Abstract
Rationale and Objective:
Progression of chronic kidney disease (CKD) in sickle cell disease (SCD) and its risk factors remain poorly defined. We identified characteristics associated with CKD as well as decline of estimated glomerular filtration rate (eGFR) and presence of proteinuria over time in adults with SCD.
Study Design:
Retrospective, observational study.
Setting & Participants:
SCD patients ≥ 18 years old in a single center from 2004–2013.
Predictors:
Baseline clinical and laboratory measures, comorbidities, SCD-related complications and severity of genotypes defined as severe (HbSS/HbSβ0 thalassemia) or mild (HbSC/HbSβ+ thalassemia).
Outcomes:
Presence at baseline of CKD, defined as eGFR < 90 mL/min/1.73 m2 or proteinuria as ≥1+ greater on urinalysis or current kidney transplant or dialysis; change in eGFR; and incident proteinuria.
Analytical Approach:
Logistic regression for baseline CKD. Linear mixed effects model for eGFR decline and generalized linear mixed effects model for proteinuria during the study period evaluating for interaction with time. Stratified by genotype severity.
Results:
Among 427 patients, 331 had ≥ 2 measures of creatinine. Over median follow-up of 4.01 (interquartile range, 1.66–7.19) years, the annual eGFR decline was 2.05 mL/min/1.73 m2 for severe genotypes (p<0.001) and 1.16 mL/min/1.73 m2 (p=0.02) for mild genotypes. At baseline, 21.4% of patients with severe genotypes had CKD versus 17.2% of those with mild genotypes. For severe genotypes, ACE-I/ARB use (OR, 6.10; 95% CI, 2.03–18.29; p=0.001) and avascular necrosis (OR, 0.40; 95% CI, 0.16–0.97; p=0.04) were associated with baseline CKD. Among those with mild genotypes, higher hemoglobin level was associated with a lower probability of CKD (OR per 1 g/dL greater hemoglobin level, 0.63; 95% CI, 0.43–0.93; p=0.02). Rate of eGFR decline was inversely related to hemoglobin levels (β= 0.46 [SE, 0.23]; p=0.04) within the severe genotype subgroup. No factors were identified to be associated with proteinuria over time.
Limitations:
Retrospective observational study, limited direct measures of albuminuria.
Conclusions:
Patients with SCD exhibit rapid decline in eGFR over time. Decline in eGFR is associated with markers of disease severity and associated comorbidities.
Keywords: sickle cell disease, chronic kidney disease (CKD), estimated glomerular filtration rate (eGFR), eGFR decline, proteinuria, sickle hemoglobin, genotype, hyperfiltration, thalassemia
INTRODUCTION
Sickle cell disease (SCD) is a genetic disorder whose pathology is attributed to expression of sickle hemoglobin (HbS). Polymerization of HbS in the deoxygenated state alters red cell structure and rheology and results in hemolysis.1, 2 Adhesion of cellular elements to the endothelium produces vaso-occlusive events. These events and the consequences of hemolysis lead to the clinical manifestations of SCD. The inner renal medulla, being an environment with low pH, high osmolality, low oxygen tension, and sluggish venous flow, is particularly susceptible to vaso-occlusion and hemolysis.3 As a result, chronic kidney disease (CKD) is common among adult SCD patients, with prevalence of up to 26–68%.2
Overall disease severity in sickle cell patients is characterized by the frequency and severity of vaso-occlusive events and associated comorbidities and is partly driven by the inherited genotype. HbSS disease (homozygous HbS disease) and HbSβ0 thalassemia are identified as severe genotypes; HbSC disease (double heterozygous inheritance of HbS and HbC) and HbSβ+ thalassemia are milder genotypes. Albuminuria, as in many diseases with glomerular involvement, serves as an early marker of progressive decline in kidney function and is more common among patients with severe genotypes.4, 5
Most studies of kidney disease in SCD have focused on prevalence, and the majority of risk factors associated with kidney disease have been identified in this context.2 Very little data presently exist regarding progression of kidney function in patients with SCD, particularly regarding which factors predict greater loss of estimated glomerular filtration rate (eGFR) over time.5–8
In a retrospective study of patients with SCD followed at a single center, clinical and laboratory data were collected to assess factors associated with baseline CKD, progressive eGFR decline, and development of incident proteinuria. Because of the previously observed differences in kidney disease between patients with severe versus mild genotypes, we hypothesized that we would observe differences in CKD risk between the two groups.
METHODS
Study design and participants
We reviewed the medical records of adult SCD patients (≥ 18 years-old) followed from 2004 to 2013. Patients with common genotypes of SCD were included (e.g. homozygous sickle cell disease [HbSS], hemoglobin SC disease [HbSC], sickle-β0-thalassemia [HbSβ0], sickle-β+-thalassemia [HbSβ+], etc). We excluded patients with history of bone marrow transplantation, systemic lupus erythematosus, HIV, or hepatitis B or C virus infection at baseline. We stratified patients into two groups based upon presumed genotype severity – “severe” included HbSS and HbSβ0 and “mild” included HbSC and HbSβ+. The baseline visit was defined as the first measure of serum creatinine (for eGFR calculation). Longitudinal evaluations of eGFR were obtained in the non-crisis, “steady state.” Only patients without a kidney transplant or dialysis requirement at baseline were eligible for evaluation of eGFR decline over time or incident proteinuria. The study was approved by our Institutional Review Board (14–2673) with a waiver of consent for analysis of deidentified data.
