This combined observational and genetic study investigates the association between chronic kidney disease and risk of spontaneous intracerebral hemorrhage.
Key Points
Question
Is chronic kidney disease associated with higher risk of spontaneous intracerebral hemorrhage?
Findings
In observational analyses using data from 2 large studies, results indicated that chronic kidney disease was associated with higher risk of spontaneous intracerebral hemorrhage; genomic analyses evaluating genetically determined chronic kidney disease confirmed this association.
Meaning
These findings suggest a causal association between chronic kidney disease and spontaneous intracerebral hemorrhage.
Abstract
Importance
The evidence linking chronic kidney disease (CKD) to spontaneous intracerebral hemorrhage (ICH) is inconclusive owing to possible confounding by comorbidities that frequently coexist in patients with these 2 diseases.
Objective
To determine whether there is an association between CKD and ICH risk.
Design, Setting, and Participants
A 3-stage study that combined observational and genetic analyses was conducted. First, the association between CKD and ICH risk was tested in the Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH) study, a multicenter case-control study in the US. All participants with available data on CKD from ERICH were included. Second, this analysis was replicated in the UK Biobank (UKB), an ongoing population study in the UK. All participants in the UKB were included in this study. Third, mendelian randomization analyses were implemented in the UKB using 27 CKD-related genetic variants to test for genetic associations. ERICH was conducted from August 1, 2010, to August 1, 2017, and observed participants for 1 year. The UKB enrolled participants between 2006 and 2010 and will continue to observe them for 30 years. Data analysis was performed from November 11, 2019, to May 10, 2022.
Exposures
CKD stages 1 to 5.
Main Outcomes and Measures
The outcome of interest was ICH, ascertained in ERICH via expert review of neuroimages and in the UKB via a combination of self-reported data and International Statistical Classification of Diseases, Tenth Revision, codes.
Results
In the ERICH study, a total of 2914 participants with ICH and 2954 controls who had available data on CKD were evaluated (mean [SD] age, 61.6 [14.0] years; 2433 female participants [41.5%]; 3435 male participants [58.5%]); CKD was found to be independently associated with higher risk of ICH (odds ratio [OR], 1.95; 95% CI, 1.35-2.89; P < .001). This association was not modified by race and ethnicity. Replication in the UKB with 1341 participants with ICH and 501 195 controls (mean [SD] age, 56.5 [8.1] years; 273 402 female participants [54.4%]; 229 134 male participants [45.6%]) confirmed this association (OR, 1.28; 95% CI, 1.01-1.62; P = .04). Mendelian randomization analyses indicated that genetically determined CKD was associated with ICH risk (OR, 1.56; 95% CI, 1.13-2.16; P = .007).
Conclusions and Relevance
In this 3-stage study that combined observational and genetic analyses among study participants enrolled in 2 large observational studies with different characteristics and study designs, CKD was consistently associated with higher risk of ICH. Mendelian randomization analyses suggest that this association was causal. Further studies are needed to identify the specific biological pathways that mediate this association.
Introduction
Spontaneous, nontraumatic, intracerebral hemorrhage (ICH) is a devastating disease. Forty percent of affected patients die within 30 days and less than 30% of survivors regain the ability to walk.1,2 The high morbidity and mortality of this disease are partly attributable to a lack of effective therapies to improve outcomes.3,4,5,6 Primary prevention is thus an important method of reducing the burden of ICH, but preventive strategies remain limited to strict blood pressure control, which has been challenging on a population basis.7,8 Identification of novel biologic pathways underlying the occurrence of ICH is needed to inform the development of new preventive and therapeutic strategies.
Several studies indicate that chronic kidney disease (CKD) is associated with higher risk9,10,11 and worse outcomes after ischemic stroke.11,12,13 However, the link between CKD and ICH, the most frequent type of hemorrhagic stroke, is less clear. Studies evaluating this association have been limited by small sample sizes and lack of racial and ethnic diversity, yielding inconsistent results in White participants and limited applicability to Asian, Black, and Hispanic populations.14,15,16 Furthermore, these studies have been unable to assess causal relationships between CKD and ICH owing to their observational design.
To overcome these limitations, we combined observational and genetic data from 2 large landmark studies to evaluate the association between CKD and ICH. We first tested for an association between CKD and ICH risk using data from the observational Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH) study, the largest, to our knowledge, multiancestry study of this condition completed to date.17 We subsequently pursued an independent replication in the UK Biobank (UKB), one of the largest population-based studies with available genome-wide data.18 Finally, we further evaluated the encountered associations by conducting a mendelian randomization analysis in the UKB. Owing to their random assortment during meiosis, genetic variants associated with CKD are relatively immune to confounding by exposures that exert their effect after birth and can thus be used as instruments to evaluate the causal association between CKD and risk of ICH.19
Methods
Study Design and Inclusion Criteria
We conducted a 3-stage study that combined observational and genetic analyses (eFigure in the Supplement). First, we tested for associations between CKD and ICH risk in ERICH, a multicenter, prospective, case-control study that enrolled 3000 participants with ICH and 3000 controls matched by age, sex, and racial and ethnic group from August 1, 2010, to August 1, 2017.17 Cases and controls had equal representation of Black, Hispanic or Latino, and White participants. Second, we conducted an independent replication of this analysis in the UKB, an observational cohort study that enrolled more than 500 000 Britons in the UK from 2006 to 2010.18 The UKB included participants of the following race and ethnicity categories: Asian, Black, mixed, White and other or unknown race and ethnicity. Third, we implemented a mendelian randomization genetic analysis in the UKB to evaluate whether genetically determined CKD was associated with ICH risk. In secondary analyses, we examined the association between CKD and functional outcome in the ERICH study, as several well-established risk factors for ICH have also been shown to associate with poor outcome after the hemorrhage.20 The study protocols for both the ERICH study and UKB were approved by their respective institutional review boards. All participants or their legally designated surrogates provided written informed consent. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Ascertainment of CKD in the ERICH and UKB Studies
In the ERICH study, CKD was ascertained from a combination of in-person interviews and medical record abstraction using a dedicated form that identified the variables of interest to the study. In the UKB, CKD was ascertained using data from baseline in-person interviews and previously validated International Statistical Classification of Diseases, Tenth Revision (ICD-10), codes21,22 obtained from electronic health records of hospital admissions and visits to the general practitioner (eTable 1 in the Supplement). We defined CKD to include all stages (1-5). Because the ERICH study lacked specific data on end-stage kidney disease (ESKD), we defined this condition as the combination of CKD and dialysis for this study. For the secondary analysis on functional outcome in the ERICH study, we included additional cases of CKD and ESKD extracted from free-text fields for medical history on the medical record abstraction form, which was available only for participants with ICH.
