Summary
Background and objectives
The Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) was created to identify markers of disease progression in patients with autosomal dominant polycystic kidney disease (ADPKD).
Design, setting, participants, & measurements
Linear mixed models were utilized to model effects of baseline parameters on changes in natural-log (ln)-transformed total kidney volume (TKV) and iothalamate clearance (GFR) across time in CRISP participants (creatinine clearance at entry >70 ml/min). Stepwise selection was used to obtain a final main effect model.
Results
TKV increased from year to year, whereas GFR uncorrected for body surface area (BSA) decreased only at year 6. Higher lnTKV and urine sodium excretion (UNaV), lower serum HDL-cholesterol, and younger age at baseline associated with greater lnTKV growth from baseline to year 3 and to year 6. Higher lnTKV at baseline associated with greater GFR decline from year 1 to year 3 and to year 6. Higher BSA and 24-hour urine osmolality at baseline associated with greater GFR decline from year 1 to year 6. Higher UNaV and lower serum HDL-cholesterol at baseline associated with greater GFR decline from year 1 to year 6 by univariate analysis only. Associations seen during year 1 to year 6 (not seen during year 1 to year 3) reflect the time lag between structural and functional disease progression.
Conclusions
Serum HDL-cholesterol, UNaV, and 24-hour urine osmolality likely affect ADPKD progression. To what extent their modification may influence the clinical course of ADPKD remains to be determined.
Introduction
The National Institutes of Health-sponsored Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) was created to develop imaging techniques and analyses to follow disease progression and evaluate treatments for autosomal dominant polycystic kidney disease (ADPKD) (1–5). CRISP followed 241 subjects (age 15 to 46 years; creatinine clearance at entry >70 ml/min) between January 5, 2001 and August 26, 2005 (CRISP I) at Emory University, Kansas University Medical Center, the Mayo Clinic, and the University of Alabama–Birmingham with baseline and three yearly (YR1 to YR3) visits including measurements of total kidney volume (TKV) by magnetic resonance (MR) and GFR by iothalamate clearance.
CRISP I participants seen at Emory University and the Mayo Clinic (n = 131) had renal blood flow (RBF) measurements by MR. Analysis of baseline predictors of disease progression in this subset showed that TKV, RBF, and urinary sodium excretion (UNaV) independently predicted TKV increase during 3 years of follow-up (5). Only TKV and total cyst volume (TCV) were independent predictors of GFR decline. Short follow-up and stability of GFR during CRISP I limited the ability to detect significant predictors of GFR decline.
To further evaluate the association of baseline covariates with structural (increase in TKV) or functional (decline in GFR) disease progression, we have extended the analysis to include the entire CRISP I cohort and up to 6 years of follow-up in most patients (CRISP II).
Materials and Methods
Study Organization
CRISP clinical centers included the Mayo Clinic and University of Alabama–Birmingham, Emory University, and the University of Kansas Medical Center. Washington University during CRISP I and the University of Pittsburgh during CRISP II served as the data coordinating and image analysis center (1–5), which collected images transferred from the collaborating institutions in Digital Imaging and Communications in Medicine format, segmented and analyzed images, stored and recorded results, and performed statistical analysis.
Study Protocol
Detailed descriptions of the study protocol and baseline characteristics have been published (1–5). ADPKD subjects were eligible if they were older than 15 and younger than 46 years of age and had a creatinine clearance >70 ml/min (1). Hypertension status was defined as previous diagnosis and current use of antihypertensive medications or as systolic and diastolic BP >140/90 mmHg on three consecutive visits.
After informed written consent, enrolled subjects were scheduled for two-day evaluations in the General Clinical Research Center at baseline and 1 (YR1), 2 (YR2), and 3 (YR3) between January 5, 2001 and August 26, 2005. Approximately 2 years after completion of CRISP I, participants were contacted and after informed written consent enrolled subjects were scheduled for a baseline CRISP II visit (YR6). Enrollment into CRISP II started in July 2007. CRISP II participants are followed by telephone or in center visits every 6 months; measurements of serum creatinine every year; and MR measurements of TKV, TCV, and iothalamate clearance every 2 years. Before each visit, participants were instructed to continue their medications, to discontinue any nonsteroidal anti-inflammatory medications for at least 7 days before evaluation, and not to initiate diuretic therapy within 14 days of evaluation. During CRISP I, subjects collected a 24-hour urine sample for determination of creatinine, urea nitrogen, UNaV, and urine albumin excretion (UalbV) on the day before admission. Weight, height, body surface area (BSA), and body mass index (BMI) were measured at admission. BP was measured in the morning, before antihypertensive medication intake, in the left and right arms after being seated for at least 5 minutes on three occasions 3 minutes apart using an oscillometric measuring device. Blood was obtained before GFR studies for the determination of blood hemoglobin, serum electrolytes, liver enzymes, and lipid profile during CRISP I and serum electrolytes and lipid profile only during CRISP II. GFR was measured by a nonradiolabeled iothalamate clearance technique with sonographic monitoring of bladder emptying (4). The mean coefficient of variation is 4.9%. GFR was also estimated (eGFR) using the Modification of Diet in Renal Disease (MDRD) equation: 186.3 × serum creatinine−1.154 × age−0.203 × 0.742 (if female) × 1.212 (if black) (6). Protein intake (g/d) was estimated using the equation 6.25 × [urine urea nitrogen in g/d + (0.03 × weight in kg)] (7). Urine osmolality averaged over 24 hours (24-h Uosm) was estimated using the equation urine urea nitrogen in mg/L/28 + 2 × (urine sodium + potassium in mEq/L) in a 24-hour urine collection.
