hemoglobin (g/dL) | 14.4 | 14.3 | 14.3 | 14.3 | 14.3 | 14.2 | 14.1 | 14.4 | 14.4 | 14.4 | 0.807 |
| Uric acid (mg/dL) | 5.1 | 5.1 | 5.2 | 5.2 | 5.4 | 5.2 | 5.3 | 5.3 | 5.3 | 5.5 | < 0.001 |
| LDL cholesterol (mg/dL) | 118 | 124 | 121 | 123 | 120 | 119 | 121 | 123 | 125 | 127 | 0.014 |
| HDL cholesterol (mg/dL) | 51 | 51 | 52 | 52 | 52 | 51 | 51 | 52 | 55 | 54 | < 0.001 |
| Triglycerides (mg/dL) | 143 | 135 | 136 | 137 | 137 | 142 | 140 | 137 | 137 | 148 | 0.702 |
| Urinary ACR (mg/g) | 22.9 | 22.0 | 21.1 | 20.8 | 25.9 | 21.2 | 22.0 | 16.8 | 14.3 | 15.5 | < 0.001 |
| Creatinine (mg/dL) | 0.77 | 0.76 | 1.78 | 0.77 | 0.81 | 0.79 | 0.84 | 0.79 | 0.79 | 0.85 | 0.002 |
| eGFR (mL/min/1.73 m2) | 81.0 | 81.7 | 79.8 | 82.1 | 80.5 | 81.2 | 77.1 | 82.9 | 79.4 | 75.0 | 0.001 |
Continuous data were expressed as mean or geometric mean (for triglycerides and urinary ACR) and analyzed using Jonckheere–Terpstra trend test. Binary data were shown as percent and analyzed using Cochran–Armitage trend tests
RA renin–angiotensin, HbA1c hemoglobin A1c, LDL low-density lipoprotein, HDL high-density lipoprotein, eGFR estimated glomerular filtration rate, ACR albumin-to-creatinine ratio
The geometric mean levels of urinary ACR exhibited a significantly decreasing trend over the 10-year period when examined using the Jonckheere–Terpstra test (Table 1; p < 0.001). When albuminuria was treated categorically (binary data), no significantly increasing trend was observed in the frequency of patients with albuminuria in the single regression analysis (Fig. 1a: p = 0.956) or the Cochran–Armitage trend test (p = 0.981). In the univariate logistic regression analysis, the first-visiting calendar year was not associated with increased prevalence of albuminuria; however, after adjustment for other clinical data in the multivariate analysis, the first visit year became statistically significant with an odds ratio (OR) of 0.966 per year (Table 2, p = 0.029). The logistic regression analysis after adjustment for the propensity score that was calculated using the 17 covariates yielded the identical association between calendar year and prevalence of albuminuria (Table 2).
Fig. 1.
Serial changes in the frequency of patients with albuminuria ( ≥ 30 mL/min/1.73 m2, left panel) and reduced eGFR ( < 60 mL/min/1.73 m2, right panel) in patients with type 2 diabetes who first visited between 2004 and 2013. Shaded area indicates 95% confidence interval of the regression line and dashed lines indicate lower and upper limit of 95% predictive interval
Table 2.