Ascertainment of ICH in the ERICH and UKB Studies
The ERICH study enrolled participants with ICH who were at least 18 years of age, lived within 75 miles of a recruiting center, and had neuroimaging-confirmed brain intraparenchymal hemorrhage with accompanying acute neurologic symptoms. The diagnosis of ICH was confirmed by study neurologists via review of clinical and neuroimaging data, and each brain hemorrhage was classified as lobar (when located in superficial regions of the brain) or nonlobar (when located in the basal ganglia, thalamus, cerebellum, or brainstem). ICH attributable to brain tumors, vascular malformations, and hemorrhagic conversion of an ischemic stroke were excluded. ICH cases in the UKB were ascertained by an outcome adjudication group using algorithmic combinations of coded information from the baseline in-person interview and International Classification of Diseases, Ninth Revision, and ICD-10 codes (eTable 1 in the Supplement) appropriate for primary, nontraumatic ICH from hospital admissions and death registries.23 Both prevalent (present on enrollment) and incident (occurred during follow-up) ICH cases were included in the analysis.
Ascertainment of Functional Outcome in the ERICH Study
Functional outcome was ascertained 3 months after the ICH by trained study staff using the modified Rankin Scale (mRS), a score that ranges from 0 (normal neurologic examination) to 6 (death). Following the approach used in several other studies, the mRS score obtained 3 months after ICH was dichotomized, with poor outcome defined as a score of 4 to 6.24
Genomic Data in the UKB
Genotyping was carried out by the Affymetrix Research Services Laboratory in 106 sequential batches of approximately 4700 samples each. A subset of 49 950 participants involved in the UK Biobank Lung Exome Variant Evaluation study was genotyped using the Applied Biosystems UK BiLEVE Axiom Array (Affymetrix) that contains 807 411 markers. Another 438 427 participants were genotyped using the closely related Applied Biosystems UK Biobank Axiom Array that contains 825 927 markers, of which 95% are shared with the UK BiLEVE Axiom Array. Standard quality-control procedures for genome-wide data were performed centrally by the UKB research team, as previously reported,18 on markers that were present in both the UK BiLEVE Axiom Array and the UK Biobank Axiom Array. Principal component analysis was used to evaluate and assign ancestry. This pipeline yielded 488 377 samples and 805 426 markers that entered the imputation process, which was completed using the IMPUTE4 software (University of Oxford) and a combination of 3 reference panels: Haplotype Reference Consortium, UK10K haplotype, and 1000 Genomes Phase 3.25,26,27 The result of the imputation process was a data set with 93 095 623 autosomal single nucleotide variations (SNVs), short indels, and large structural variants in 487 320 individuals.
Statistical Analyses
Stage 1: CKD and ICH Risk in the ERICH Study
Discrete variables are presented as count (percentage) and continuous variables as mean (SD), as appropriate. We fitted univariable and multivariable logistic regression models to evaluate the association between CKD and risk of ICH. Multivariable model building proceeded in several steps: first, covariates with P < .10 in univariable analyses were included in the model; second, universal confounders (age, sex, and race and ethnicity) were forced into the model; third, covariates with P > .10 were backward eliminated; and fourth, collinear covariates, as expressed by a variance inflation factor greater than 5, were identified and removed from the model. We evaluated whether race and ethnicity modified the association between CKD and ICH by adding product terms to our regression model. We also evaluated hemorrhage location as an effect modifier. Because information on location was not available for controls, we implemented a case-only analysis,28 using multivariable logistic regression, where the effect modifier (hemorrhage location) was the outcome and CKD remained the exposure. In sensitivity analyses, we evaluated ESKD instead of CKD as the exposure of interest.
Stage 2: Replication in the UKB and Meta-analysis
We fitted multivariable logistic regression models to replicate the association between CKD and ICH risk. Model building proceeded as described for stage 1. In a sensitivity analysis, we calculated propensity scores based on covariates included in the multivariable logistic regression model, matched ICH cases and non-ICH cases to a 1:10 ratio using the nearest-neighbor method, and then fitted conditional logistic regression models. We pooled the study-specific estimates from the ERICH study and the UKB using inverse variance, random-effects meta-analysis, assessing the heterogeneity across estimates using the restricted maximum-likelihood method, with corresponding P values and I2 estimates to quantify the amount of heterogeneity.
