KIM-1 and Kidney Disease Progression in Autosomal Dominant Polycystic Kidney Disease: HALT-PKD Results

Abstract

Background:

Cyst compression of renal tubules plays a role in progression of autosomal dominant polycystic kidney disease (ADPKD) and may induce expression of kidney injury molecule-1 (KIM-1). Whether urinary KIM-1 indexed for creatinine (uKIM-1/Cr) is a prognostic marker of disease progression in ADPKD is unknown. In this secondary analysis of a prospective cohort study we sought to determine whether patients with high uKIM-1/CR as opposed to low at baseline had greater rates of eGFR loss and height-adjusted total kidney volume (HtTKV) increase.

Methods:

Baseline uKIM-1/Cr values were obtained from 754 participants in the Halt Progression of Polycystic Kidney Disease (HALT-PKD) studies A (early ADPKD) and B (late ADPKD). The predictor was uKIM-1/Cr which was dichotomized by the median value of 0.2417 pg/g, and the primary outcomes were measured longitudinally over time. Mixed-effects linear models were used in the analysis to calculate the annual slope of change in eGFR and HtTKV.

Results:

Patients with high uKIM-1/Cr (above the median) had an annual percent decline in eGFR that was 0.47 mL/min greater than those with low uKIM-1/Cr (p=.0015) after adjustment for all considered covariates. This association was seen in Study B patients alone (0.45 mL/min, p=.009), but not in Study A patients alone (0.42 mL/min, p=0.06). High baseline uKIM-1/Cr was asociated with higher HtTKV in the baseline cross-sectional analysis than low uKIM-1/Cr (p=.02), but there was no difference between the groups in the mixed-effect model annual slopes.

Conclusion:

Elevated baseline uKIM-1/Cr is associated with greater decline in eGFR over time. Further research is needed to determine if uKIM-1/Cr improves risk stratification in patients with ADPKD.

Introduction

Autosomal dominant polycystic kidney disease (ADPKD) is the most common single gene disorder causing kidney disease and the fourth leading cause of end stage kidney disease (ESKD) worldwide(1). ADPKD affects up to 12.5 million individuals and is characterized by bilateral progressively enlarging cysts that lead to progressive decline in kidney function(2). The root cause of the disorder lies with mutations in the PKD1 and PKD2 genes that encode the polycystin proteins (1 and 2, respectively) which are localized to the primary cilium on the apical membrane of the tubular cells of the kidneys(3).

The best current predictor of disease progression is height-adjusted total kidney volume (HtTKV) adjusted for age, which has been shown to outperform baseline creatinine, baseline blood urea nitrogen (BUN), and proteinuria for the prediction of progression to CKD 3(2, 4). The Mayo classification system categorizes patients into classes 1A through 1E based on age and HtTKV, with higher classes shown to be at greater risk of progression to ESKD (2.4% and 66.9% progression to ESKD at 10 years for stages 1A and 1E, respectively)(5). These HtTKV measurements are most often obtained from MRI imaging, and so are costly and in rare cases not feasible for patients with certain devices or hardware(6). Enrollment in clinical trials and treatment with Tolvaptan are currently reserved for patients with higher stages of disease, which makes accurate classification of great clinical importance(7).

Kidney Injury Molecule-1 (KIM-1) is a transmembrane protein that is upregulated in renal tubular cells after ischemic injury(8). KIM-1 is not expressed in the normal kidney but is expressed in a variety of human kidney diseases, predominantly in the apical membrane of proximal tubular cells(9). While KIM-1 is primarily used in the setting of drug-induced acute kidney injury (AKI), recent studies suggest that KIM-1 may be useful in predicting chronic kidney disease (CKD) progression as well(10, 11). Cyst growth plays an important role in ADPKD progression, and the degree of compression that enlarging cysts exert on the neighboring renal tubules may upregulate KIM-1 expression. Urinary KIM-1 may therefore signal higher risk of kidney disease progression in ADPKD(12). Given the plausible mechanism of KIM-1 upregulation in ADPKD and recent studies showing that KIM-1 can predict CKD progression in non-cystic diseases, we sought to determine whether KIM-1 predicts kidney disease progression in ADPKD. We hypothesized that baseline urinary KIM-1 is an independent predictor of future decline in estimated glomerular filtration rate (eGFR) and of increase in HtTKV(13)^,(14). We evaluated our hypotheses using urinary samples from the Halt Progression of Polycystic Kidney Disease (HALT-PKD) studies.