Univariate and multivariate logistic regression analysis examining the associations of prevalence of albuminuria and reduced eGFR with calendar year
| Variable | Albuminuria | Wald Chi-square | Reduced eGFR | Wald Chi-square | |||||
|---|---|---|---|---|---|---|---|---|---|
| Odds ratio | 95% CI | p value | Odds ratio | 95% CI | p value | ||||
| Univariate | |||||||||
| First-visit year | 1.000 | 0.980–1.021 | 0.981 | 0.0 | 1.050 | 1.025–1.077 | < 0.001 | 15.5 | |
| Multivariate | |||||||||
| First-visit year | 0.966 | 0.936–0.996 | 0.029 | 4.8 | 1.030 | 0.986–1.075 | 0.185 | 0.2 | |
| Age | 1.053 | 0.937–1.184 | 0.385 | 0.8 | 3.281 | 2.727–3.948 | < 0.001 | 158.6 | |
| Male gender | 1.361 | 1.071–1.730 | 0.012 | 6.4 | 1.672 | 1.211–2.308 | 0.002 | 9.8 | |
| Diabetes duration | 1.090 | 0.991–1.199 | 0.075 | 3.2 | 1.075 | 0.955–1.209 | 0.232 | 1.4 | |
| Smoking | 1.433 | 1.171–1.754 | 0.001 | 12.2 | 0.825 | 0.628–1.085 | 0.167 | 1.9 | |
| Retinopathy | 3.801 | 3.137–4.605 | < 0.001 | 185.9 | 1.694 | 1.282–2.221 | < 0.001 | 14.6 | |
| BMI | 1.143 | 1.031–1.267 | 0.012 | 6.5 | 1.351 | 1.141–1.516 | < 0.001 | 14.3 | |
| Use of RAS inhibitors | 1.420 | 1.150–1.754 | 0.001 | 10.6 | 1.874 | 1.453–2.416 | < 0.001 | 23.4 | |
| SBP | 1.502 | 1.320–1.709 | < 0.001 | 38.0 | 0.976 | 0.831–1.148 | 0.772 | 0.1 | |
| DBP | 0.905 | 0.796–1.028 | 0.125 | 2.4 | 1.208 | 1.021–1.419 | 0.028 | 4.9 | |
| Hemoglobin | 0.641 | 0.573–0.718 | < 0.001 | 60.0 | 0.436 | 0.376–0.507 | < 0.001 | 118.5 | |
| HbA1c | 1.321 | 1.200–1.455 | < 0.001 | 32.0 | 0.920 | 0.798–1.061 | 0.253 | 1.3 | |
| Uric acid | 1.262 | 1.140–1.397 | < 0.001 | 20.2 | 2.623 | 2.268–3.033 | < 0.001 | 168.8 | |
| Triglycerides | 1.268 | 1.142–1.408 | < 0.001 | 19.9 | 1.386 | 1.197–1.605 | < 0.001 | 19.0 | |
| PS-adjusted | |||||||||
| First-visit year | 0.966 | 0.938–0.996 | 0.028 | 4.9 | 1.029 | 0.987–1.073 | 0.159 | 1.78 | |
BMI body mass index, RAS renin–angiotensin system, SBP systolic blood pressure, DBP diastolic blood pressure, HbA1c hemoglobin A1c, CI confidence interval, AUROC area under the receiver operating characteristic curve, PS propensity score
Mean levels of eGFR decreased significantly over the 10-year period (Table 1; Jonckheere–Terpstra test for trend: p = 0.001), with the number of patients with eGFR < 60 mL/min/1.73 m2 increasing significantly (p = 0.005; Cochran–Armitage test for trend: p < 0.001; univariate logistic regression analysis: OR, 1.05; p < 0.001). However, the statistical significance for the first-visit year disappeared after adjustment for the covariates as well as for the propensity scores (Table 2).
Other variables significantly associated with the presence of albuminuria and with reduced eGFR are listed in Table 2. According to the Wald Chi-square, the following factors had higher predictability for albuminuria in descending order: presence of retinopathy, hemoglobin level, systolic blood pressure, and HbA1c. The factors with higher predictability for reduced eGFR were as follows in descending order: serum uric acid level, age, hemoglobin, and use of RAS inhibitors.
Finally, the Granger causality test was used to examine the time-series data that affected the time-series data on the frequency of patients with albuminuria and reduced kidney function. As shown in Table 3, only the time series for age significantly impacted the changes in albuminuria over time. The time series for use of RAS inhibitors and BMI was factors that influenced the time-series data on reduced eGFR. Of these, BMI and reduced eGFR had a bidirectional relationship. All other factors were one-way causalities.
Table 3.