Outcomes and Covariates
GFR was estimated using the creatinine-based Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation.9 CKD was defined by a modified version of KDIGO (Kidney Disease Improving Global Outcomes) CKD guideline, which incorporates eGFR and levels of albuminuria.10 We defined CKD as either the presence of proteinuria (at least 1+ on dipstick urinalysis) or eGFR < 90ml/min/1.73 m2. This eGFR cutoff was used since creatinine-based estimates of eGFR may underestimate decreased kidney function in SCD patients.11, 12 Hyperfiltration was defined as eGFR > 130 mL/min /1.73 m2 for women and > 140 mL/min/1.73 m2 for men.13, 14 We assessed change in eGFR over time as a continuous measure and assessed the proportion of patients with rapid eGFR decline, defined as eGFR loss of >3.0 ml/min/1.73 m2 per year.15, 16 For these analyses, we included patients who had at least two eGFRs during the observation period. Proteinuria was assessed semi-quantitatively by urinalysis. We defined proteinuria as a binary variable: absent if urine dipstick results were 0 to trace and present if 1+ or greater. Hemoglobinuria was defined as dipstick urinalysis with blood, 1+ or greater, but with less than 5 red blood cells per high power field by microscopy. Baseline proteinuria status was assessed using urinalysis results conducted at the time baseline creatinine was measured. If urinalysis results were not available at the time the baseline creatinine was measured, results from the first urinalysis within 3 months thereafter were used.
Variables were obtained at study entry and at least yearly, while the patient was in the non-crisis, “steady state” until the end of the study period. We recorded routine demographics, clinical parameters, other clinical comorbidities, SCD-related complications and relevant treatment data. Time in the study was defined from first measure of creatinine (baseline) to last recorded measure during the study period.
Statistical analyses
All variables of interest were summarized by counts and percentages if categorical, or by median and interquartile ranges (IQR) if continuous. Analyses were stratified by “severe genotype and “mild genotype.” For CKD at baseline, all covariates of interest were first examined individually via age- and sex-adjusted logistic regression analysis. All covariates felt to be clinically relevant excluding those with severe missing data were evaluated in a multivariable model also adjusted for age and sex. The initial multivariable model for backward selection for the overall sample included baseline values for WBC count, hemoglobin, reticulocyte count, lactate dehydrogenase, total bilirubin, weight, acute chest syndrome, stroke, ongoing need for RBC transfusion, hydroxyurea therapy, leg ulcers, avascular necrosis, systolic blood pressure, angiotensin converting enzyme inhibitor/angiotensin receptor blocker (ACE inhibitor/ARB) therapy, and history of diabetes mellitus. We performed variable selection by backward elimination until all variables left in the model had p-values < 0.05.
To evaluate the eGFR change over time, we constructed a linear mixed effects model for the longitudinal eGFR assessments with random intercept and random time effect adjusted for baseline age and sex. Time from baseline was used as the time scale. Covariates were evaluated individually in an age- and sex-adjusted model by including an interaction term between the covariate and time. Variables chosen for more extensive multivariable modeling included proteinuria, systolic blood pressure, hemoglobin level, ACE-I/ARB therapy, hydroxyurea therapy, history of diabetes mellitus and baseline eGFR. Those covariates whose interaction terms with time were statistically significant (p<0.05) were considered to be associated with the outcome. Coefficients for interaction terms between the covariates and time represent the covariates’ effect on the trend of eGFR over time, and estimates for the coefficients for interaction terms are presented.
In addition to proteinuria at baseline, we assessed probability of proteinuria over the observation period. We utilized a generalized linear mixed effects model for the longitudinal proteinuria status for each covariate of interest with random intercept and random time effect adjusted for baseline age and sex. This model included an interaction term between the covariate and time from baseline. Covariates included in our multivariable model were systolic blood pressure, hemoglobin level, ACE-I/ARB therapy, hydroxyurea therapy, history of diabetes mellitus and baseline eGFR. Those covariates whose interaction terms with time reached statistical significance (p<0.05) were considered to be associated with the outcome. All analyses were conducted using SAS software, Version [9] for Windows Copyright © [2014] SAS Institute Inc., Cary, NC, USA.
RESULTS
Baseline characteristics
Four hundred and twenty-seven patients (HbSS: 269 [63.0%]; HbSC: 98 [23.0%]; HbSβ+: 30 [7.0%]; HbSβ0: 21 [4.9%]; HbSD: 2 [0.5%]; HbSE: 3 [0.7%]; and SHPFH [sickle cell trait in association with hereditary persistence of fetal hemoglobin]: 4 [0.9%]) met inclusion criteria. Fourteen patients were excluded due to comorbidities - 8 for hepatitis C virus, 2 for HIV, 1 for lupus, 3 for bone marrow transplantation. The median age of patients was 30 (IQR, 20–41) years, with more than half female (230 [54.0%]). Multiple baseline laboratory and clinical variables appeared to differ between “mild” (HbSC/HbSβ+) and “severe” (HbSS/HbSβ0) genotypes (Table 1). Individuals (n=9) with other forms of SCD (HbSD, HbSE, and SHPFH) were excluded from stratified analyses. The median eGFR for all patients was 134.6 (IQR, 113.1–152.1) ml/min/1.73m2. This was higher in patients with severe genotypes compared to those with mild genotypes (140.2 [IQR, 118.7–155.9] v. 123.7 [IQR, 106.3–137.3] ml/min/1.73m2). Among all subjects, 210 (49.3%) exhibited hyperfiltration by our definition. Proteinuria was present in 16.3% of individuals and was more prevalent in severe genotype subjects (21.5% v. 5.7%). Ferritin, specific gravity, hemoglobin F, height, direct and indirect bilirubin had severe missingness (in a range from 47.3% to 74.9%) and were excluded from multivariable analyses. All other included variables had less than 20% of observations missing.