Stage 3: Genetic Analysis in the UKB
We conducted a 1-sample mendelian randomization analysis using data from UKB participants of genetically confirmed European ancestry. As instruments, we used 27 independent (r2 < 0.1, a measure of correlation between genetic variants) SNVs previously reported to be associated with CKD at genome-wide significant levels (P < 5 × 10−8) (eTable 2 in the Supplement).29 SNVs with palindromic alleles (A/T or C/G) were not considered. The selected SNVs were tested for association against the risk of both CKD and ICH, and the results were used to implement the inverse variance–weighted method. In secondary analyses, we implemented the more conservative weighted-median approach. We assessed horizontal pleiotropy via the mendelian randomization Egger (MR-Egger) and mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) estimates. To account for residual confounding, we performed a sensitivity analysis in which we excluded 2 SNVs that were also associated with hypertension and diabetes, types 1 and 2.
Secondary Analysis on Functional Outcome in the ERICH Study
Using data from the ERICH study, we fitted multivariable logistic regression models to evaluate the association between CKD and poor functional outcome. Model building proceeded as described for stage 1. We tested for effect modification by race and ethnicity and lobar or nonlobar location by adding product terms to our regression models. In sensitivity analyses, we dichotomized the mRS as 0 to 2 vs 3 to 6 (instead of 0-3 vs 4-6) and used ordinal logistic regression to model the mRS as a discrete variable. We did not pursue outcome analyses in the UKB owing to the lack of information on functional status after ICH. We used R for all statistical analyses (R Foundation), PLINK, version 1.930 for genetic analyses, and the MendelianRandomization and MR-PRESSO packages in R for mendelian randomization analyses. All P values were 2-sided, and P < .05 indicated statistical significance. Data analysis was performed from November 11, 2019, to May 10, 2022.
Results
Chronic Kidney Disease and Risk of Intracerebral Hemorrhage in the ERICH Study
Among the 6000 study participants enrolled in the ERICH study, we included 2914 participants with ICH and 2954 controls with available data on CKD (mean [SD] age, 61.6 [14.0] years; 2433 female participants [41.5%]; 3435 male participants [58.5%]). Study participants from the following races and ethnicities were included: 1951 Black (33.2%), 1946 Hispanic or Latino (33.2%), and 1971 White (33.6%) (Table 1). The prevalence of CKD was 4.3% (125 of 2914) in ICH cases and 1.3% (39 of 2954) in controls (unadjusted P < .001). Baseline demographics and comorbidities by diagnosis of CKD are shown in eTable 3 in the Supplement. CKD was associated with higher odds of ICH in unadjusted analyses (odds ratio [OR], 3.35; 95% CI, 2.33-4.82; P < .001) and adjusted analyses (OR, 1.95; 95% CI, 1.35-2.89; P < .001) (Tables 2 and 3). In sensitivity analyses, ESKD was associated with even higher odds of ICH in unadjusted analyses (OR, 5.44; 95% CI, 2.85-10.37; P < .001) and adjusted analyses (OR, 2.93; 95% CI, 1.51-5.67; P < .001; Table 3). The association between CKD and ICH risk was not modified by race and ethnicity or hemorrhage location.
Table 1. Demographics and Clinical Characteristics of ERICH and UKB Participants With Available Data on Chronic Kidney Disease.
| Variable | No. (%) | |
|---|---|---|
| ERICH (n = 5868) | UKB (n = 502 536) | |
| Study design | Case/control study | Population study |
| Demographics | ||
| Age, mean (SD), y | 61.6 (14.0) | 56.5 (8.1) |
| Female | 2433 (41.5) | 273 402 (54.4) |
| Male | 3435 (58.5) | 229 134 (45.6) |
| Race/ethnicity | ||
| Asian | 0 | 11 456 (2.3) |
| Black | 1951 (33.2) | 8061 (1.6) |
| Hispanic/Latino | 1946 (33.2) | 0 |
| White | 1971 (33.6) | 472 725 (94.2) |
| Othera | 0 | 9396 (1.9) |
| Vascular risk factors | ||
| Diabetes, types 1 and 2 | 1463 (24.9) | 28 174 (5.6) |
| Hyperlipidemia | 2668 (45.9) | 96 234 (19.1) |
| Hypertension | 4013 (68.5) | 169 747 (33.8) |
| Smoking history | 2983 (50.9) | 226 049 (45.2) |
| Comorbidities | ||
| Atrial fibrillation | 474 (8.1) | 22 879 (4.6) |
| Congestive heart failure | 466 (8.0) | 10 056 (2.0) |
| Myocardial infarction | 962 (16.4) | 38 637 (7.7) |
| Alcohol use | 3232 (55.3) | 346 516 (69.2) |
| Medications | ||
| Antiplatelet use | 2540 (43.3) | 73 676 (14.7) |
| Anticoagulant use | 427 (7.3) | 5751 (1.1) |
| Antihypertensive use | 3134 (53.4) | 83 605 (16.6) |
| Measurements | ||
| Blood pressure, mean (SD), mm Hg | ||
| Systolic | 135.3 (19.9) | 137.8 (18.7) |
| Diastolic | 77.3 (13.4) | 82.2 (10.2) |
Abbreviations: ERICH, Ethnic/Racial Variations of Intracerebral Hemorrhage; UKB, UK Biobank.
Other includes participants who are of mixed or unknown ancestry.