Granger causality test examining the association of temporal trend in clinical characteristics and laboratory data with temporal tend in albuminuria and reduced eGFR
| Temporal trend in X | Temporal trend in albuminuria (Y) | Temporal trend in reduced eGFR (Y) | ||||||
|---|---|---|---|---|---|---|---|---|
| From X to Y* | From Y to X* | From X to Y* | From Y to X* | |||||
| Chi-square | p value | Chi-square | p value | Chi-square | p value | Chi-square | p value | |
| Age | 3.90 | 0.048 | 0.84 | 0.359 | 0.15 | 0.697 | 0.01 | 0.925 |
| Sex | 0.25 | 0.618 | 0.55 | 0.457 | 0.47 | 0.494 | 0.11 | 0.737 |
| Diabetes duration | 2.66 | 0.103 | 1.22 | 0.270 | 0.01 | 0.937 | 3.80 | 0.051 |
| Smoking | 0.13 | 0.720 | 1.72 | 0.189 | 1.84 | 0.175 | 2.01 | 0.157 |
| Use of RAS blockers | 0.01 | 0.930 | 2.67 | 0.102 | 8.63 | 0.003 | 0.12 | 0.733 |
| Coronary heart disease | 0.27 | 0.605 | 1.85 | 0.174 | 0.90 | 0.342 | 5.10 | 0.024 |
| Stroke | 0.78 | 0.378 | 0.14 | 0.713 | 0.04 | 0.837 | 0.07 | 0.796 |
| Retinopathy | 0.03 | 0.860 | 0.62 | 0.431 | 0.62 | 0.433 | 0.19 | 0.663 |
| SBP | 0.29 | 0.590 | 0.06 | 0.807 | 0.79 | 0.373 | 0.00 | 0.966 |
| DBP | 1.61 | 0.204 | 3.43 | 0.064 | 0.48 | 0.489 | 6.71 | 0.010 |
| BMI | 0.02 | 0.900 | 5.12 | 0.024 | 7.62 | 0.006 | 34.54 | < 0.001 |
| HbA1c | 0.19 | 0.666 | 4.05 | 0.044 | 0.85 | 0.356 | 0.91 | 0.341 |
| Hemoglobin | 0.27 | 0.605 | 0.47 | 0.493 | 0.00 | 0.980 | 0.30 | 0.582 |
| Uric acid | 0.00 | 0.948 | 8.39 | 0.004 | 1.09 | 0.295 | 0.57 | 0.449 |
| LDL cholesterol | 0.56 | 0.454 | 0.42 | 0.518 | 1.43 | 0.232 | 0.64 | 0.425 |
| HDL cholesterol | 0.00 | 0.976 | 0.07 | 0.790 | 1.23 | 0.267 | 0.18 | 0.671 |
| Triglycerides | 1.82 | 0.177 | 0.10 | 0.751 | 0.01 | 0.941 | 1.80 | 0.180 |
The direction from X to Y indicates that the temporal trend in X, for example age, influences the temporal trend in albuminuria or reduced eGFR during the 10 years more precisely and significantly than only X itself. The direction from Y to X examined the inverse causality
Discussion
This single-center serial cross-sectional study aimed to determine whether the frequency of albuminuria and reduced kidney function changed significantly during the last 10 years in Japanese patients with type 2 diabetes. The univariate analysis showed that the frequency of patients with albuminuria did not change; however, the multivariate analysis yielded a significant decreasing trend. In contrast, while the apparent frequency of patients exhibiting reduced kidney function (eGFR < 60 mL/min/1.73 m2) increased significantly, the significance disappeared in the multivariate analysis. In the Granger causality test, which was used to infer causality between time series, there was a significant causality between the time series on albuminuria and age and between the time series on reduced kidney function and the use of RAS inhibitors and BMI.