Table 1.
All SCD* (All genotypes) | HbSS/HbSβ0-Thalassemia (Severe genotypes) | HbSC/HbSβ+-Thalassemia (Mild genotypes) | ||||
---|---|---|---|---|---|---|
No. with data | Value | No. with data | Value | No. with data | Value | |
Age (years) | 427 | 30 (20, 41) | 290 | 28 (21, 39) | 128 | 34 (20, 43) |
Female sex | 427 | 230 (54.0) | 290 | 159 (54.8) | 128 | 68 (53.1) |
Weight (Kg) | 418 | 68.1 (59.3, 80.2) | 283 | 65.1 (56.9, 74.8) | 126 | 75.5 (65.8, 92.9) |
Systolic BP (mm Hg) | 424 | 121 (111, 133) | 289 | 120 (110, 132) | 126 | 128 (116, 137) |
Diastolic BP (mm Hg) | 424 | 72 (65, 79) | 289 | 70 (63, 76) | 126 | 77 (70, 83) |
Hemoglobin (g/dL) | 426 | 9.7 (8.3, 11.2) | 290 | 9.0 (7.9, 10.1) | 127 | 11.7 (10.5, 12.9) |
Reticulocyte count (%) | 394 | 181.4 (115.9, 267.8) | 274 | 217.8 (142.3, 296.4) | 113 | 122.4 (86.9, 164.3) |
LDH (U/L) | 343 | 826 (613, 1155) | 233 | 957 (741, 1287) | 102 | 586 (503, 767) |
Total bilirubin (mg/dl) | 381 | 1.8 (1.1, 3.2) | 262 | 2.5 (1.5, 4.1) | 111 | 1.0 (0.8, 1.5) |
Direct bilirubin (mg/dl) | 108 | 0.1 (0.1, 0.2) | 79 | 0.1 (0.1, 0.2) | 27 | 0.1 (0.1, 0.2) |
Indirect bilirubin (mg/dl) | 107 | 1.9 (3.3, 1.1) | 78 | 2.5 (1.5, 3.8) | 27 | 1.0 (0.6, 1.2) |
WBC count (x103/uL) | 426 | 10.1 (7.6, 12.4) | 290 | 10.8 (8.4, 13.0) | 127 | 8.5 (6.5, 10.5) |
Ferritin (μg/L) | 195 | 224 (69, 761) | 131 | 298 (115, 1090) | 60 | 110 (43, 289) |
Hemoglobin F (%) | 187 | 4.6 (1.7, 9.4) | 140 | 6.1 (3.0, 10.5) | 43 | 0.9 (0.5, 2.0) |
Medical history | ||||||
Diabetes | 427 | 21 (4.9) | 290 | 7 (2.4) | 128 | 14 (10.9) |
Acute chest syndrome | 401 | 312 (77.8) | 275 | 236 (85.8) | 117 | 73 (62.4) |
Stroke | 387 | 52 (13.4) | 266 | 45 (16.9) | 112 | 7 (6.3) |
Leg ulcers | 347 | 52 (15.0) | 240 | 49 (20.4) | 99 | 3 (3.1) |
Avascular necrosis | 294 | 109 (37.1) | 199 | 73 (36.7) | 87 | 33 (37.9) |
Priapism* | 272 | 63 (40.7) | 108 | 49 (45.4) | 45 | 14 (31.1) |
Serum creatinine (mg/dl)† | 420 | 0.70 (0.60, 0.90) | 285 | 0.70 (0.60, 0.83) | 126 | 0.80 (0.70, 1.00) |
eGFR (ml/min/1.73m2)† | 420 | 135.0 (114.1, 152.1) | 285 | 140.7 (120.4, 156.1) | 126 | 124.7 (107.6, 137.4) |
Serum urea nitrogen (mg/dl)† | 389 | 8.0 (6.0, 11.0) | 270 | 8.0 (6.0, 11.0) | 110 | 9.0 (7.0, 11.0) |
Urine specific gravity† | 224 | 1.01 (1.01, 1.01) | 150 | 1.01 (1.01, 1.01) | 68 | 1.01 (1.01, 1.01) |
Proteinuria† | 238 | 38 (16.0) | 162 | 34 (21.0) | 70 | 4 (5.7) |
Hemoglobinuria† | 223 | 22 (9.9) | 149 | 21 (14.1) | 68 | 1 (1.5) |
Ongoing need for RBC transfusion | 427 | 17 (4.0) | 290 | 14 (4.8) | 128 | 3 (2.3) |
Hydroxyurea use | 425 | 136 (32.0) | 288 | 119 (41.3) | 128 | 17 (13.3) |
ACEi/ARB therapy | 425 | 50 (11.8) | 289 | 32 (11.1) | 127 | 18 (14.2) |
Continuous data presented as Median [IQR]; categorical data as count (percent).
Includes all assessed variants of SCD (HbSS, HbSC, HbSβ0, HbSβ+, HbSD, etc). Nine additional patients were included in the “All SCD” group who had other genotype variants.
eGFR= estimated glomerular filtration rate, ACEi = Angiotensin-converting enzyme inhibitor, ARB = angiotensin receptor blocker; WBC, White blood cell; BP, blood pressure; LDH, Lactate dehydrogenase; eGFR, _____; RBC, _______.