Methods

Study population:

The HALT-PKD studies are prospective, randomized, double-blind, placebo-controlled trials of 2 groups of patients with ADPKD. Study A included 558 individuals aged 15–49 years with normal kidney function defined as eGFR > 60 mL/min/1.73m² (“early” PKD)(14). Study B included 486 individuals aged 18–64 years with advanced chronic kidney disease (CKD) defined as eGFR 25–60 mL/min/1.73m² (“late” PKD)(15). The trials tested whether a lower blood pressure (BP) target (intensive group) delayed kidney disease progression compared with standard BP targets (study A), and whether multilevel blockade of the renin-angiotensin-aldosterone system (RAAS) delayed progression of kidney disease compared to monolevel blockade (studies A and B) over a follow-up duration of 96 months(16).

Study A included serial MRI HtTKV measurements in addition to serial eGFR determinations (by Chronic Kidney Disease Epidemiology Collaboration equation) and serial urinary albumin/creatinine ratios (uACR)(17). The results of study A showed that the combination of lisinopril and telmisartan did not significantly alter rates of HtTKV increase compared to monotherapy. Intensive BP control was associated with less increase in HtTKV, but not a change in eGFR. Study B, similarly, did not show a difference in eGFR changes with addition of an ARB compared to ACE inhibitor alone. For purposes of this analysis, we identified individuals from both studies (A and B) with an adequate volume of residual urine samples to measure KIM-1. All participants provided informed consent and permission to biobank their samples for future studies of ADPKD.

Predictors:

Urinary KIM-1 indexed for creatinine (uKIM-1/Cr) was measured using baseline samples available from studies A and B of HALT-PKD. Urine creatinine was measured using an enzymatic procedure (Roche), and urine albumin was measured using a nephelometric method (Seimens). uKIM-1 was measured using a multiplex assay on a MESO Scale Diagnostics platform and was measured in duplicate with the average of the results used. Dichotomized uKIM-1/Cr of high and low levels by observed median value of 0.2417 pg/g was used in the analysis. Creatinine indexing may be inappropriate in the setting of acute kidney injury(18) due to inconsistent creatinine excretion, but in the setting of chronic disease including the ADPKD population, creatinine indexing is more common(19).

Outcomes:

The primary outcomes were eGFR and HtTKV that were measured longitudinally over time. The decline in eGFR was examined in the combined A and B cohorts, and then in Study A and Study B separately. The HtTKV outcome could only be used for patients in study A because study B did not measure HtTKV.

Covariates:

The covariates adjusted for in the analysis included age, gender, PKD genotype (PKD 1 or PKD 2), baseline systolic and diastolic BP, baseline body mass index (BMI), baseline eGFR (for HtTKV analysis only), and baseline uACR. These variables were chosen for inclusion due to their strong epidemiologic associations with progression of disease in ADPKD(20).

Statistical Analysis:

Baseline characteristics and demographics were summarized according to uKIM-1/Cr level (greater than or less than the median of 0.2417 pg/g) using sample mean and standard deviation values or counts and percentages as appropriate. eGFR in the HALT-PKD studies was estimated using the Chronic Kidney Disease Epidemiology Collaboration equation. The chi-square test, t-test and rank sum test were used in the comparison between the two groups. Patients were excluded if they did not have complete data for the predictor, outcomes, or adjustment variables.

Linear mixed effects models with random intercept and random slope were used to evaluate whether baseline uKIM-1/Cr was associated with eGFR and HtTKV that were measured longitudinally. The association was assessed in an incremental series of models by first examining the unadjusted association and then adjusting for covariates in steps to see if the association might have been due to other factors. Dichotomized uKIM-1/Cr of high and low levels by the median was used as predictor in the analysis. As in the original linear mixed models in the HALT-PKD studies, baseline eGFR and HtTKV were retained as outcome measures in the mixed models, not adjusted for as covariates. In the analysis of HtTKV, the natural log value of HtTKV was used in the analysis and the slope was converted into annual percent change using the formula 100(e^β−1) as in the main analyses of the HALT PKD trials [14]. With mixed effects analysis of longitudinal data presented here, the interaction effect of uKIM-1/Cr and time is used to assess a relationship of interest, which is how uKIM-1/Cr associates with the change in the outcome variable over time (i.e. slope of time). In the analysis of mixed effects models with adjustment for a covariate, the effects of the covariate on both the intercept and slope were considered by including both the covariate and its interaction with time in the models. The confounders (i.e. covariates) in the final model included age, gender, PKD genotype, baseline systolic and diastolic BP, baseline body mass index (BMI), baseline eGFR (for HtTKV analysis only), and baseline uACR. The confounders were added stepwise as follows: Model 1 included adjustments for randomization group (treatment vs control) and study group (A vs B; only in the analysis of combined data); Model 2 added age and sex; Model 3 genotype, systolic blood pressure, diastolic blood pressure, body mass index, baseline eGFR (for HtTKV analysis only) and albumin creatinine ratio. The analysis was also repeated for studies A and B separately (and consequently study group was not considered a covariate). Associations were considered statistically significant at a p-value <0.05. All analyses were performed using SAS Institute, version 9.4, and R, version 3.1.3.