Measurements of TKV
MR imaging studies were performed in the morning before medication intake and breakfast. During CRISP I, coronal T2-weighted images (single-shot fast spin echo/half-Fourier acquired single-shot turbo spin-echo) and gadolinium-enhanced, three-dimensional, volume-interpolated, spoiled-gradient echo coronal T1-weighted images were obtained (3-mm slice thickness). The volumes of individual kidneys were measured in T1-weighted images with a stereology method and calculated from the set of contiguous images by summing the products of the area measurements and slice thickness. The reliability coefficient was 0.998 for TKV. The average coefficient of variation of the TKV in repeated analysis of 99 images was 0.01%. A region-based thresholding method was used to calculate cyst volumes using T2-weighted images. During CRISP II, gadolinium enhanced T1-weighted images are no longer obtained because of concerns raised in 2006 about the role of gadolinium in nephrogenic systemic fibrosis (8). In addition to T2-weighted imaging, a fast imaging sequence, 2D true-FISP T2/T1-weighted imaging of the kidneys without fat saturation, is obtained to help delineate the kidney borders. The image analyst displayed these images concurrently as a visual guide and performed kidney volume measurement on T1-weighted images using the stereology method just as in the CRISP I image analysis.
Statistical Methods
The data were analyzed using Stata version 10 (StataCorp, College Station, TX). Changes in baseline covariates across time were examined by comparing each of the follow-up years to the baseline year using paired t tests or the McNemar test. Descriptive statistics for TKV and GFR were calculated to investigate the nature of the outcome variables. Because of the skewed nature of TKV, the natural log (lnTKV) was taken to normalize the data. Linear mixed models were utilized to model the change in outcome across time. In each model, there were fixed effects for year, the baseline covariates of interest, and their pairwise interactions. To account for the correlated observations within each subject, a random effect was also included. Of interest is a significant interaction between a particular baseline covariate and year, which would suggest that the covariate affects the trajectory of the outcome across time. This analysis was done for CRISP I subjects (baseline to YR3) as well as CRISP II subjects (baseline to YR6). Because of the nonlinear nature of the GFR outcome, the baseline year was not included in the assessment of change. This also helped mitigate regression to the mean.
Regression analyses were conducted in the following way. For each of the outcomes, univariate relationships were investigated by fitting a linear mixed model with year, the baseline covariate, and their respective interaction. Baseline covariates of interest are shown in Table 1. Covariates that significantly affected the outcome across time at the 0.15 level were included in a “full” model. Subsequently, if baseline covariates were significant at the 0.10 level in the full model, they were included in the pool of potential covariates for the “final” model. Backward selection was used in the final model by removing nonsignificant interactions (and, if possible, nonsignificant main effects) at the 0.05 significance level.
Table 1.
Relevant clinical, radiologic, and laboratory parameters at baseline and at YR1 to YR6
| Parameter | Baseline (n = 195 to 241) | Y1 (n = 174 to 241) | Y2 (n = 163 to 241) | Y3 (n = 168 to 241) | Y6 (n = 169 to 241) |
|---|---|---|---|---|---|
| Female gender (%) | 60.2 | 60.2 | 60.2 | 60.2 | 60.3 |
| Age (years) | 32.4 ± 8.9 | 33.6 ± 8.9c | 34.9 ± 8.8c | 35.8 ± 8.8c | 39.1 ± 8.7c |
| BSA (m2) | 1.89 ± 0.25 | 1.90 ± 0.25a | 1.92 ± 0.25c | 1.92 ± 0.25c | 1.93 ± 0.25c |
| BMI (kg/m2) | 25.9 ± 5.3 | 26.1 ± 5.3b | 26.4 ± 5.6c | 26.8 ± 5.8c | 27.1 ± 5.2c |
| Protein intake (g/day) | 71.9 ± 22.9 | 74 ± 28.1 | 76.8 ± 23.4a | 81.8 ± 27.5c | NA |
| Hypertension (%) | 61.0 | 64.2b | 69.3c | 73.3c | 74.2c |
| MAP (mmHg) | 92.7 ± 11.8 | 90.8 ± 11.7a | 90.8 ± 11.1a | 92 ± 11.9 | 91.8 ± 11.2 |
| TKV (ml) | 1073 ± 663 | 1123 ± 692c | 1240 ± 789c | 1277 ± 826c | 1510 ± 1106c |
| TCV (ml) | 534 ± 528 | 610 ± 586c | 686 ± 651c | 744 ± 714c | — |
| iGFR (ml/min) | 106.5 ± 27.8 | 108.2 ± 29.4 | 106.3 ± 33.6 | 104.6 ± 34.8 | 92 ± 41.4c |
| iGFR (ml/min per 1.73 m2) | 97.8 ± 24.7 | 98.8 ± 25.7 | 96.2 ± 29.7 | 94.5 ± 28.7a | 83.5 ± 38.3c |
| eGFR (ml/min per 1.73 m2) | 89.1 ± 27.7 | 85.3 ± 23.7c | 83.6 ± 26.9c | 81 ± 25.4c | 72.1 ± 28.3c |
| Serum sodium (mEq/L) | 138.1 ± 2.2 | 138 ± 2.7 | 138.3 ± 2.3 | 138.2 ± 2.3 | 137.6 ± 2.3a |