Assessed only in male patients.
For these data, 7 patients with kidney transplant or receiving dialysis were excluded.
CKD Prevalence
At baseline, 85 (19.9%) patients were noted to have CKD, of whom 7 (2%) were receiving dialysis (n=4; 1%) or had a kidney transplant (n=3; 1%). Data for proteinuria at baseline was missing in 188 patients. Of those with available data, 39 of 239 (16.3%) had proteinuria, 29 of whom had eGFR ≥90 ml/min/1.73m2. Forty-nine patients had eGFR <90 ml/min/1.73m2 of whom 10 had proteinuria. Of these patients, 29 had eGFR ≥60 and <90 ml/min/1.73m2, with proteinuria in 3 patients. Of the 20 patients with eGFR <60 ml/min/1.73m2, 7 had proteinuria. A greater proportion with severe genotypes had CKD (21.4%) versus those with mild genotypes (17.2%).
In initial age- and sex-adjusted covariate analyses among those with severe SCD, higher hemoglobin and history of avascular necrosis were associated with lower probability of CKD (Table 2). The probability of CKD was higher among those with hemoglobinuria and those receiving angiotensin converting enzyme inhibitor/angiotensin receptor blocker (ACE-I/ARB) therapy. Higher systolic (SBP) and diastolic blood pressure (DBP) were associated with prevalent CKD. Among mild genotype patients, higher ferritin, higher total bilirubin, and history of stroke had statistically significant associations with CKD. Higher hemoglobin was associated with lower probability of CKD. Diabetes mellitus and hydroxyurea therapy were not associated with prevalent CKD in either subgroup.
Table 2.
HbSS/HbSβ0-Thalassemia (Severe genotypes) | HbSC/HbSβ+-Thalassemia (Mild genotypes) | |||
---|---|---|---|---|
OR (95%CI) | P | OR (95%CI) | P | |
Hemoglobin | 0.73 (0.59, 0.89) | 0.003 | 0.63 (0.43, 0.93) | 0.02 |
LDH | 1.00 (1.00, 1.00) | 0.1 | 1.00 (1.00, 1.00) | 0.4 |
Reticulocyte count | 1.00 (1.00, 1.00) | 0.1 | 1.00 (0.99, 1.01) | 0.5 |
Hemoglobin F | 1.02 (0.96, 1.08) | 0.6 | *** | |
WBC count | 1.03 (0.94, 1.12) | 0.6 | 0.93 (0.79, 1.10) | 0.4 |
Ferritin | 1.00 (1.00, 1.00) | 0.9 | 1.00 (1.00, 1.00) | 0.02 |
Total bilirubin | 1.01 (0.88, 1.15) | 0.9 | 2.37 (1.07, 5.28) | 0.03 |
Direct bilirubin | 1.04 (0.56, 1.92) | 0.9 | *** | |
Indirect bilirubin | 1.03 (0.82, 1.29) | 0.8 | 4.32 (0.55, 33.96) | 0.2 |
Hemoglobinuria† | 11.41 (3.87, 33.67) | <0.001 | *** | |
Urine specific gravity | *** | *** | ||
Systolic BP | 1.03 (1.01, 1.05) | 0.01 | 0.99 (0.96, 1.02) | 0.5 |
Diastolic BP | 1.02 (1.00, 1.05) | 0.05 | 0.97 (0.93, 1.01) | 0.1 |
Weight | 0.99 (0.98, 1.01) | 0.5 | 0.97 (0.94, 1.01) | 0.1 |
History of diabetes | 3.48 (0.64, 18.95) | 0.1 | 0.99 (0.22, 4.47) | 0.9 |
History of stroke | 1.66 (0.74, 3.70) | 0.2 | 7.61 (1.05, 55.30) | 0.04 |
History of acute chest syndrome | 0.85 (0.36, 2.03) | 0.7 | 0.70 (0.20, 2.47) | 0.6 |
History of avascular necrosis | 0.33 (0.14, 0.78) | 0.01 | 0.69 (0.19, 2.55) | 0.6 |
History of leg ulcers | 0.67 (0.28, 1.58) | 0.4 | *** | |
ACEi/ARB therapy | 10.03 (4.04, 24.92) | <0.001 | 2.37 (0.67, 8.36) | 0.2 |
Hydroxyurea therapy | 1.02 (0.55, 1.89) | 0.9 | *** | |
Ongoing need for RBC transfusion | 1.08 (0.22, 5.26) | 0.9 | 11.02 (0.70, 174.51) | 0.09 |
Results from minimally adjusted (age and sex) logistic regression models for each covariate.
OR – odds ratio, CI – confidence interval; ACEi = Angiotensin-converting enzyme inhibitor, ARB = angiotensin receptor blocker; RBC, _______; LDH, _____; WBC, _____; BP, blood pressure; eGFR, estimated glomerular filtration rate.
Estimates were not calculable due to sample size.
In the final multivariable model for severe genotype patients (Table 3), ACE-I/ARB therapy retained an association with higher probability of CKD at baseline (OR, 6.10; 95% CI, 2.03–18.29; p=0.001) and avascular necrosis was associated with lower probability of CKD at baseline (OR, 0.40; 95% CI, 0.16–0.97; p=0.04). Among mild genotype patients, higher hemoglobin level was associated with lower probability of CKD at baseline (OR per 1 g/dL greater hemoglobin level, 0.63; 95% CI, 0.43–0.93; p=0.02).