Results:

Baseline characteristics:

Out of 1,044 patients in the HALT-PKD studies A+B, 754 (70.8%) had complete data and were included in the analysis. The main reason for patient exclusion was the absence of a baseline uKIM-1/Cr value (228 patients, accounting for 79% of exclusions). Table 1 shows the baseline characteristics of patients with uKIM-1/Cr above and below the median uKIM-1/Cr value of 0.2417 pg/g. Patients in the high uKIM-1/Cr group were more likely to be from Study B and more likely to be female. There were also small but statistically significant differences in baseline eGFR and uACR. There were no differences between the groups in age, PKD genotype, baseline blood pressure, or BMI (Table 1). After adjustments for all covariates, baseline uKIM-1/Cr was associated with baseline HtTKV (p=0.009), but not baseline eGFR (p=0.8)

Table 1.

Baseline characteristics based on having a baseline uKIM-1/Cr above (high uKIM-1/Cr) or below (low uKIM-1/Cr) the median value of 0.2417 pg/g, as well as the combined baseline characteristics for the cohort.

Variables	High uKIM-1/Cr N=377	Low uKIM-1/Cr N=377	P-value
Age (years), Mean ± SD	43±10	43±10	.715
Early ADPKD (Study A), n (%)	184 (48.8)	226 (59.9)	.002
Treatment Group, n (%)	180 (47.7)	194 (51.5)	.308
Male Gender, n (%)	170 (45.1)	201 (53.3)	.024
PKD 1 genotype, n (%)	287 (76.1)	294 (78.0)	.544
Systolic BP (mmHg), Mean ± SD	127.6 ± 15.0	128.1 ± 13.9	.668
Diastolic BP (mmHg), Mean ± SD	80.0 ± 10.7	79.4 ± 10.0	.399
BMI (kg/m2), Mean ± SD	27.7 ± 5.0	27.5 ± 5.0	.557
Estimated GFR (mL/min/1.73 m2), Mean ± SD	69.8 ± 25.9	73.6 ± 24.8	.039
uACR (mg/g), median (IQR)	0.02 (0.01–0.05)	0.02(0.01–0.03)	<.001
uKIM-1/Cr (pg/g), median (IQR)	0.50 (0.34–0.75)	0.12 (0.07–0.18)	<.001

ADPKD, autosomal dominant polycystic kidney disease; BP, blood pressure; BMI, body mass index; GFR, glomerular filtration rate; uACR, urinary albumin-to-creatinine ratio; uKIM-1, urinary kidney injury molecule 1. The P values were calculated based on the chi-square test for categorical variables, rank sum test for uACR and uKIM-1/Cr, and t-test for other continuous variables.

Association of uKIM-1/Cr and change in eGFR in the combined cohort (HALT-PKD A and B):

In the final model, high baseline uKIM-1/Cr values were associated with an annual eGFR decline that was 0.47 mL/min greater than the low uKIM-1/Cr group, a result that was statistically significant (P = 0.002, Table 2).

Table 2.

Difference in annual decline in eGFR in patients with uKIM-1/Cr above (high uKIM-1/Cr) or below (low uKIM-1/Cr) the median value of 0.2417 pg/g.