| Serum potassium (mEq/L) | 4.04 ± 0.46 | 3.97 ± 0.42a | 3.99 ± 2.07 | 3.81 ± 0.49c | 3.92 ± 0.42a |
| Serum HDL (mg/dl) | 47.8 ± 13.3 | 45.4 ± 12.9c | 44.6 ± 12.9c | 45.7 ± 12.1b | 46.4 ± 14.7 |
| Serum LDL (mg/dl) | 102 ± 34 | 101.1 ± 27.7 | 103.4 ± 28.9 | 100.9 ± 28.5 | 101.6 ± 31.6 |
| Serum uric acid (mg/dl) | 5 ± 1.51 | 5.09 ± 1.51 | 5.26 ± 1.58b | 5.42 ± 1.54c | NA |
| Urine volume (ml/24 h) | 2452 ± 1140 | 2777 ± 1216c | 2763 ± 1170b | 2794 ± 1231c | NA |
| Urine osmolality (mOsm/L) | 368 ± 159 | 332 ± 138b | 348 ± 162a | 342 ± 144a | NA |
| Urine albumin (mg/24 h) | 45.8 ± 63 | 43.3 ± 52.9 | 44.8 ± 62.5 | 53.1 ± 84.3a | NA |
| Urine citrate (mg/24 h) | 509 ± 350 | 490 ± 272 | 444 ± 238a | 441 ± 235 | NA |
| Urine urea nitrogen (mg/24 h) | 9194 ± 3396 | 9527 ± 4237 | 9947 ± 3416b | 10740 ± 4114c | NA |
| Urine potassium (mEq/24 h) | 58.9 ± 23.3 | 59.4 ± 23.9 | 59.5 ± 20.4 | 59.9 ± 21.3 | NA |
| Urine sodium (mEq/24 h) | 193.2 ± 86.1 | 203.1 ± 83.5 | 195.3 ± 75.9 | 190.6 ± 69.2 | NA |
Data are shown as mean ± SD. Numbers in parentheses represent minimal and maximal sample sizes. iGFR, GFR measured by iothalamate clearance; NA, data not available.
P < 0.05,
P < 0.01,
P < 0.001 compared to baseline.
Results
Baseline Characteristics and Longitudinal Changes
Two-hundred and forty-one patients participated in CRISP I and 203 re-enrolled into CRISP II. Table 1 summarizes the relevant clinical, radiologic, and laboratory parameters. Sixty percent of participants in CRISP I and in CRISP II are female. BSA and BMI increased steadily from baseline to YR6. Protein intake (estimated from urine urea nitrogen excretion) was significantly higher at YR2 and YR3 compared with baseline; urine urea nitrogen excretion was not measured at YR6. TKV and TCV increased significantly from year to year, whereas GFR uncorrected for BSA remained unchanged from baseline to YR3 and decreased significantly at YR6. This is consistent with a lag time between structural and functional changes in ADPKD.
Potential Baseline Predictors of lnTKV Slopes across Time
Table 2 shows which baseline covariates significantly predicted the change in lnTKV as reflected by a significant year × covariate interaction coefficient in the univariate models. Models were built to assess lnTKV change from baseline to YR3 or from baseline to YR6. Male gender and presence of hypertension at baseline were associated with higher TKV slopes. In addition, the higher the baseline BSA, BMI, estimated protein intake, mean arterial pressure, lnTKV, lnTCV, serum sodium and uric acid concentrations, lnUalbV, and UNaV, the higher was the increase in lnTKV over time. Baseline GFR and eGFR and serum HDL-cholesterol were negatively correlated with baseline to YR3 changes in lnTKV slopes, implying that the larger the baseline values, the lesser the increase in lnTKV over time. Associations between baseline covariates and lnTKV slopes from baseline to YR6 were similar to those between baseline covariates and baseline to YR3 slopes.
Table 2.
Baseline variables as univariate predictors of lnTKV across time
| Variable | Baseline YR3 |
Baseline YR6 |
||
|---|---|---|---|---|
| Year*Variable Coefficient | P | Year*Variable Coefficient | P | |
| Gender (% female) | −0.011 | 0.002 | −0.016 | <0.001 |
| Hypertension | 0.015 | <0.001 | 0.008 | 0.023 |
| Age (years) | −0.0002 | 0.465 | −0.001 | <0.001 |
| BMI (kg/m2) | 0.001 | <0.001 | 0.0004 | 0.172 |
| BSA (m2) | 0.037 | <0.001 | 0.032 | <0.001 |
| Protein intake (g/day) | 0.0003 | 0.002 | 0.0002 | 0.013 |
| MAP (mmHg) | 0.0003 | 0.048 | 0.0001 | 0.668 |
| lnTKV (ml) | 0.024 | <0.001 | 0.015 | <0.001 |
| lnTCV (ml) | 0.013 | <0.001 | 0.007 | <0.001 |
| iGFR (ml/min) | −0.000003 | 0.967 | 0.00001 | 0.898 |
| iGFR (ml/min per 1.73 m2) | −0.0002 | 0.008 | −0.0002 | 0.020 |
| Serum sodium (mEq/L) | 0.002 | 0.039 | 0.002 | 0.008 |
| Serium potassium (mEq/L) | 0.004 | 0.292 | −0.001 | 0.814 |
| Serum HDL (mg/dl) | −0.001 | <0.001 | −0.001 | <0.001 |
| Serum LDL (mg/dl) | 0.0001 | 0.318 | 0.00002 | 0.696 |
| Serum uric acid (mg/dl) | 0.006 | <0.001 | 0.006 | <0.001 |
| Urine volume (mL) | 0.000003 | 0.067 | 0.000004 | 0.011 |
| Urine osmolality (mOsm/L) | 0.00001 | 0.472 | 0.000003 | 0.781 |
| lnUalb (mg/24 h) | 0.008 | <0.001 | 0.004 | 0.033 |
| Urine citrate (mg/24 h) | 0.000001 | 0.899 | −0.00001 | 0.064 |
| Urine urea nitrogen (mg/24 h) | 0.000002 | 0.009 | 0.000001 | 0.037 |
| Urine potassium (mEq/24 h) | 0.0001 | 0.136 | 0.00001 | 0.883 |
| Urine sodium (mEq/24 h) | 0.0001 | <0.001 | 0.0001 | <0.001 |
Bold values indicate P ≤ 0.05. MAP, mean arterial pressure.