Table 2. Distribution of Intracerebral Hemorrhages and Functional Outcomes Across Strata of Chronic Kidney Disease.
| ICH or outcome status | Chronic kidney disease, No./total No. (%) | Unadjusted P value | |
|---|---|---|---|
| No | Yes | ||
| Risk analyses | |||
| ERICH | |||
| ICH | 2789/2914 (95.7) | 125/2914 (4.3) | <.001 |
| No ICH | 2915/2954 (98.7) | 39/2954 (1.3) | |
| UK Biobank | |||
| ICH | 1255/1341 (93.6) | 86/1341 (6.4) | <.001 |
| No ICH | 491 831/501 195 (98.1) | 9364/501 195 (1.9) | |
| Outcome analyses | |||
| ERICH (n = 2521 ICH cases) | |||
| Good outcome (mRS score of 0-3) | 1204/1291 (93.3) | 87/1291 (6.7) | .003 |
| Poor outcome (mRS score of 4-6) | 1106/1230 (89.9) | 124/1230 (10.1) | |
| mRS score | |||
| 0 | 107/111 (96.4) | 4/111 (3.6) | .008 |
| 1 | 373/396 (94.2) | 23/396 (5.8) | |
| 2 | 376/404 (93.1) | 28/404 (6.9) | |
| 3 | 348/380 (91.6) | 32/380 (8.4) | |
| 4 | 434/472 (91.9) | 38/472 (8.1) | |
| 5 | 185/205 (90.3) | 20/205 (9.7) | |
| 6 | 487/553 (88.1) | 66/553 (11.9) | |
Abbreviations: ERICH, Ethnic/Racial Variations of Intracerebral Hemorrhage; ICH, intracerebral hemorrhage; mRS, modified Rankin Scale.
Table 3. Association Between Chronic Kidney Disease, End-stage Kidney Disease, and Risk of Intracerebral Hemorrhage.
| Study | Chronic kidney disease | End-stage kidney disease | ||||||
|---|---|---|---|---|---|---|---|---|
| Univariable model | Multivariable model | Univariable model | Multivariable model | |||||
| OR (95% CI) | P value | OR (95% CI) | P value | OR (95% CI) | P value | OR (95% CI) | P value | |
| ERICHa | 3.35 (2.33-4.82) | <.001 | 1.95 (1.35-2.89) | <.001 | 5.44 (2.85-10.37) | <.001 | 2.93 (1.51-5.67) | .001 |
| UKBb | 3.60 (2.89-4.48) | <.001 | 1.28 (1.01-1.62) | .04 | 7.00 (2.88-16.97) | <.001 | 3.21 (1.31-7.87) | .01 |
| Meta-analysis, random effects | 3.53 (2.93-4.26) | <.001 | 1.54 (1.02-2.32) | .04 | 5.93 (3.52-10.00) | <.001 | 3.03 (1.78-5.15) | <.001 |
| Heterogeneity, I2, % | 0 | .74 | 71 | .07 | 0 | .76 | 2 | .31 |
Abbreviations: ERICH, Ethnic/Racial Variations of Intracerebral Hemorrhage; OR, odds ratio; UKB, UK Biobank.
Multivariable model adjusted for age, sex, race and ethnicity, alcohol use, anticoagulant use, congestive heart failure, diabetes, hypertension, myocardial infarction, and smoking history.
Multivariable model adjusted for age, sex, race and ethnicity, atrial fibrillation, alcohol use, anticoagulant use, antiplatelet use, congestive heart failure, hyperlipidemia, hypertension, and myocardial infarction.
Replication in the UKB and Meta-analysis
Among 502 536 participants enrolled in the UKB, we identified 1341 participants with ICH (883 incident ICH [65.8%]; 458 prevalent ICH [34.2%]) and 501 195 controls (mean [SD] age, 56.5 [8.1] years, 273 402 female participants [54.4%]; 229 134 male participants [45.6%]). Study participants from the following races and ethnicities were included: 11 456 Asian (2.3%), 8061 Black (1.6%), 472 725 White (94.2%), and 9396 other race and ethnicity (1.9%) (Table 1). The prevalence of CKD was 6.4% (86 of 1341) in ICH cases and 1.9% (9364 of 501 195) in controls (unadjusted P < .001). Baseline demographics and comorbidities by diagnosis of CKD are shown in eTable 4 in the Supplement. CKD was associated with higher odds of ICH in unadjusted analyses (OR, 3.60; 95% CI, 2.89-4.48; P < .001) and adjusted analyses (OR, 1.28; 95% CI, 1.01-1.62; P = .04) (Table 3). ESKD was associated with even higher odds of ICH in unadjusted analyses (OR, 7.00; 95% CI, 2.88-16.97; P < .001) and adjusted analyses (OR, 3.21; 95% CI, 1.31-7.87; P = .01) (Table 3). Sensitivity analyses matching ICH cases and controls by their propensity score produced similar results for CKD in unadjusted analyses (OR, 1.31; 95% CI, 1.02-1.68; P = .04) and adjusted analyses (OR, 1.29; 95% CI, 1.01-1.66; P = .04). Random-effects models that pooled the estimates from the ERICH study and the UKB yielded consistent findings (Table 3).
Mendelian Randomization Analysis in the UKB
Mendelian randomization analyses supported a causal association between CKD and ICH. Among 487 320 participants with available genome-wide data in the UKB, 408 898 (83.9%) were of European ancestry. Among these, we identified 1078 ICH cases (mean [SD] age, 61.1 [6.7] years; 467 female participants [43.3%]; 611 male participants [56.7%]) and 407 820 controls (mean [SD] age, 56.9 [8.0] years; 220 545 female controls [54.1%]; 187 275 male controls [45.9%]). Genetically determined CKD was associated with higher odds of ICH when using both the inverse variance–weighted method (OR, 1.56; 95% CI, 1.13-2.16; P = .007) and the more conservative weighted-median approach (OR, 1.72; 95% CI, 1.06-2.82; P = .03) (Figure). There were no signs of horizontal pleiotropy, as expressed by the null results obtained using the MR-Egger (intercept OR, 1.00; 95% CI, 0.96-1.03; P = .81) and MR-PRESSO global tests (residual sum of squares, 23.22; P = .73) (Figure). Sensitivity analyses excluding CKD-related SNVs also associated with hypertension or diabetes yielded similar results (inverse variance–weighted approach: OR, 1.57; 95% CI, 1.12-2.19; P = .008) (Figure).