Model	Difference in annual change in eGFR (mL/min), high versus low uKIM-1/Cr	p-value
Unadjusted	−0.64	<.001
Model 1	−0.58	<.001
Model 2	−0.56	<.001
Model 3	−0.47	.002

Model 1: Treatment group (Treatment vs Control), Study (A vs B)

Model 2: variables in model 1 + age, gender

Model 3: variables in model 2 + baseline systolic BP, baseline diastolic BP, baseline BMI, PKD genotype, and baseline uACR

Association of uKIM-1/Cr and change in eGFR in Study A and Study B:

We analyzed groups A and B separately to determine if baseline uKIM-1/Cr was associated with percent eGFR decline in the “early” and “late” ADPKD cohorts. Out of 558 patients in HALT-PKD study A and 486 in study B, 410 (73.5%) and 344 (70.7%), respectively, had complete data and were included in the analyses. In the final model for Study A patients, high uKIM-1/Cr values had a trend towards association with a decrease in eGFR that was 0.42 mL/min greater than the low uKIM-1/Cr group, although the result missed statistical significance (p=0.06). In Study B, there was a 0.45 mL/min annual difference between the groups which was statistically significant (p=.009, Table 3). Of note, using KIM-1 without indexing did not significantly impact results. Finally, the eGFR difference in the separate Study A and Study B cohorts were slightly lower than the combined cohort because the combined cohort featured an additional adjustment for study group.

Table 3.

Difference in annual eGFR change in patients with uKIM-1/Cr above (high uKIM-1/Cr) or below (low uKIM-1/Cr) the median value of 0.2417 pg/g, in patients in study A and B separately.

	Study A		Study B
Model	Difference in annual change in eGFR (mL/min), high versus low uKIM-1/Cr	p-value	Difference in annual change in eGFR (mL/min), high versus low uKIM-1/Cr	p-value
Unadjusted	−0.54	.02	−0.62	.001
Model 1	−0.55	.02	−0.62	.001
Model 2	−0.56	.02	−0.53	.003
Model 3	−0.42	.06	−0.45	.009

Model 1: Treatment group (Treatment vs Control)

Model 2: variables in model 1 + age, gender

Model 3: variables in model 2 + baseline systolic BP, baseline diastolic BP, baseline BMI, PKD genotype, and baseline uACR

Association of uKIM-1/Cr and change in HtTKV:

HtTKV data was not available in the study B participants, and so only the 486 patients with complete data from study A were included. Baseline uKIM-1/Cr was associated with baseline HtTKV but was not associated with annual changes (i.e. slope) in HtTKV (p=0.2).

Discussion

This study is the largest to date evaluating the prognostic role of uKIM-1/Cr levels in both early and late stage PKD. Patients with a baseline uKIM-1/Cr level above the median had greater decreases in eGFR than those below the median. Levels of uKIM-1/Cr above the median were associated with increased eGFR decline in patients with late ADPKD, with a trend towards association in early ADPKD (p=0.06). Finally, baseline uKIM-1/Cr levels were associated with baseline HtTKV values, although not with annual change in HtTKV over time. These findings suggest that uKIM-1/Cr may be a useful predictor of eGFR loss, especially in later stage ADPKD.

ADPKD is the most common single gene disorder causing kidney disease(1), and is characterized by bilateral progressively enlarging cysts that lead to progressive decline in kidney function(2). The root cause of the disorder lies with mutations in the PKD1 and PKD2 genes that encode the polycystin proteins (1 and 2, respectively) which are localized to the primary cilium on the apical membrane of the tubular cells of the kidneys(3). While many molecular mechanisms appear to underlie cyst formation and growth(3), cyst growth ultimately leads to compression and disruption of the renal tubular structure as well as tubular cell proliferation(21).

KIM-1 is a transmembrane protein with Ig-like and mucin domains in its ectodomain that functions as a putative epithelial cell adhesion molecule(8). KIM-1 levels are undetectable in the normal kidney but are increased dramatically in the setting of ischemia or renal tubular cell disruption(22), especially in the setting of drug-induced injury(23). The predominant site of expression following injury appears to be the apical membrane of proximal tubular cells(9). KIM-1 levels rise as early as 2 hours after injury(24), and KIM-1 expression also coincides with the expression of dedifferentiation and proliferation markers in the injured renal tubular cells, suggesting a role for KIM-1 in renal regeneration(8). In addition to its role in the detection of acute kidney injury, KIM-1 has also been shown to predict the development of chronic kidney disease (CKD)(10, 11). The exact mechanism underlying these associations remains unclear.

KIM-1 expression is upregulated in renal cell carcinoma (RCC). Interestingly, tubular cells adjacent to RCC often express KIM-1 even when the cancer cells do not, suggesting that compression of tubular cells by tumor cells may upregulate KIM-1 expression(25). Compression of tubular cells by cysts is thought to be a major reason for loss of kidney function in ADPKD(12), and it is therefore plausible that KIM-1 expression would be similarly upregulated in PKD, and that KIM-1 could predict disease progression in a manner similar to HtTKV. However, while uKIM-1/Cr did associate with baseline HtTKV in our study, it did not associate with annual change in HtTKV.