Regression Analysis for Changes in lnTKV across Time
When stepwise selection was used to obtain a final main effect model (Table 3), baseline lnTKV and UNaV were positively and HDL-cholesterol and age were negatively and independently associated with changes in lnTKV slopes across time from baseline to YR3 and from baseline to YR6. The higher the baseline UNaV, the greater was the increase of lnTKV over the 3 and 6 years on study. Higher baseline HDL and older age at the start of the study were independently associated with slower lnTKV growth.
Table 3.
Final regression models predicting structural disease progression measured as lnTKV across time from baseline parameters
| Parameter | Baseline to YR3 (n = 205) |
Baseline to YR6 (n = 165) |
||
|---|---|---|---|---|
| Coefficient | P | Coefficient | P | |
| Year | −0.072 | 0.004 | 0.011 | 0.655 |
| Age | 0.0002 | 0.705 | 0.001 | 0.466 |
| lnTKV | 1.001 | <0.001 | 1.014 | <0.001 |
| Serum HDL | 0.00004 | 0.915 | 0.001 | 0.205 |
| Urine sodium | −0.000003 | 0.960 | 0.00001 | 0.930 |
| Year*age | −0.0005 | 0.049 | −0.001 | <0.001 |
| Year*lnTKV | 0.022 | <0.001 | 0.014 | <0.001 |
| Year*serum HDL | −0.001 | <0.001 | −0.001 | <0.001 |
| Year*urine sodium | 0.0001 | 0.001 | 0.0001 | 0.008 |
| Intercept | −0.017 | 0.776 | −0.155 | 0.118 |
Bold values indicate P ≤ 0.05.
Potential Baseline Predictors of GFR Slopes across Time
Table 4 shows which baseline covariates significantly predicted the change in GFR across time from YR1 to YR3 or from YR1 to YR6 as reflected by a significant year × covariate interaction coefficient in the univariate models. Baseline GFR values were not used in the analysis to minimize regression to the mean. Higher lnTKV, lnTCV, MAP, age, and hypertension status at baseline were significantly associated with steeper GFR declines over time from YR1 to YR3 and more significantly from YR1 to YR6. Higher baseline protein intake, serum uric acid, 24-h Uosm, lnUalbV, and UNaV were associated with steeper declines, whereas higher serum HDL-cholesterol values were associated with less decline in GFR from YR1 to YR6, but not from YR1 to YR3. Higher significance of associations and emerging new associations between baseline covariates and GFR slopes from YR1 to YR6 as compared with slopes from YR1 to YR3 are consistent with the stability of the GFR during CRISP I and a time lag between structural and functional disease progression. Female gender was negatively associated with GFR slopes from YR1 to YR3, but not from YR1 to YR6.
Table 4.
Baseline variables as univariate predictors of uncorrected GFR across time
| YR1 to YR3 |
YR3 to YR6 |
|||
|---|---|---|---|---|
| Year*Variable Coefficient | P | Year*Variable Coefficient | P | |
| Gender (% female) | −4.8917 | 0.002 | 1.1240 | 0.148 |
| Hypertension | −3.3713 | 0.036 | −2.7624 | <0.001 |
| Age (years) | −0.1745 | 0.051 | −0.1175 | 0.008 |
| BMI (kg/m2) | −0.0690 | 0.638 | −0.2982 | <0.001 |
| BSA (m2) | 2.4554 | 0.427 | −6.9990 | <0.001 |
| Protein intake (g/day) | 0.0372 | 0.315 | −0.0615 | 0.001 |
| MAP (mmHg) | −0.1694 | 0.011 | −0.0926 | 0.005 |
| lnTKV (ml) | −4.7005 | 0.001 | −3.7777 | <0.001 |
| lnTCV (ml) | −2.6099 | <0.001 | −1.6222 | <0.001 |
| iGFR (ml/min) | 0.0575 | 0.043 | −0.0029 | 0.833 |
| iGFR (ml/min per 1.73 m2) | 0.0479 | 0.141 | 0.0375 | 0.023 |
| eGFR (ml/min per 1.73 m2) | 0.0175 | 0.560 | 0.0257 | 0.073 |
| Serum sodium (mEq/L) | 0.2920 | 0.414 | 0.0520 | 0.764 |
| Serum HDL (mg/dl) | −0.0797 | 0.188 | 0.0630 | 0.036 |
| Serum LDL (mg/dl) | 0.0278 | 0.251 | 0.0051 | 0.694 |
| Serum uric acid (mg/dl) | 1.5012 | 0.003 | −0.8473 | 0.002 |
| Urine volume (mL) | 0.0005 | 0.471 | 0.0003 | 0.333 |
| Urine osmolality (mOsm/L) | 0.0094 | 0.067 | −0.0084 | <0.001 |
| lnUalb (mg/24 h) | −1.0919 | 0.211 | −1.3488 | <0.001 |
| Urine citrate (mg/24 h) | 0.0019 | 0.446 | 0.0004 | 0.737 |
| Urine urea nitrogen (mg/24 h) | 0.0003 | 0.314 | −0.0003 | 0.007 |
| Urine sodium (mEq/24 h) | 0.0205 | 0.035 | −0.0116 | 0.010 |
Bold values indicate P ≤ 0.05.