Table 3.
HbSS/HbSβ0-Thalassemia (Severe genotypes) | HbSC/HbSβ+-Thalassemia (Mild genotypes) | |||
---|---|---|---|---|
OR (95%CI) | P | OR (95%CI) | P | |
Hemoglobin, per 1-g/dL greater | 0.63 (0.43, 0.93) | 0.02 | ||
History of avascular necrosis | 0.40 (0.16, 0.97) | 0.04 | ||
ACEi/ARB therapy | 6.1 (2.03, 18.29) | 0.001 |
Results from age- and sex- adjusted multivariable logistic regression model
OR – odds ratio, CI – confidence interval; ACEi = Angiotensin-converting enzyme inhibitor, ARB = angiotensin receptor blocker.
Decline in eGFR
For severe genotype patients, the median number of serum creatinine observations was 6 (IQR, 2–11) and for the mild genotype patients the median was 3 (IQR, 1–6). Among the 331 patients with at least two measures of creatinine, the median follow-up was 4.01 (IQR, 1.66–7.19) years. The rate of decline overall in eGFR was 1.82 mL/min/1.73 m2 per year adjusted for baseline age and sex. Patients with severe genotypes, compared with those with mild genotypes, demonstrated nominally steeper eGFR decline (−2.05 mL/min/1.73 m2 per year versus −1.16 mL/min/1.73 m2 per year; Figure 1) although the difference did not reach statistical significance (p=0.1). Rapid decline of eGFR (>3.0 ml/min/1.73 m2 per year) was noted in 103 (31.1%) patients: 80 (33.9%) with severe genotype and 21 (23.6%) with mild genotype. Over the course of the study, 6 (1.4%) additional patients required dialysis and none received transplantation after the baseline timepoint.
We examined factors that could potentially influence the rate of eGFR decline. Table 4 presents estimates of the coefficients of the interaction terms of examined covariates with time. Table S1 and S2 provide additional information on the estimates for the main time effect as well as the annual rate of decline. The annual rate of decline is reported for the mean value of continuous covariates or for the group with the condition of interest for binary covariates.
Table 4.
HbSS/HbSβ0-Thalassemia (Severe genotypes) | HbSC/HbSβ+-Thalassemia (Mild genotypes) | |||
---|---|---|---|---|
Estimate (SE) | P | Estimate (SE) | P | |
Hemoglobin | 0.48 (0.16) | 0.003 | 0.22 (0.30) | 0.5 |
LDH | −0.00097 (0.00052) | 0.07 | −0.0056 (0.0022) | 0.01 |
Reticulocyte count | −0.00088 (0.0023) | 0.7 | −0.0029 (0.0060) | 0.6 |
Hemoglobin F | 0.082 (0.056) | 0.1 | 0.27 (0.30) | 0.4 |
WBC count | 0.095 (0.075) | 0.2 | 0.016 (0.13) | 0.9 |
Ferritin | −0.00044 (0.00022) | 0.04 | −0.00093 (0.00072) | 0.2 |
Total bilirubin | 0.18 (0.12) | 0.1 | 0.89 (0.90) | 0.3 |
Direct bilirubin | −1.36 (5.43) | 0.8 | −6.75 (16.23) | 0.7 |
Indirect bilirubin | 0.34 (0.25) | 0.2 | 6.53 (3.42) | 0.06 |
Baseline eGFR | 0.018 (0.0082) | 0.03 | 0.0081 (0.020) | 0.7 |
Baseline proteinuria | −2.14 (0.91) | 0.02 | −1.68 (2.32) | 0.5 |
Hemoglobinuria | −0.73 (1.04) | 0.5 | *** | *** |
Urine specific gravity | 63.39 (86.25) | 0.5 | 148.50 (170.90) | 0.5 |
Systolic BP | −0.029 (0.017) | 0.08 | −0.086 (0.025) | <0.001 |
Diastolic BP | −0.017 (0.019) | 0.4 | −0.11 (0.044) | 0.01 |
Weight | 0.0028 (0.017) | 0.9 | −0.0071 (0.023) | 0.8 |
History of diabetes | −0.73 (1.42) | 0.6 | −4.25 (1.41) | 0.003 |
History of stroke | −1.54 (0.75) | 0.04 | 0.37 (3.51) | 0.9 |
History of acute chest syndrome | −0.42 (0.81) | 0.6 | −0.23 (1.14) | 0.8 |
History of avascular necrosis | −0.073 (0.60) | 0.9 | −2.13 (1.18) | 0.07 |
History of leg ulcers | 0.12 (0.65) | 0.9 | −0.30 (3.70) | 0.9 |
ACEi/ARB therapy | −0.52 (0.43) | 0.2 | 0.40 (0.72) | 0.6 |
Hydroxyurea therapy | 0.10 (0.28) | 0.7 | −0.26 (0.60) | 0.7 |
Ongoing need for RBC transfusion | 1.68 (1.38) | 0.2 | 1.24 (14.42) | 0.9 |
Estimates expressed in mL/min/1.73 m2 per year; for continuous covariates, estimate is per 1-unit increase. Results from minimally adjusted (age and sex) linear mixed effects model with random intercept and random time effect for each covariate.
ACEi = Angiotensin-converting enzyme inhibitor, ARB = angiotensin receptor blocker; RBC, _______; LDH, _____; WBC, _____; BP, blood pressure; eGFR, estimated glomerular filtration rate; SE, _____.
Too few patients to model.