Our findings appear to be consistent with epidemiological studies from the NHANES [11, 12], as well as a single-center study reported most recently from Japan [13]. However, in the study from the NHANES, the prevalence of reduced eGFR was only adjusted for age, sex, and ethnicity but not for the use of RAS inhibitors [12]. The Japanese group has simply examined covariation showing only crude incidence of renal parameters over time, lacking considerations of the correlation or association between the trends [13]. Although both of these authors speculated that decreased incidence of albuminuria and increased incidence of reduced kidney function were associated with increased incidence of use of the RA inhibitors [12, 13], the associations were not examined by appropriate statistical analyses. In our study, the associations were examined by multiple logistic regression analysis and Granger causality test. Pharmacological blockade of the RAS by angiotensin-converting enzyme (ACE) inhibitors and angiotensin II receptor blockers acutely reduce intraglomerular pressure and GFR by dilating the afferent arterioles of the renal glomeruli [17, 18]. Relatively recent guidelines for hypertension have recommended RAS inhibitors as the first-line therapy for hypertension in patients with diabetes regardless of albuminuria [19–21], promoting physicians to prescribe RAS inhibitors more preferentially. This may help explain, at least in part, why the prevalence of reduced eGFR recently increased significantly. In our study, however, use of RAS inhibitors was also associated with increased prevalence of albuminuria. These seemingly paradoxical results, frequently observed in observational or cross-sectional studies, may be attributed, at least in part, to reverse causation bias [22] due to the preferential use of RAS inhibitors especially in patients with diabetic nephropathy. Another possibility that use of RAS inhibitors, especially ACE inhibitors, is rather associated with increased risk of progression of nephropathy [23, 24] may not be completely denied.
Beside the use of RAS inhibitors, higher serum levels of uric acid, older age, and lower hemoglobin were significantly associated with reduced kidney function. Hyperuricemia is known to have a bidirectional relationship with chronic kidney diseases including diabetic nephropathy [25, 26]. Renal clearance of uric acid decreases as renal function reduces, resulting in elevation of serum uric acid levels. In addition, hyperuricemia is a well known risk factor for decline in kidney function. We have recently showed that, in Japanese patients with type 2 diabetes and GFR ≥ 60 mL/min/1.73 m2, higher serum levels of uric acid were significantly associated with increased risk of 30% and 50% reductions in eGFR, as well as progression to end-stage kidney disease [27]. The slightly but significantly increasing trend of serum levels of uric acid in the current study may have contributed to the increase in patients with an eGFR < 60 mL/min/1.73 m2 over time. The Japanese study showed that age-dependent increase in arterial stiffness was associated with reduced eGFR [13], although independent effects of age and increased stiffness with albuminuria were not determined.
Close associations of higher age and lower levels of hemoglobin or anemia [28, 29] with reduced renal function have been well demonstrated. Most of covariates significantly associated with albuminuria, including presence of retinopathy, lower hemoglobin level [28, 29], greater systolic blood pressure, and higher levels of HbA1c are well known risk factors for development and progression of albuminuria [30–32]. Nonetheless, the fundamental factors associated with the decrease in the prevalence of albuminuria were not clarified in this study.
Finally, the Granger causality test is a unique statistical hypothesis test used to infer whether a time series is useful for predicting another time series [15]. Using this test, we found that changes in the use of RAS inhibitors and BMI showed a significant ability to predict changes over time in the frequency of reduced eGFR. Only the use of RAS inhibitors was consistent with the results of the multivariate logistic regression analysis, strongly suggesting that the recent increase in RAS inhibitor use was closely associated with the increase in patients with reduced kidney function. The reasons for the discrepancy between the results of the logistic regression analysis and the Granger causality test are unknown but it may be due to the differences in the predictive performance between these statistical analyses.