Figure. Mendelian Randomization (MR) Results for Genetically Determined Chronic Kidney Disease and Risk of Intracerebral Hemorrhage.
A, Association of single-nucleotide variations (SNVs) with chronic kidney disease and intracerebral hemorrhage. The blue summary line corresponds to the inverse variance–weighted slope. B, Forest plot of different MR methods. MR pleiotropy residual sum and outlier (PRESSO) corresponds to the outlier corrected estimate. IVW indicates inverse variance weighted; OR, odds ratio; WM, weighted median.
Chronic Kidney Disease and Functional Outcome After Intracerebral Hemorrhage in the ERICH Study
Among 3000 ICH patients enrolled in the ERICH study, 2521 (84.0%) had information on both CKD and outcome. The distribution of mRS scores 3 months after ICH is shown in Table 2. Baseline demographics and comorbidities by mRS are shown in eTable 5 in the Supplement. Three months after ICH, 1230 study participants (48.8%) had a poor functional outcome (ie, mRS 4-6). Participants with CKD experienced a higher risk of poor outcome (58.8% [124 of 211] vs 47.9% [1106 of 2310]; unadjusted P = .003) (Table 2). These findings were confirmed in multivariable regression analyses (OR, 2.05; 95% CI, 1.13-3.07; P < .001) (Table 4). The association between CKD and poor outcome was modified by hemorrhage location; CKD was associated with higher odds of poor outcome in lobar ICH (OR, 4.03; 95% CI, 1.69-10.12; P = .002) compared with nonlobar ICH (OR, 1.74; 95% CI, 1.10-2.76; P = .02). The association between CKD and poor outcome was not modified by race and ethnicity. Sensitivity analyses dichotomizing the mRS as 0 to 2 vs 3 to 6 and considering the mRS as a discrete variable yielded similar results for all analyses (Table 4).
Table 4. Association Between Chronic Kidney Disease and Poor Outcome After Intracerebral Hemorrhagea.
| Analysis | Univariable model | Multivariable model | ||
|---|---|---|---|---|
| OR (95% CI) | P value | OR (95% CI) | P value | |
| Dichotomized mRS | ||||
| 0-3 vs 4-6 | 1.55 (1.17-2.07) | .003 | 2.05 (1.13-3.07) | <.001 |
| 0-2 vs 3-6 | 1.67 (1.22-2.31) | .002 | 2.01 (1.32-3.12) | .001 |
| Ordinal mRS (range 0 to 6) | 1.99 (1.41-2.82) | <.001 | 1.89 (1.40-2.52) | <.001 |
Abbreviations: ICH, intracerebral hemorrhage; mRS, modified Rankin Scale; OR, odds ratio.
The final multivariable model included: age, sex, race/ethnicity, hypertension, hours to CT scan, ICH volume (natural log transformed), IVH volume (natural log transformed), and hemorrhage location.
Discussion
We report the results of a combined epidemiologic and genetic case-control study that evaluated the role of CKD in ICH. We found that CKD was independently associated with ICH risk and that the magnitude of this association was larger in analyses focused on ESKD. Mendelian randomization analyses found a significant association between genetically determined CKD and ICH risk, supporting a causal association between these 2 conditions.19 Finally, we found that CKD was independently associated with poor functional outcome after ICH. Of note, we did not find evidence of effect modification by race and ethnicity, suggesting that the observed association affected all evaluated ancestry groups equally.
Our findings provide important new evidence for a role of CKD in determining the risk of ICH. Previous studies have shown an association between CKD and neuroimaging markers of cerebral small vessel disease, the pathophysiological process underlying ICH.31,32,33 A recent genetic study has also shown an association between urine-albumin creatinine ratio and risk of ICH.34 Our results from the ERICH study suggest that CKD was associated with a higher risk of ICH in a large, racially and ethnically diverse population. This association was significant after adjusting for possible confounders, including hypertension and diabetes, suggesting an independent contribution of kidney impairment to ICH risk. Although residual confounding is possible, this association was also found in mendelian randomization analyses, which accounts for both measured and unmeasured confounding,19 strengthening the likelihood that CKD is an independent and causal contributor to ICH risk. Importantly, the independent contribution of CKD to ICH risk was present in both lobar and nonlobar hemorrhages, which for many patients have distinct pathophysiological processes—the former is mainly linked to cerebral amyloid angiopathy, whereas the latter is caused by hypertension-related small-vessel disease.35 Our study also confirms previous findings showing that CKD is associated with poor functional outcome after ICH.15,16 These results are in line with mounting evidence indicating that factors influencing ICH risk also determine ICH severity.36,37 Notably, we found that this association was modified by hematoma location. Patients with CKD have a higher risk of poor functional outcome after a lobar hemorrhage compared with a nonlobar hemorrhage.