In severe genotype patients, presence of proteinuria, higher ferritin, or history of stroke were associated with faster eGFR decline (Table 4). Lower baseline eGFR in this group was associated with more rapid eGFR loss. Maintenance of eGFR was associated with higher hemoglobin. For the annual rate of decline (Table S1), patients with proteinuria or history of stroke had the fastest eGFR decline (−3.51 ml/min/1.73m2 per year [p<.001] and −3.30 ml/min/1.73m2/year [p<0.001], respectively). Declines in eGFR were also significantly associated with ferritin, baseline eGFR, and hemoglobin (eGFR change associated with the mean value of these parameters was −1.29 (p=0.002), −2.10 (p<0.001) and −1.99 (p<0.001) ml/min/1.73m2 per year, respectively). Notably, diabetes mellitus did not have a statistically significant influence on eGFR decline in these patients.
Among mild genotype patients, higher lactate dehydrogenase, diabetes mellitus, higher systolic and diastolic blood pressures were associated with faster decline in eGFR. From the estimates for annual rates of decline (Table S2), mild genotype patients with diabetes had the fastest rate of decline in eGFR (−4.93 ml/min/1.73m2 per year, p<0.001). Annual rate of decline evaluated at the mean value of the corresponding continuous variables were as follows: lactate dehydrogenase (−1.23 ml/min/1.73m2 per year, p=0.02), systolic blood pressure (−1.09 ml/min/1.73m2 per year, p=0.01), and diastolic blood pressure (−1.01 ml/min/1.73m2 per year, p=0.03).
Multivariable analysis performed in the severe genotype group identified only baseline hemoglobin to be associated with eGFR change over time (0.46 [SE, 0.23] ml/min/1.73m2 increase per year for each 1 g/dl greater hemoglobin [p=0.04]). No variables were identified to have a significant influence on eGFR change over time in the mild genotype group.
Incident Proteinuria
Of those SCD patients assessed at baseline, 200 individuals had no overt proteinuria. Over the observation period, 84 (20%) of all patients exhibited proteinuria at some point. However, the risk of proteinuria did not change over time (β= −0.016 [SE, 0.04], p=0.7). Among severe genotype patients (Table 5), higher weight was associated with faster decrease in the log-odds of proteinuria over time for those weighing more than 61.9 kg, and a slower increase in the log-odds of proteinuria over time for those were less than 61.9 kg (β for interaction between weight and time, −0.0067 [SE, 0.0032], p=0.04). In multivariable analysis, no covariates had a statistically significant interaction with time. We did not evaluate those with mild genotypes due to small numbers of those with proteinuria (n=4).
Table 5.
HbSS/HbSβ0-Thalassemia (Severe genotypes) | ||
---|---|---|
Estimate (SE) | P | |
Hemoglobin | −0.046 (0.032) | 0.1 |
LDH | 0.000032 (0.000084) | 0.7 |
Reticulocyte count | 0.00041 (0.00039) | 0.3 |
Hemoglobin F | −0.011 (0.014) | 0.4 |
WBC count | −0.012 (0.014) | 0.4 |
Ferritin | 0.000021 (0.000052) | 0.7 |
Total bilirubin | 0.012 (0.024) | 0.6 |
Direct bilirubin | 1.89 (1.57) | 0.2 |
Indirect bilirubin | 0.041 (0.058) | 0.5 |
Baseline eGFR | −0.00001 (0.0015) | 0.9 |
Hemoglobinuria | −0.27 (0.22) | 0.2 |
Urine specific gravity | −32.68 (21.78) | 0.1 |
Systolic BP | −0.0028 (0.0028) | 0.3 |
Diastolic BP | −0.0049 (0.0030) | 0.1 |
Weight | −0.0067 (0.0032) | 0.04 |
Diabetes mellitus | −0.33 (0.24) | 0.2 |
History of stroke | 0.044 (0.12) | 0.7 |
History of acute chest syndrome | 0.45 (0.24) | 0.06 |
History of avascular necrosis | 0.079 (0.10) | 0.4 |
History of leg ulcers | 0.0069 (0.10) | 0.9 |
ACEi/ARB therapy | −0.11 (0.098) | 0.3 |
Hydroxyurea therapy | −0.13 (0.085) | 0.1 |
Ongoing need for RBC transfusion | −0.41 (0.65) | 0.5 |
Estimates expressed log-odds of proteinuria over time; for continuous covariates, estimate is per 1-unit increase.
Results from minimally adjusted (age and sex) generalized linear mixed effects model with random intercept and random time effect for each covariate.
ACEi = Angiotensin-converting enzyme inhibitor, ARB = angiotensin receptor blocker; RBC, _______; LDH, _____; WBC, _____; BP, blood pressure; eGFR, estimated glomerular filtration rate; SE, _____.