Regression Analysis for Changes in GFR across Time
In the final stepwise regression model (Table 5), higher baseline lnTKV and serum uric acid concentration were associated with lower baseline GFR. Higher baseline lnTKV was also associated with a steeper decline in GFR, but higher baseline serum uric acid was associated with lower changes in GFR over time between YR1 and YR3. The relationship between lnTKV and GFR remained when adding YR6, but serum uric acid was no longer significant. Instead, higher baseline BSA was associated with higher baseline GFR and steeper decreases in GFR over time. This was also true for 24-h Uosm. Although UNaV was negatively and serum cholesterol concentration was positively associated with the change in GFR from YR1 to YR6, neither was an independently significant predictor of GFR decline in the final regression model.
Table 5A.
Final regression models predicting functional disease progression measured as GFR (uncorrected iothalamate clearance) across time from baseline parameters, YR1 to YR3
| YR1 to YR3 (n = 235) |
||
|---|---|---|
| Coefficient | P | |
| Year | 26.41 | 0.005 |
| lnTKV | −10.29 | 0.016 |
| Serum uric acid | −1.728 | 0.270 |
| Year*lnTKV | −5.483 | <0.001 |
| Year*serum uric acid | 1.821 | <0.001 |
| Intercept | 189.1 | <0.001 |
Bold values indicate P ≤ 0.05.
Table 5B.
Final regression models predicting functional disease progression measured as GFR (uncorrected iothalamate clearance) across time from baseline parameters, YR1 to YR6
| YR1 to YR6 (n = 182) |
||
|---|---|---|
| Coefficient | P | |
| Year | 29.94 | <0.001 |
| BSA | 60.97 | <0.001 |
| lnTKV | −18.26 | <0.001 |
| Urine osmolality | 0.0533 | <0.001 |
| Year*BSA | −4.799 | 0.004 |
| Year*lnTKV | −3.230 | <0.001 |
| Year*urine osmolality | −0.0065 | 0.009 |
| Intercept | 103.2 | <0.001 |
Bold values indicate P ≤ 0.05.
Serum HDL-Cholesterol, UNaV, and 24-h Uosm Are Constant Throughout CRISP
This analysis of CRISP data has revealed three potentially modifiable predictors of disease progression in ADPKD: serum HDL-cholesterol (a possible surrogate marker of vascular disease), UNaV (a surrogate marker of salt intake), and Uosm (a surrogate marker of vasopressin effect on the kidney). During CRISP, average (Table 1) and within-patient (Table 6) serum HDL-cholesterol concentration, UNaV, and 24-h Uosm remained mostly constant over time.
Table 6.
Spearman correlation coefficients among baseline, YR1, YR2, and YR3 serum HDL-cholesterol, UNaV, and 24-h Uosm values
| Year | Y1 | Y2 | Y3 | Y6 |
|---|---|---|---|---|
| Serum HDL-cholesterol | ||||
| baseline | ||||
| r | 0.705 | 0.712 | 0.703 | 0.708 |
| P | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
| n | 226 | 215 | 192 | 195 |
| Y1 | ||||
| r | — | 0.800 | 0.759 | 0.672 |
| P | — | <0.0001 | <0.0001 | <0.0001 |
| n | — | 213 | 191 | 192 |
| Y2 | ||||
| r | — | — | 0.817 | 0.715 |
| P | — | — | <0.0001 | <0.0001 |
| n | — | — | 182 | 181 |
| Y3 | ||||
| r | — | — | — | 0.703 |
| P | — | — | — | <0.0001 |
| n | — | — | — | 170 |
| UNaV | ||||
| baseline | ||||
| r | 0.425 | 0.395 | 0.369 | NA |
| P | <0.0001 | <0.0001 | <0.0001 | NA |
| n | 168 | 166 | 172 | NA |
| Y1 | ||||
| r | — | 0.444 | 0.418 | NA |
| P | — | <0.0001 | <0.0001 | NA |
| n | — | 161 | 166 | NA |
| Y2 | ||||
| r | — | — | 0.335 | NA |
| P | — | — | <0.0001 | NA |
| n | — | — | 166 | NA |
| 24-h Uosm | ||||
| baseline | ||||
| r | 0.408 | 0.495 | 0.467 | NA |
| P | <0.0001 | <0.0001 | <0.0001 | NA |
| n | 210 | 201 | 205 | NA |
| Y1 | ||||
| r | — | 0.682 | 0.601 | NA |
| P | — | <0.0001 | <0.0001 | NA |
| n | — | 207 | 210 | NA |
| Y2 | ||||
| r | — | — | 0.684 | NA |
| P | — | — | <0.0001 | NA |
| n | — | — | 206 | NA |
NA, data not available.