Although our study provides important evidence to support a causal association between CKD and ICH, the pathways mediating this association remain to be identified. One possible pathway is the exacerbation of existing hypertension. The kidneys play a central role in regulating blood pressure and fluid balance via neurohormonal pathways and pressure natriuresis. These processes are disrupted in individuals with CKD, as diseased kidneys are unable to effectively excrete sodium.38,39 Kidney retention of sodium, in addition to increased sympathetic activity, may exacerbate existing hypertension and, therefore, increase the risk of ICH. Another possible pathway mediating the association involves the harmful biologic effects of uremia. Individuals with advanced CKD have higher levels of uremic toxins circulating in the blood,40 which have been associated with endothelial dysfunction,41 reduced platelet retention and aggregation,42 and impaired cerebral autoregulation.43 Endothelial damage by uremic toxins can contribute to a loss in vessel integrity leading to a higher risk of sustaining a brain hemorrhage. Combined with the effects on platelets and cerebral autoregulation, these conditions may serve as a catalyst for ICH. Finally, these results may point to a shared pathway contributing to kidney and cerebral manifestations of small-vessel disease in genetically susceptible individuals.44,45 Nitric oxide deficiency,46 hyperphosphatemia and arterial calcification,47 and deficiency of the Klotho protein48,49 have been proposed as direct mechanisms acting on both organs. Further research is needed to identify new biologic targets involved in both kidney and cerebral small-vessel disease.
Strengths and Limitations
One important strength of this study was the use of 2 large studies that jointly evaluated more than 4000 ICH cases, allowing both appropriate analytical power and independent replication of its main findings. Another relevant strength was the combination of complementary epidemiologic and genetic analyses that facilitate casual inference. Our study also had a few limitations. First, as a meta-analysis of 2 observational studies with different inclusion and exclusion criteria and ascertainment methods, the combined cohort is inherently heterogeneous, supporting our use of a random-effects model. There were significant differences in the baseline demographics and clinical characteristics between participants in the 2 studies. Participants in the ERICH study were more likely to be Black and Hispanic or Latino, older, and have a higher prevalence of medical comorbidities than those in the UKB. Second, our genetic analyses in the UKB were limited to individuals of European ancestry, which limits the generalizability of our results to other populations. Third, we did not have access to data on estimated glomerular filtration rate or CKD staging for most participants, limiting CKD diagnosis to physician- or self-report and preventing analyses that evaluated more granular categorization of CKD severity.
Conclusions
In conclusion, we report the results of a study that combined epidemiologic and genetic analyses to examine the role of CKD in ICH risk. We found that CKD was associated with a higher risk of ICH and that the magnitude of this association was larger in those with ESKD. Results from mendelian randomization analyses focused on genetically determined CKD suggest that this association was causal. We also found that CKD was associated with poor functional outcome after ICH. Future research is needed to elucidate the biologic mechanisms responsible for the associations of CKD with the risk and severity of ICH and to study the genetic associations reported here in Asian, Black, and Hispanic or Latino populations.
eTable 1. ICD-9 and ICD-10 Codes Used to Ascertain Chronic Kidney Disease and Intracerebral Hemorrhage
eTable 2. Independent Single Nucleotide Variations Associated With Chronic Kidney Disease
eTable 3. ERICH Baseline Demographics and Comorbidities by Diagnosis of Chronic Kidney Disease
eTable 4. UKB Baseline Demographics and Comorbidities by Diagnosis of Chronic Kidney Disease
eTable 5. Baseline Demographics and Comorbidities by 3-Month After ICH mRS
eFigure. Flowchart of Participant Inclusion and Exclusion Criteria
References
- 1.Qureshi AI, Mendelow AD, Hanley DF. Intracerebral haemorrhage. Lancet. 2009;373(9675):1632-1644. doi: 10.1016/S0140-6736(09)60371-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.van Asch CJ, Luitse MJ, Rinkel GJ, van der Tweel I, Algra A, Klijn CJ. Incidence, case fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: a systematic review and meta-analysis. Lancet Neurol. 2010;9(2):167-176. doi: 10.1016/S1474-4422(09)70340-0 [DOI] [PubMed] [Google Scholar]
- 3.Hemorrhagic Stroke Academia Industry (HEADS) Roundtable Participants . Unmet needs and challenges in clinical research of intracerebral hemorrhage. Stroke. 2018;49(5):1299-1307. doi: 10.1161/STROKEAHA.117.019541 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Qureshi AI, Palesch YY, Barsan WG, et al. ; ATACH-2 Trial Investigators and the Neurological Emergency Treatment Trials Network . Intensive blood-pressure lowering in patients with acute cerebral hemorrhage. N Engl J Med. 2016;375(11):1033-1043. doi: 10.1056/NEJMoa1603460 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hanley DF, Thompson RE, Rosenblum M, et al. ; MISTIE III Investigators . Efficacy and safety of minimally invasive surgery with thrombolysis in intracerebral haemorrhage evacuation (MISTIE III): a randomised, controlled, open-label, blinded endpoint phase 3 trial. Lancet. 2019;393(10175):1021-1032. doi: 10.1016/S0140-6736(19)30195-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Anderson CS, Heeley E, Huang Y, et al. ; INTERACT2 Investigators . Rapid blood-pressure lowering in patients with acute intracerebral hemorrhage. N Engl J Med. 2013;368(25):2355-2365. doi: 10.1056/NEJMoa1214609 [DOI] [PubMed] [Google Scholar]