DISCUSSION
In this retrospective, longitudinal study of SCD patients, we report clinical and laboratory characteristics associated with prevalent CKD, progressive decline in eGFR, and incident proteinuria. Proteinuria and subsequent kidney disease is common among individuals with SCD.2 The present understanding of sickle cell nephropathy is that repeated sickling events contribute to ischemic injury and increased oxidant stress.1, 2, 11 Microinfarction and worsening hypoxia trigger additional prostaglandin release and subsequent vasodilation, increasing renal blood flow.1, 11 As a result, GFR first increases, manifesting in pathologic hyperfiltration and eventually producing albuminuria or overt proteinuria, glomerular sclerosis and subsequently, reduced eGFR.1, 2, 11, 17Endothelial dysfunction, driven by endothelin 1 elevation, increased soluble fms-like tyrosine kinase 1 (sFLT1), and hemolysis, also likely contributes to CKD.18, 19
Our data confirm the high prevalence of CKD in SCD, although our reported prevalence likely represents an underestimation as moderately increased albuminuria is not captured by dipstick urinalysis. We utilized a modified CKD definition (eGFR cutoff of <90 ml/min/1.73 m2) which could arguably overestimate CKD. However, with the typical early onset of hyperfiltration in SCD, an eGFR of 90 ml/min/1.73m2 would be unlikely to indicate truly preserved kidney function. This decision is supported by the noted bias of current creatinine-based equations in overestimating GFR in sickle cell patients.12, 20, 21 In multivariable analysis, the only laboratory measure associated with baseline CKD was hemoglobin level. The association of baseline CKD with lower hemoglobin levels may reflect disease severity or conversely may be due to reduced erythropoietin production subsequent to CKD.22 Hemoglobinuria, a reflection of ongoing intravascular hemolysis, was associated with CKD in initial age- and sex-adjusted analyses, consistent with prior studies demonstrating this relationship with APOL1 (apolipoprotein L1) risk variants as well as HMOX1 (heme oxygenase) variants.8, 23 However, we were unable to demonstrate this relationship in multivariable analysis.
Adult SCD patients exhibit accelerated decline in eGFR over time. The loss of eGFR in patients with severe genotypes in this cohort (−2.06 ml/min/1.73m2 per year) is greater than that expected for healthy individuals, which by report ranges from decreases of 0.36 to 1.21 ml/min/1.73m2 per year.24 The magnitude of eGFR loss in our cohort is similar to that reported for older individuals with diabetes, but lower than that recently reported by Asnani et al. in Jamaican patients with sickle cell anemia (HbSS)7, 24, who had a loss of GFR of 3.2 ml/min/1.73m2. However, the Jamaican cohort was smaller (41 individuals) and utilized radionucleotide measurement of GFR, which may have influenced their result. Notably, in the Jamaican cohort, higher hemoglobin was also associated with less GFR decline, and their results for the relationship with albuminuria suggested an influence on GFR decline.
We did not see an association between hemoglobin and eGFR decline in the mild genotype group, in which hemolysis was milder and hemoglobin better preserved. While other covariates were not identified in multivariable analyses, we can draw inferences from our initial age- and sex-adjusted analyses. In the severe genotype subgroup, ferritin and hemoglobin were associated with eGFR decline in relationships that suggest correlation with disease severity. Proteinuria had the expected relationship with eGFR decline in this group as well. Baseline eGFR seemed to have more importance in eGFR decline, perhaps driven by the high prevalence of hyperfiltration. Traditional CKD risk factors (blood pressure and diabetes mellitus) did not associate with eGFR decline in severe genotype patients.
Greater comorbid disease would suggest greater disease severity, and this relationship seems consistent with CKD risk. History of stroke demonstrated this relationship when examined in the severe genotype subgroup. The association of eGFR loss with stroke may be due to a shared pathophysiology, and we have previously described an association between albuminuria and stroke in SCD.25
The mild genotype subgroup demonstrated much slower eGFR decline (−1.16 ml/min/1.73m2 per year) than the severe genotype group. Among these patients, diabetes mellitus and blood pressure -- traditional CKD risk factors -- were associated with decline in eGFR. The only marker of disease severity in this group associated with eGFR decline was lactate dehydrogenase. Progression of CKD in this group, which was also older, may be driven by these underlying traditional risk factors.
Because hyperfiltration was common, decline in eGFR could potentially be attributed to regression to the mean. However, although this was not statistically significant, patients with hyperfiltration had nominally slower decline in eGFR, arguing against this possibility. Hyperfiltration has been observed in individuals with low muscle mass;26 unfortunately, we lacked data on body mass index to assess this possibility. Patients with hyperfiltration were younger and more likely to be female (Table S3).
In assessing the probability of proteinuria over time, we did not detect any specific relationships from the evaluated covariates. These analyses were hampered by the lack of systematic and quantitative assessment of proteinuria. These limitations underscore the need to collect these data in future studies. Specific treatments for proteinuria and CKD are lacking in SCD. Short-term studies have demonstrated ACE-I/ARB therapy reduces proteinuria.17, 27–29 Hydroxyurea may also reduce proteinuria, as demonstrated in case series, cross-sectional studies, and a recent prospective study.29–32 In our current analyses, we could not evaluate whether these agents afforded lesser eGFR loss due to lack of data on total treatment course or dose of these agents for the entire cohort. In fact, ACE-I/ARB therapy was more likely associated with baseline CKD in severe genotype patients. While counterintuitive, this association may exist because these agents were administered to those at high risk for kidney disease. In a recent evaluation of a subgroup of this cohort with proteinuria at baseline, ACE-I/ARB therapy was associated with slower decline in eGFR.32a
This study was retrospective and in a single center, and its findings must be viewed in context of its limitations. All laboratory data were measured as part of routine clinical practice rather than by specified protocols with standard intervals and measures, leading to variation in the number of data points available among subjects. Our use of single assessments of eGFR and proteinuria at baseline may have overestimated CKD prevalence typically defined by sustained evidence of reduced eGFR or proteinuria. Use of creatinine-based estimating equations may have also complicated our analyses. Accuracy of the CKD-EPI equation is reduced when eGFR > 60 ml/min/1.73m2, as much of this population demonstrates. However, studies of SCD patients that included patients with radionucleotide GFR measurements in the range of our population suggest that for creatinine-based equations the CKD-EPI equation is better than the MDRD (Modification of Diet in Renal Disease) Study equation.20, 21 We also evaluated longitudinal assessments of eGFR for CKD progression so we believe that noted trends still have validity. Future prospective studies should consider systematic use of cystatin C-based equations for this population. Proteinuria was assessed by dipstick and only reported in a semi-quantitative fashion. Further, the influence of therapeutic measures on kidney disease is difficult to ascertain given their employment via routine clinical practice. However, the large size of the cohort and the consistency in practice at our center allow us to draw the noted valuable inferences.