In addition to its role in the detection of acute kidney injury, KIM-1 has also been shown to predict the development of chronic kidney disease (CKD)(10, 11). In a recently published study by Schulz et al., plasma KIM-1 was measured in 4739 participants of the population-based Malmö Diet and Cancer Study. The risk of CKD development over a mean follow-up period of 16 years was 45% higher in the highest KIM-1 quartile compared to the lowest, even after adjustments for covariates(11). Similarly, in a secondary analysis of the Multi-Ethnic Study of Atherosclerosis (MESA) trial, patients in the highest decile of urinary KIM-1 levels at baseline had a doubling in their odds of development of incident CKD at 5 years(10). The findings of these studies associating urinary KIM-1 with CKD (other than ADPKD), suggest that KIM-1 upregulation reflects multifactorial pathophysiology in the kidney beyond cyst compression of renal tubules alone.

We are unaware of any studies that have evaluated uKIM-1/Cr as a predictor of CKD progression in ADPKD. KIM-1 has been shown to correlate with HtTKV in cross sectional analyses of small samples of patients with ADPKD(26, 27). In one cross-sectional study of 118 PKD patients, uKIM-1 was significantly increased compared to healthy controls. Interestingly, uKIM-1 correlated well with baseline HtTKV, although not with baseline eGFR(26). Similar observations were reported by Petzold et al. who also showed that KIM-1 correlated with baseline kidney volume, but not with baseline eGFR. Using HALT-PKD data, we were able to evaluate whether baseline uKIM-1 levels associate with kidney function decline by evaluating changes in eGFR and HtTKV over time. Similar to previous studies, baseline uKIM-1/Cr was not associated with eGFR at baseline in the HALT-PKD population, but higher uKIM-1/Cr at baseline did predict a statistically significant increase in eGFR loss over time.

In our study, uKIM-1/Cr was associated with HtTKV at baseline, but was not associated with the changes (i.e. slope) in HtTKV over time. This finding suggests that increased uKIM-1/Cr levels are not solely related to the increase in cyst size. Given the rise of uKIM-1/Cr in tubular cell disruption, uKIM-1/Cr may additionally be a marker of the number of cysts in addition to the size of the cysts. Alternatively, uKIM-1/Cr has been used to predict CKD progression in non-cystic disease, and uKIM-1/Cr elevations may therefore reflect underlying damage not directly related to cyst size or cyst growth. While further research is needed into the possible mechanisms of uKIM-1/Cr elevation in ADPKD, our results show that uKIM-1/Cr adds prognostic value beyond HtTKV alone. It is therefore possible that the incorporation of uKIM-1/Cr into the current risk stratification system could improve the prognostic evaluation. For example, clinical studies have focused on patients with Mayo stage 1C ADPKD or higher. It is possible that patients with 1B disease and elevated uKIM-1/Cr may be at high risk and could warrant inclusion in future studies. Conversely, patients with stage 1C disease and low uKIM-1/Cr may be at lower risk than suggested by HtTKV alone. Future research should examine how inclusion of uKIM-1/Cr impacts clinical classification.

Our study has several strengths including the large HALT-PKD database used for analysis, as well as the ability to associate uKIM-1/Cr with changes in eGFR and HtTKV over time. However, our study has several notable limitations. First, we cannot exclude the possibility of residual confounding despite our use of multivariable analysis. Secondly, there is no currently established cut-off for uKIM-1/Cr, which renders clinical application of these findings difficult. Missing data points, especially missing uKIM-1/Cr values, resulted in the exclusion of about 30% of patients, which reduced the power of our analysis. Finally, HtTKV data was not available in the late ADPKD group.

Our study demonstrates that baseline uKIM-1/Cr values are independently associated with eGFR decline over time. uKIM-1/Cr did not associate with changes in HtTKV over time, suggesting that the underlying mechanism of eGFR in these patients is not dependent solely of cyst growth. Further research is needed to validate these findings and to determine if a threshold level of uKIM-1/Cr or changes in uKIM-1/Cr can be utilized clinically in the assessment and follow up of patients with ADPKD.

Figure 1.

Forest plot of the differences in annual percent decline between the high uKIM-1/Cr and low uKIM-1/Cr groups in Study A, Study B, and in the combined cohort, as well as the difference in annual percent increase in HtTKV between the high and low uKIM-1/Cr groups in Study A. The results are based on the fully adjusted models.