- 7.Keep RF, Hua Y, Xi G. Intracerebral haemorrhage: mechanisms of injury and therapeutic targets. Lancet Neurol. 2012;11(8):720-731. doi: 10.1016/S1474-4422(12)70104-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Muntner P, Hardy ST, Fine LJ, et al. Trends in blood pressure control among US adults with hypertension, 1999-2000 to 2017-2018. JAMA. 2020;324(12):1190-1200. doi: 10.1001/jama.2020.14545 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lee M, Saver JL, Chang KH, Liao HW, Chang SC, Ovbiagele B. Low glomerular filtration rate and risk of stroke: meta-analysis. BMJ. 2010;341:c4249. doi: 10.1136/bmj.c4249 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Masson P, Webster AC, Hong M, Turner R, Lindley RI, Craig JC. Chronic kidney disease and the risk of stroke: a systematic review and meta-analysis. Nephrol Dial Transplant. 2015;30(7):1162-1169. doi: 10.1093/ndt/gfv009 [DOI] [PubMed] [Google Scholar]
- 11.Cherng YG, Lin CS, Shih CC, et al. Stroke risk and outcomes in patients with chronic kidney disease or end-stage renal disease: 2 nationwide studies. PLoS One. 2018;13(1):e0191155. doi: 10.1371/journal.pone.0191155 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yahalom G, Schwartz R, Schwammenthal Y, et al. Chronic kidney disease and clinical outcome in patients with acute stroke. Stroke. 2009;40(4):1296-1303. doi: 10.1161/STROKEAHA.108.520882 [DOI] [PubMed] [Google Scholar]
- 13.Sutherland LJ, Diprose WK, Wang MTM, Barber PA. Chronic kidney disease and outcome following endovascular thrombectomy for acute ischemic stroke. J Stroke Cerebrovasc Dis. 2020;29(4):104665. doi: 10.1016/j.jstrokecerebrovasdis.2020.104665 [DOI] [PubMed] [Google Scholar]
- 14.Khatri R, Afzal MR, Qureshi MA, Maud A, Huanyu D, Jose Rodriguez G. Preexisting renal failure increases in-hospital mortality in patients with intracerebral hemorrhage. J Stroke Cerebrovasc Dis. 2019;28(2):237-242. doi: 10.1016/j.jstrokecerebrovasdis.2018.07.036 [DOI] [PubMed] [Google Scholar]
- 15.Kim JK, Shin JJ, Park SK, Hwang YS, Kim TH, Shin HS. Prognostic factors and clinical outcomes of acute intracerebral hemorrhage in patients with chronic kidney disease. J Korean Neurosurg Soc. 2013;54(4):296-301. doi: 10.3340/jkns.2013.54.4.296 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bhattacharyya R, Miller LE. Preexisting renal failure increases in-hospital mortality in patients with intracerebral hemorrhage. J Stroke Cerebrovasc Dis. 2019;28(8):2360. doi: 10.1016/j.jstrokecerebrovasdis.2019.04.030 [DOI] [PubMed] [Google Scholar]
- 17.Woo D, Rosand J, Kidwell C, et al. The Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH) study protocol. Stroke. 2013;44(10):e120-e125. doi: 10.1161/STROKEAHA.113.002332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bycroft C, Freeman C, Petkova D, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018;562(7726):203-209. doi: 10.1038/s41586-018-0579-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Grover S, Del Greco M F, Stein CM, Ziegler A. Mendelian randomization. Methods Mol Biol. 2017;1666:581-628. doi: 10.1007/978-1-4939-7274-6_29 [DOI] [PubMed] [Google Scholar]
- 20.Gross BA, Jankowitz BT, Friedlander RM. Cerebral intraparenchymal hemorrhage: a review. JAMA. 2019;321(13):1295-1303. doi: 10.1001/jama.2019.2413 [DOI] [PubMed] [Google Scholar]
- 21.Vlasschaert MEO, Bejaimal SAD, Hackam DG, et al. Validity of administrative database coding for kidney disease: a systematic review. Am J Kidney Dis. 2011;57(1):29-43. doi: 10.1053/j.ajkd.2010.08.031 [DOI] [PubMed] [Google Scholar]
- 22.Roy L, Zappitelli M, White-Guay B, Lafrance JP, Dorais M, Perreault S. Agreement between administrative database and medical chart review for the prediction of chronic kidney disease G category. Can J Kidney Health Dis. 2020;7:2054358120959908. doi: 10.1177/2054358120959908 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Schnier C, Bush K, Nolan J, Sudlow C. Definitions of stroke for UK Biobank phase 1 outcomes adjudication. Accessed September 24, 2021. https://biobank.ndph.ox.ac.uk/showcase/ukb/docs/alg_outcome_stroke.pdf
- 24.Broderick JP, Adeoye O, Elm J. Evolution of the modified Rankin Scale and its use in future stroke trials. Stroke. 2017;48(7):2007-2012. doi: 10.1161/STROKEAHA.117.017866 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Howie B, Fuchsberger C, Stephens M, Marchini J, Abecasis GR. Fast and accurate genotype imputation in genome-wide association studies through prephasing. Nat Genet. 2012;44(8):955-959. doi: 10.1038/ng.2354 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.McCarthy S, Das S, Kretzschmar W, et al. ; Haplotype Reference Consortium . A reference panel of 64 976 haplotypes for genotype imputation. Nat Genet. 2016;48(10):1279-1283. doi: 10.1038/ng.3643 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Chou WC, Zheng HF, Cheng CH, et al. A combined reference panel from the 1000 Genomes and UK10K projects improved rare variant imputation in European and Chinese samples. Sci Rep. 2016;6(1):39313. doi: 10.1038/srep39313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Vittinghoff E, Bauer DC. Case-only analysis of treatment-covariate interactions in clinical trials. Biometrics. 2006;62(3):769-776. doi: 10.1111/j.1541-0420.2006.00511.x [DOI] [PubMed] [Google Scholar]