Discussion
The effects of multiple baseline covariates on the rate of renal enlargement in CRISP I and CRISP II participants followed for 6.3 years in the study presented here confirm and extend the conclusions of the previous study in a subset of CRISP I patients with 3 years of follow-up (5). As in that study, TKV and UNaV are independently predictive of renal growth. RBF, which was found to be an important predictor of disease progression, is not included in the study presented here because it was measured in only two of the four participating clinical centers. Exclusion of RBF from this analysis may have uncovered HDL-cholesterol as an independent predictor of renal growth (see below).
The effects of dietary salt on the development of hypertension, cardiovascular health, and progression of chronic kidney disease have received much attention (9–11). The study presented here suggests that it may also affect the progression of ADPKD. This can occur by several mechanisms. Modest increases in salt intake cause changes in plasma sodium, which are partially buffered by the movement of fluid from the intracellular to the extracellular compartment (12,13). Concomitant small increases in osmolality activate mechanisms to retain water (stimulation of vasopressin secretion and thirst) and elimination of salt (release of natriuretic factors such as atrial natriuretic factor) (13–15). Persistent high salt intake increases plasma levels of endogenous cardiotonic steroids and renal production of TGF-β (16–18). Vasopressin promotes cystogenesis (19). Atrial natriuretic peptide could have a cystogenic effect via generation of cGMP (20). Cardiotonic steroids induce extracellular signal-regulated kinase activation and proliferation of human ADPKD cyst-derived epithelial cells (18). TGF-β plays a role in the development of epithelial-mesenchymal transition and interstitial fibrosis in ADPKD (17). Increased salt consumption in the last decades underscores the relevance of dietary salt as a potential modifying factor in ADPKD (21,22).
Another observation of this study is the association of lower baseline serum HDL-cholesterol levels with faster renal enlargement. An association of lower HDL-cholesterol and steeper GFR decline had been observed in MDRD Study A, but renal volume was not measured (23). Several mechanisms could account for an effect of HDL on cyst growth. The detection of an HDL-cholesterol effect in this but not our previous study when baseline RBF was included in the model suggests that the effect may be indirect through an effect on the vasculature. Indeed, HDL exerts antiatherogenic and anti-inflammatory actions on the vasculature through its role in the reverse transport of cholesterol and through the sphingosine-1-phosphate (S1P)-S1P1–5 receptor (a family of five G-protein-coupled receptors) and apolipoprotein A1-scavenging receptor class B type1 receptor interactions (24,25). S1P or apolipoprotein A1 could directly affect the development of polycystic kidney disease because renal sphingolipid metabolism is altered (26,27) and renal expression of apolipoprotein A1 is downregulated (28) in animal models of polycystic kidney disease.
In our previous study, no significant association was detected between UNaV or serum HDL-cholesterol and GFR decline (5). Because of the lag time between structural and functional changes in ADPKD, we re-examined the associations between baseline parameters and GFR decline in the study presented here with longer follow-up. Consistent with this lag time, we find a higher number and significance of associations between baseline covariates and GFR slopes during YR1 to YR6 compared with YR1 to YR3. With longer follow-up from YR1 to YR6, higher UNaV and lower serum HDL-cholesterol concentrations become significantly associated with faster GFR decline. However, UNaV and serum HDL-cholesterol are not independently significant in the final main effect model. It is possible that longer follow-up will be needed to detect a larger and independent effect of these parameters on renal function.
In the final regression model to predict GFR decline, two baseline covariates, in addition to lnTKV, have an independent significant effect: BSA and 24-h Uosm. In both cases, they are positively associated with GFR at baseline and negatively with GFR changes over time. The explanation for the positive association between baseline BSA and GFR uncorrected for BSA is obvious. The association between baseline 24-h Uosm and GFR likely reflects better urine concentrating capacity in the patients with higher GFR and/or the known relationship between urinary concentrating activity and GFR (29). On the other hand, negative associations of baseline BSA and 24-h Uosm with GFR change over time suggest detrimental effects on renal function. The association between BSA and rate of disease progression in ADPKD is consistent with a recent analysis in the large HALT-PKD clinical trial cohort showing a highly significant independent association between baseline BSA and renal disease severity ascertained by BSA- or height-adjusted TKV and eGFR (V.E. Torres et al., submitted). This may point to an effect of developmental programming on the clinical course of ADPKD. The association between 24-h Uosm and rate of GFR decline is consistent with a large body of evidence suggesting that vasopressin contributes to the progression of ADPKD and chronic kidney disease because 24-h Uosm can be considered a surrogate marker for vasopressin effect on the kidney (29,30–32).
A post hoc analysis of the MDRD Study A patients with ADPKD (GFR 25 to 55 ml/min per 1.73 m2; n = 139) showed an association between high urine volume or low urine osmolality with faster GFR decline (33). Because patients with the highest urine volumes tended to have lower serum sodium concentrations and hypostenuria, excess water intake and not a renal concentrating defect was deemed responsible for the high urine volumes. Pushing fluids was suggested to be detrimental, possibly by increasing intratubular pressure and promoting cyst growth. In the study presented here, urine volume and 24-h Uosm are not independent predictors of renal growth. Contrary to the observation in the MDRD study that serum sodium concentrations tended to be lower in the patients with faster GFR decline, we find a statistically significant direct correlation between serum sodium level and the rate of renal enlargement, possibly reflecting more severe disease and impaired concentrating ability or lower fluid intake. As discussed above, baseline 24-h Uosm was positively associated with GFR, but negatively associated with GFR change across time. Taken together, these results do not support the hypothesis that high fluid intake promotes cyst growth or is detrimental to renal function in ADPKD.