This study is subject to the attendant limitations. First, the serial cross-sectional study does not allow to assess causal relationships. The Granger causality test was used to complement the limitation, although the test cannot verify close causality, only demonstrating strong associations between two time-series data. Second, while the NHANES and Japanese study covered 27 years [12] and 18 year [13], respectively, our study is based on a shorter (10-year) observation. Third, we studied patients who visited an urban university hospital, likely limiting the generalizability of the results. Fourth, treatment duration of the RAS inhibitors before the study was not assessed. Finally, UACR and serum creatinine were measured on a single occasion, possibly leading to improper categorization of albuminuria and GFR. Nonetheless, we believe this did not affect the temporal trend as this limitation was non-differential throughout the study period.
This study also has strengths. It represents a large single-center study including individuals who were definitely diagnosed as type 2 diabetes by specialist of diabetes. UACR and serum creatinine were measured at a single laboratory, eliminating inter-institutional variance. In addition, we restricted timing of urine collection to first morning, which allowed minimizing the variation of albumin excretion due to exercise and diurnal fluctuations [33].
Conclusions
This large serial cross-sectional study of Japanese patients with type 2 diabetes has shown that the frequency of albuminuria did not change, while the frequency of reduced kidney function increased. However, after adjustment for clinical findings including frequency of the use of RAS inhibitors, a modest but statistically significant decrease in the frequency of albuminuria was observed, whereas the frequency of reduced kidney function remained stable. Taken with the results of the Granger causality test together, changes in clinical and environmental factors may have led to the apparent changes in the renal manifestations over the last decade. The causal nature of the environmental exposures needs to be inferred in future studies, using more sophisticated methods, including Mendelian randomization [34].
Author contributions
NT designed the study, collected the data, performed the statistical analyses, researched data, and wrote the manuscript. YYa, YYo, TM, and KH substantially contributed to collection and interpretation of data. TB is the guarantor of this work, had full access to all the data in the study, performed the statistical analyses in part, reviewed/edited the manuscript, and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Compliance with ethical standards
Conflicts of interest
The authors have nothing to declare.
Human rights statement
The present study was conducted as a part of the Prospective Cohort Study Elucidating Factors Associated with the Pathogenesis, Prognosis, and Prognosis of Diabetic Nephropathy conducted in Diabetes Center, Tokyo Women’s Medical University, the protocol of which was approved by the Ethics Committee of Tokyo Women’s Medical University School of Medicine (Approval No. 1584)
Informed consent
Since this was an observational but not a prospective intervention study, the Ethics Committee provided a waiver of informed consent.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Hasslacher CH, Ritz E, Wahl P, et al. Similar risks of nephropathy in patients with type I or type II diabetes mellitus. Nephrol Dial Transpl. 1989;4:859–863. doi: 10.1093/ndt/4.10.859. [DOI] [PubMed] [Google Scholar]
- 2.Yokoyama H, Okudaira M, Otani T, et al. Higher incidence of diabetic nephropathy in type 2 than in type 1 diabetes in early-onset diabetes in Japan. Kidney Int. 2000;58:302–311. doi: 10.1046/j.1523-1755.2000.00166.x. [DOI] [PubMed] [Google Scholar]