- 29.Wuttke M, Li Y, Li M, et al. ; Lifelines Cohort Study; V. A. Million Veteran Program . A catalog of genetic loci associated with kidney function from analyses of a million individuals. Nat Genet. 2019;51(6):957-972. doi: 10.1038/s41588-019-0407-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience. 2015;4:7. doi: 10.1186/s13742-015-0047-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Xu M, Zhang S, Liu J, et al. Kidney dysfunction is associated with a high burden of cerebral small vessel disease in primary intracerebral hemorrhage. Curr Neurovasc Res. 2018;15(1):39-46. doi: 10.2174/1567202615666180326094728 [DOI] [PubMed] [Google Scholar]
- 32.Tsai YH, Lee M, Lin LC, et al. Association of chronic kidney disease with small vessel disease in patients with hypertensive intracerebral hemorrhage. Front Neurol. 2018;9:284. doi: 10.3389/fneur.2018.00284 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ovbiagele B, Wing JJ, Menon RS, et al. Association of chronic kidney disease with cerebral microbleeds in patients with primary intracerebral hemorrhage. Stroke. 2013;44(9):2409-2413. doi: 10.1161/STROKEAHA.113.001958 [DOI] [PubMed] [Google Scholar]
- 34.Marini S, Georgakis MK, Chung J, et al. Genetic overlap and causal inferences between kidney function and cerebrovascular disease. Neurology. 2020;94(24):e2581-e2591. doi: 10.1212/WNL.0000000000009642 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Fisher CM. Lacunar strokes and infarcts: a review. Neurology. 1982;32(8):871-876. doi: 10.1212/WNL.32.8.871 [DOI] [PubMed] [Google Scholar]
- 36.Falcone GJ, Biffi A, Devan WJ, et al. ; GOCHA Investigators . Burden of blood pressure-related alleles is associated with larger hematoma volume and worse outcome in intracerebral hemorrhage. Stroke. 2013;44(2):321-326. doi: 10.1161/STROKEAHA.112.675181 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Leasure AC, Qureshi AI, Murthy SB, et al. Association of intensive blood pressure reduction with risk of hematoma expansion in patients with deep intracerebral hemorrhage. JAMA Neurol. 2019;76(8):949-955. doi: 10.1001/jamaneurol.2019.1141 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Aperia AC, Broberger CG, Söderlund S. Relationship between renal artery perfusion pressure and tubular sodium reabsorption. Am J Physiol. 1971;220(5):1205-1212. doi: 10.1152/ajplegacy.1971.220.5.1205 [DOI] [PubMed] [Google Scholar]
- 39.Soi V, Yee J. Sodium homeostasis in chronic kidney disease. Adv Chronic Kidney Dis. 2017;24(5):325-331. doi: 10.1053/j.ackd.2017.08.001 [DOI] [PubMed] [Google Scholar]
- 40.Duranton F, Cohen G, De Smet R, et al. ; European Uremic Toxin Work Group . Normal and pathologic concentrations of uremic toxins. J Am Soc Nephrol. 2012;23(7):1258-1270. doi: 10.1681/ASN.2011121175 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Six I, Flissi N, Lenglet G, et al. Uremic toxins and vascular dysfunction. Toxins (Basel). 2020;12(6):E404. doi: 10.3390/toxins12060404 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Boccardo P, Remuzzi G, Galbusera M. Platelet dysfunction in renal failure. Semin Thromb Hemost. 2004;30(5):579-589. doi: 10.1055/s-2004-835678 [DOI] [PubMed] [Google Scholar]
- 43.Sprick JD, Nocera JR, Hajjar I, O’Neill WC, Bailey J, Park J. Cerebral blood flow regulation in end-stage kidney disease. Am J Physiol Renal Physiol. 2020;319(5):F782-F791. doi: 10.1152/ajprenal.00438.2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Kelly DM, Rothwell PM. Does chronic kidney disease predict stroke risk independent of blood pressure: a systematic review and meta-regression. Stroke. 2019;50(11):3085-3092. doi: 10.1161/STROKEAHA.119.025442 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Kelly D, Rothwell PM. Disentangling the multiple links between renal dysfunction and cerebrovascular disease. J Neurol Neurosurg Psychiatry. 2020;91(1):88-97. doi: 10.1136/jnnp-2019-320526 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Baylis C. Nitric oxide deficiency in chronic kidney disease. Am J Physiol Renal Physiol. 2008;294(1):F1-F9. doi: 10.1152/ajprenal.00424.2007 [DOI] [PubMed] [Google Scholar]
- 47.Hruska KA, Mathew S, Lund R, Qiu P, Pratt R. Hyperphosphatemia of chronic kidney disease. Kidney Int. 2008;74(2):148-157. doi: 10.1038/ki.2008.130 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Hu MC, Shi M, Zhang J, et al. Klotho deficiency causes vascular calcification in chronic kidney disease. J Am Soc Nephrol. 2011;22(1):124-136. doi: 10.1681/ASN.2009121311 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Kuriyama N, Ozaki E, Mizuno T, et al. Association between α-Klotho and deep white matter lesions in the brain: a pilot case control study using brain MRI. J Alzheimers Dis. 2018;61(1):145-155. doi: 10.3233/JAD-170466 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. ICD-9 and ICD-10 Codes Used to Ascertain Chronic Kidney Disease and Intracerebral Hemorrhage
eTable 2. Independent Single Nucleotide Variations Associated With Chronic Kidney Disease
eTable 3. ERICH Baseline Demographics and Comorbidities by Diagnosis of Chronic Kidney Disease
eTable 4. UKB Baseline Demographics and Comorbidities by Diagnosis of Chronic Kidney Disease
eTable 5. Baseline Demographics and Comorbidities by 3-Month After ICH mRS
eFigure. Flowchart of Participant Inclusion and Exclusion Criteria