In summary, this study has identified three potentially modifiable phenotypic traits that likely affect the progression of ADPKD: serum HDL-cholesterol (possibly a surrogate marker of vascular disease) and UNaV and 24-h Uosm (surrogate markers of salt and vasopressin effect on the kidney). To what extent their modification may influence the clinical course of ADPKD remains to be determined.
Disclosures
None.
Acknowledgments
This study was supported by the CRISP Study UO1 cooperative agreements (DK56956, DK56943, DK56957, and DK56961), the Mayo Clinic General Clinical Research Grant MO1-RR00585, the Emory University General Clinical Research Grant MO1-RR00039, and the University of Kansas General Clinical Research Grant (number to come) of the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health. We are indebted to the study coordinators who worked with professional grace and high competence to bring this project to conclusion. We thank Dr. Lise Bankir (Institut National de la Sante′ et de la Recherche Medicale, UMRS 872, Centre de Recherche des Cordeliers, Paris, France) for helpful discussions and advice.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
Access to UpToDate on-line is available for additional clinical information at www.cjasn.org.
References
- 1. Chapman A, Guay-Woodford LM, Grantham JJ, Torres V, Bae K, Baumgarten D, Kenney P, King B, Glockner J, Wetzel L, Brummer M, O'Neill C, Robbin M, Bennett W, Klahr S, Hirschman G, Kimmel P, Thompson P, Miller J: Renal structure in early autosomal dominant polycystic kidney disease (ADPKD): the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) Cohort. Kidney Int 64: 1035–1045, 2003 [DOI] [PubMed] [Google Scholar]
- 2. Grantham JJ, Torres VE, Chapman AB, Guay-Woodford LM, Bae KT, King BF, Jr, Wetzel LH, Baumgarten DA, Kenney PJ, Harris PC, Klahr S, Bennett WM, Hirschman GN, Meyers CM, Zhang X, Zhu F, Miller JP: Volume progression in polycystic kidney disease. N Engl J Med 354: 2122–2130, 2006 [DOI] [PubMed] [Google Scholar]
- 3. King BF, Torres VE, Brummer ME, Chapman AB, Bae KT, Glockner JF, Arya K, Felmlee JP, Grantham JJ, Guay-Woodford LM, Bennett WM, Klahr S, Hirschman GH, Kimmel PL, Thompson PA, Miller JP: Magnetic resonance measurements of renal blood flow as a marker of disease severity in autosomal-dominant polycystic kidney disease. Kidney Int 64: 2214–2221, 2003 [DOI] [PubMed] [Google Scholar]
- 4. Rule AD, Torres VE, Chapman AB, Grantham JJ, Guay-Woodford LM, Bae KT, Klahr S, Bennett WM, Meyers CM, Thompson PA, Miller JP. CRISP: Comparison of methods for determining renal function decline in early autosomal dominant polycystic kidney disease: The consortium of radiologic imaging studies of polycystic kidney disease cohort. J Am Soc Nephrol 17: 854–862, 2006 [DOI] [PubMed] [Google Scholar]
- 5. Torres VE, King BF, Chapman AB, Brummer ME, Bae KT, Glockner JF, Arya K, Risk D, Felmlee JP, Grantham JJ, Guay-Woodford LM, Bennett WM, Klahr S, Meyers CM, Zhang X, Thompson PA, Miller JP: Magnetic resonance measurements of renal blood flow and disease progression in autosomal dominant polycystic kidney disease. Clin J Am Soc Nephrol 2: 112–120, 2007 [DOI] [PubMed] [Google Scholar]
- 6. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130: 461–470, 1999 [DOI] [PubMed] [Google Scholar]
- 7. Maroni BJ, Steinman TI, Mitch WE: A method for estimating nitrogen intake of patients with chronic renal failure. Kidney Int 27: 58–65, 1985 [DOI] [PubMed] [Google Scholar]
- 8. Marckmann P, Skov L, Rossen K, Dupont A, Damholt MB, Heaf JG, Thomsen HS: Nephrogenic systemic fibrosis: Suspected causative role of gadodiamide used for contrast-enhanced magnetic resonance imaging. J Am Soc Nephrol 17: 2359–2362, 2006 [DOI] [PubMed] [Google Scholar]
- 9. Manunta P, Hamilton BP, Hamlyn JM: Salt intake and depletion increase circulating levels of endogenous ouabain in normal men. Am J Physiol Regul Integr Comp Physiol 290: R553–R559, 2006 [DOI] [PubMed] [Google Scholar]
- 10. Ritz E, Koleganova N, Piecha G: Role of sodium intake in the progression of chronic kidney disease. J Ren Nutr 19: 61–62, 2009 [DOI] [PubMed] [Google Scholar]
- 11. Sanders PW: Vascular consequences of dietary salt intake. Am J Physiol Renal Physiol 297: F237–F243, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. He FJ, Markandu ND, Sagnella GA, de Wardener HE, MacGregor GA: Plasma sodium: Ignored and underestimated. Hypertension 45: 98–102, 2005 [DOI] [PubMed] [Google Scholar]
- 13. Sagnella GA, Markandu ND, Buckley MG, Miller MA, Singer DR, MacGregor GA: Hormonal responses to gradual changes in dietary sodium intake in humans. Am J Physiol 256: R1171–R1175, 1989 [DOI] [PubMed] [Google Scholar]