- 3.Pugh JA, Medina R, Ramirez M. Comparison of the course to end-stage renal disease of type 1 (insulin-dependent) and type 2 (non-insulin-dependent) diabetic nephropathy. Diabetologia. 1993;36:1094–1098. doi: 10.1007/BF02374504. [DOI] [PubMed] [Google Scholar]
- 4.Ohta M, Babazono T, Uchigata Y, et al. Comparison of the prevalence of chronic kidney disease in Japanese patients with type 1 and type 2 diabetes. Diabet Med. 2010;27:1017–1023. doi: 10.1111/j.1464-5491.2010.03049.x. [DOI] [PubMed] [Google Scholar]
- 5.Mongensen CE, Christensen CK, Vittinghus E. The stages in diabetic renal disease. With emphasis on the stage of incipient diabetic nephropathy. Diabetes. 1983;32(Suppl 2):64–78. doi: 10.2337/diab.32.2.S64. [DOI] [PubMed] [Google Scholar]
- 6.Retnakaran R, Cull CA, Thorne KI, et al. Risk factors for renal dysfunction in type 2 diabetes: U.K. Prospective diabetes study 74. Diabetes. 2006;55:1832–1838. doi: 10.2337/db05-1620. [DOI] [PubMed] [Google Scholar]
- 7.Yokoyama H, Sone H, Oishi M, et al. Prevalence of albuminuria and renal insufficiency and associated clinical factors in type 2 diabetes: the Japan diabetes clinical data management study (JDDM 15) Nephrol Dial Transpl. 2009;24:1212–1219. doi: 10.1093/ndt/gfn603. [DOI] [PubMed] [Google Scholar]
- 8.Kramer CK, Leitão CB, Pinto LC, et al. Clinical and laboratory profile of patients with type 2 diabetes with low glomerular filtration rate and normoalbuminuria. Diabetes Care. 2007;30:1998–2000. doi: 10.2337/dc07-0387. [DOI] [PubMed] [Google Scholar]
- 9.Takagi M, Babazono T, Uchigata Y. Differences in risk factors for the onset of albuminuria and decrease in glomerular filtration rate in people with Type 2 diabetes mellitus: implications for the pathogenesis of diabetic kidney disease. Diabet Med. 2015;32:1354–1360. doi: 10.1111/dme.12793. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kramer HJ, Nguyen QD, Curhan G, et al. Renal insufficiency in the absence of albuminuria and retinopathy among adults with type 2 diabetes mellitus. JAMA. 2003;289:3273–3277. doi: 10.1001/jama.289.24.3273. [DOI] [PubMed] [Google Scholar]
- 11.de Boer IH, Rue TC, Hall YN, et al. Temporal trends in the prevalence of diabetic kidney disease in the United States. JAMA. 2011;305:2532–2539. doi: 10.1001/jama.2011.861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Afkarian M, Zelnick LR, Hall YN, et al. Clinical manifestations of kidney disease among US adults with diabetes, 1988–2014. JAMA. 2016;316:602–610. doi: 10.1001/jama.2016.10924. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kume S, Araki S, Ugi S, et al. Secular changes in manifestations of kidney disease among Japanese adults with type 2 diabetes, 1996–2014. J Diabetes Investig. 2018 doi: 10.1111/jdi.12977. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Matsuo S, Imai E, Horio M, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009;53:982–992. doi: 10.1053/j.ajkd.2008.12.034. [DOI] [PubMed] [Google Scholar]
- 15.Granger CWJ. Investigation causal relations by econometric models and cross-spectral methods. Econometrica. 1969;37:424–438. doi: 10.2307/1912791. [DOI] [Google Scholar]
- 16.Stokes PA, Purdon PL. A study of problems encountered in granger causality analysis from a neuroscience perspective. Proc Natl Acad Sci USA. 2017;114:E7063–E7072. doi: 10.1073/pnas.1704663114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Anderson S, Meier TW, Rennke HG, Brenner BM. Control of glomerular hypertension limits glomerular injury in rats with reduced renal mass. J Clin Invest. 1986;76:612–619. doi: 10.1172/JCI112013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Anderson S, Rennke HG, Brenner BM. Therapeutic advantage of converting enzyme inhibitors in arresting progressive renal disease associated with systemic hypertension in the rat. J Clin Invest. 1986;77:1993–2000. doi: 10.1172/JCI112528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.tandards of medical care in diabetes–2015 summary of revisions. Diabetes Care. 2015;38(Suppl 1):S4. doi: 10.2337/dc15-S003. [DOI] [PubMed] [Google Scholar]