- 14. Buckley MG, Saggar-Malik AK, Markandu ND, Sagnella GA, MacGregor GA: Effects of changes in dietary sodium intake on plasma brain natriuretic peptide levels in patients with autosomal dominant polycystic kidney disease. J Hum Hypertens 10: 415–416, 1996 [PubMed] [Google Scholar]
- 15. Szinnai G, Morgenthaler NG, Berneis K, Struck J, Muller B, Keller U, Christ-Crain M: Changes in plasma copeptin, the C-terminal portion of arginine vasopressin during water deprivation and excess in healthy subjects. J Clin Endocrinol Metab 92: 3973–3978, 2007 [DOI] [PubMed] [Google Scholar]
- 16. Bagrov AY, Shapiro JI, Fedorova OV: Endogenous cardiotonic steroids: Physiology, pharmacology, and novel therapeutic targets. Pharmacol Rev 61: 9–38, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Chea SW, Lee KB: TGF-beta mediated epithelial-mesenchymal transition in autosomal dominant polycystic kidney disease. Yonsei Med J 50: 105–111, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Nguyen AN, Wallace DP, Blanco G: Ouabain binds with high affinity to the Na,K-ATPase in human polycystic kidney cells and induces extracellular signal-regulated kinase activation and cell proliferation. J Am Soc Nephrol 18: 46–57, 2007 [DOI] [PubMed] [Google Scholar]
- 19. Wang X, Wu Y, Ward CJ, Harris PC, Torres VE: Vasopressin directly regulates cyst growth in polycystic kidney disease. J Am Soc Nephrol 19: 102–108, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Wang X, Ward CJ, Harris PC, Torres VE: Cyclic nucleotide signaling in polycystic kidney disease. Kidney Int 77: 129–140, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Havas S, Dickinson BD, Wilson M: The urgent need to reduce sodium consumption. JAMA 298: 1439–1441, 2007 [DOI] [PubMed] [Google Scholar]
- 22. He FJ, MacGregor GA: A comprehensive review on salt and health and current experience of worldwide salt reduction programmes. J Hum Hypertens 23: 363–384, 2009 [DOI] [PubMed] [Google Scholar]
- 23. Klahr S, Breyer J, Beck G, Dennis V, Hartman J, Roth D, Steinman T, Wang S, Yamamoto M: Dietary protein restriction, blood pressure control, and the progression of polycystic kidney disease modification of diet in renal disease study group. J Am Soc Nephrol 5: 2037–2047, 1995 [DOI] [PubMed] [Google Scholar]
- 24. Okajima F, Sato K, Kimura T: Anti-atherogenic actions of high-density lipoprotein through sphingosine 1-phosphate receptors and scavenger receptor class B type I. Endocr J 56: 317–334, 2009 [DOI] [PubMed] [Google Scholar]
- 25. Jo SK, Bajwa A, Awad AS, Lynch KR, Okusa MD: Sphingosine-1-phosphate receptors: Biology and therapeutic potential in kidney disease. Kidney Int 73: 1220–1230, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Chatterjee S, Pandey A: The Yin and Yang of lactosylceramide metabolism: Implications in cell function. Biochim Biophys Acta 1780: 370–382, 2008 [DOI] [PubMed] [Google Scholar]
- 27. Deshmukh G, Radin N, Gattone V, Shayman J: Abnormalities of glycosphingolipid sulfatide, and ceramide in the polycystic (cpk/cpk) mouse. J Lipid Res 35: 1611–1618, 1994 [PubMed] [Google Scholar]
- 28. Yoo KH, Kim YN, Lee MJ, Seong JK, Park JH: Identification of apolipoproteinA1 reduction in the polycystic kidney by proteomics analysis of the Mxi1-deficient mouse. Proteomics 9: 3824–3832, 2009 [DOI] [PubMed] [Google Scholar]
- 29. Bankir L: Antidiuretic action of vasopressin: Quantitative aspects and interaction between V1a and V2 receptor-mediated effects. Cardiovasc Res 51: 372–390, 2001 [DOI] [PubMed] [Google Scholar]
- 30. Meijer E, Bakker SJ, de Jong PE, Homan van der Heide JJ, van Son WJ, Struck J, Lems SP, Gansevoort RT: Copeptin, a surrogate marker of vasopressin, is associated with accelerated renal function decline in renal transplant recipients. Transplantation 88: 561–567, 2009 [DOI] [PubMed] [Google Scholar]
- 31. Meijer E, Bakker SJ, van der Jagt EJ, de Jong D, Struck J, Gansevoort RT: Copeptin, a surrogate marker of vasopressin, is associated with disease severity in autosomal dominant polycystic kidney disease. Clin J Am Soc Nephol [Epub ahead of print] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Torres VE, Bankir L, Grantham JJ: A case for water in the treatment of polycystic kidney disease. Clin J Am Soc Nephrol 4: 1140–1150, 2009 [DOI] [PubMed] [Google Scholar]
- 33. Hebert LA, Greene T, Levey A, Falkenhain ME, Klahr S: High urine volume and low urine osmolality are risk factors for faster progression of renal disease. Am J Kidney Dis 41: 962–971, 2003 [DOI] [PubMed] [Google Scholar]