- 20.Weber MA, Schiffrin EL, White WB, et al. Clinical practice guidelines for the management of hypertension in the community a statement by the American Society of Hypertension and the International Society of Hypertension. J Hypertens. 2014;32:3–15. doi: 10.1097/HJH.0000000000000065. [DOI] [PubMed] [Google Scholar]
- 21.Shimamoto K, Ando K, Fujita T, et al. The Japanese Society of Hypertension guidelines for the management of hypertension (JSH 2014) Hypertens Res. 2014;37:253–390. doi: 10.1038/hr.2013.80. [DOI] [PubMed] [Google Scholar]
- 22.MacMahon S, Collins R. Reliable assessment of the effects of treatment on mortality and major morbidity, II: observational studies. Lancet. 2001;357:455–462. doi: 10.1016/S0140-6736(00)04017-4. [DOI] [PubMed] [Google Scholar]
- 23.Suissa S, Hutchinson T, Brophy JM, et al. ACE-inhibitor use and the long-term risk of renal failure in diabetes. Kidney Int. 2006;69:913–919. doi: 10.1038/sj.ki.5000159. [DOI] [PubMed] [Google Scholar]
- 24.Schmidt M, Mansfield KE, Bhaskaran K, et al. Serum creatinine elevation after renin-angiotensin system blockade and long term cardiorenal risks: cohort study. BMJ. 2017;356:j791. doi: 10.1136/bmj.j791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Edwards NL. The role of hyperuricemia and gout in kidney and cardiovascular disease. Cleve Clin J Med. 2008;75:S13–S16. doi: 10.3949/ccjm.75.Suppl_5.S13. [DOI] [PubMed] [Google Scholar]
- 26.Zhu P, Liu Y, Han L, et al. Serum uric acid is associated with incident chronic kidney disease in middle-aged populations: a meta-analysis of 15 cohort studies. PLoS ONE. 2014;9:e100801. doi: 10.1371/journal.pone.0100801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hanai K, Tauchi E, Nishiwaki Y, et al. Effects of uric acid on kidney function decline differ depending on baseline kidney function in type 2 diabetic patients. Nephrol Dial Transplant. 2018 doi: 10.1093/ndt/gfy138. [DOI] [PubMed] [Google Scholar]
- 28.Babazono T, Hanai K, Suzuki K, et al. Lower haemoglobin level and subsequent decline in kidney function in type 2 diabetic adults without clinical albuminuria. Diabetologia. 2006;49:1387–1393. doi: 10.1007/s00125-006-0247-y. [DOI] [PubMed] [Google Scholar]
- 29.Mohanram A, Zhang Z, Shahinfar S, Keane WF, Brenner BM, Toto RD. Anemia and end-stage renal disease in patients with type 2 diabetes and nephropathy. Kidney Int. 2004;66:1131–1138. doi: 10.1111/j.1523-1755.2004.00863.x. [DOI] [PubMed] [Google Scholar]
- 30.UK Prospective Diabetes Study (UKPDS) Group Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33) Lancet. 1998;352:83–853. doi: 10.1016/S0140-6736(98)85005-8. [DOI] [PubMed] [Google Scholar]
- 31.Adler AI, Stratton IM, Neil HA, et al. Association of systolic blood pressure with macrovascular and microvascular complications of type 2 diabetes (UKPDS 36): prospective observational study. BMJ. 2000;321:412–419. doi: 10.1136/bmj.321.7258.412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zoungas S, Arima H, Gerstein HC. Effects of intensive glucose control on microvascular outcomes in patients with type 2 diabetes: a meta-analysis of individual participant data from randomised controlled trials. Lancet Diabetes Endocrinol. 2017;5:431–437. doi: 10.1016/S2213-8587(17)30104-3. [DOI] [PubMed] [Google Scholar]
- 33.Babazono T, Takahashi C, Iwamoto Y. Definition of microalbuminuria in first-morning and random spot urine in diabetic patients. Diabetes Care. 2004;27:1838–1839. doi: 10.2337/diacare.27.7.1838. [DOI] [PubMed] [Google Scholar]
- 34.Davey SG, Ebrahim S. ’Mendelian randomization’: Can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol. 2003;32:1–22. doi: 10.1093/ije/dyg070. [DOI] [PubMed] [Google Scholar]