We find that loss of eGFR among SCD patients is accelerated, even when evaluating a cohort including individuals with hyperfiltration. SCD-associated kidney disease was associated with disease severity as measured by hemoglobin levels as well as by some markers of hemolysis and SCD-related complications, suggesting a shared underlying pathophysiology. Development of proteinuria is influenced by disease severity, and in those with mild SCD genotypes, diabetes mellitus and blood pressure may have a more substantial role. Our results highlight the need for large prospective studies with systematic measures of albuminuria and eGFR and inclusion of the varied SCD genotypes. Our observations are informative for such a study, which could also identify the utility of presently available interventions.
Supplementary Material
Support:
Direct funding for the study was provided by the National Heart, Lung, and Blood Institute, R01HL111659 (KIA, VKD, JC) and U01HL117659 (KIA, JC). We also acknowledge the assistance of the NC Translational and Clinical Sciences (NC TraCS) Institute, supported by the National Center for Advancing Translational Sciences (NCATS), National Institutes of Health, through Grant Award Number UL1TR001111. Funding agencies had no role in study design; collection, analysis, and interpretation of data; writing the report, or the decision to submit the report for publication
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Financial Disclosure: The authors declare that they have no relevant financial interests.
References:
- 1.Nath KA, Hebbel RP. Sickle cell disease: renal manifestations and mechanisms. Nat Rev Nephrol. 2015;11(3): 161–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ataga KI, Derebail VK, Archer DR. The glomerulopathy of sickle cell disease. Am J Hematol. 2014;89(9): 907–914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Scheinman JI. Sickle cell disease and the kidney. Nat Clin Pract Nephrol. 2009;5(2): 78–88. [DOI] [PubMed] [Google Scholar]
- 4.Guasch A, Navarrete J, Nass K, Zayas CF. Glomerular involvement in adults with sickle cell hemoglobinopathies: Prevalence and clinical correlates of progressive renal failure. J Am Soc Nephrol. 2006;17(8): 2228–2235. [DOI] [PubMed] [Google Scholar]
- 5.Gosmanova EO, Zaidi S, Wan JY, Adams-Graves PE. Prevalence and progression of chronic kidney disease in adult patients with sickle cell disease. J Investig Med. 2014;62(5): 804–807. [DOI] [PubMed] [Google Scholar]
- 6.Powars DR, Elliott-Mills DD, Chan L, et al. Chronic renal failure in sickle cell disease: risk factors, clinical course, and mortality. Ann Intern Med. 1991;115(8): 614–620. [DOI] [PubMed] [Google Scholar]
- 7.Asnani M, Serjeant G, Royal-Thomas T, Reid M. Predictors of renal function progression in adults with homozygous sickle cell disease. British journal of haematology. 2016;173(3): 461–468. [DOI] [PubMed] [Google Scholar]
- 8.Saraf SL, Zhang X, Kanias T, et al. Haemoglobinuria is associated with chronic kidney disease and its progression in patients with sickle cell anaemia. British journal of haematology. 2014;164(5): 729–739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9): 604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chapter 1: Definition and classification of CKD. Kidney International Supplements.3(1): 19–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Becker AM. Sickle cell nephropathy: challenging the conventional wisdom. Pediatr Nephrol. 2011;26(12): 2099–2109. [DOI] [PubMed] [Google Scholar]
- 12.Thompson J, Reid M, Hambleton I, Serjeant GR. Albuminuria and renal function in homozygous sickle cell disease: observations from a cohort study. Arch Intern Med. 2007;167(7): 701–708. [DOI] [PubMed] [Google Scholar]
- 13.Vazquez B, Shah B, Zhang X, Lash JP, Gordeuk VR, Saraf SL. Hyperfiltration is associated with the development of microalbuminuria in patients with sickle cell anemia. Am J Hematol. 2014;89(12): 1156–1157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Haymann JP, Stankovic K, Levy P, et al. Glomerular hyperfiltration in adult sickle cell anemia: a frequent hemolysis associated feature. Clin J Am Soc Nephrol. 2010;5(5): 756–761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rifkin DE, Shlipak MG, Katz R, et al. Rapid kidney function decline and mortality risk in older adults. Arch Intern Med. 2008;168(20): 2212–2218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Shlipak MG, Katz R, Kestenbaum B, et al. Rapid decline of kidney function increases cardiovascular risk in the elderly. J Am Soc Nephrol. 2009;20(12): 2625–2630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Falk RJ, Scheinman J, Phillips G, Orringer E, Johnson A, Jennette JC. Prevalence and pathologic features of sickle cell nephropathy and response to inhibition of angiotensin-converting enzyme. N Engl J Med. 1992;326(14): 910–915. [DOI] [PubMed] [Google Scholar]
- 18.Ataga KI, Brittain JE, Jones SK, et al. Association of soluble fms-like tyrosine kinase-1 with pulmonary hypertension and haemolysis in sickle cell disease. British journal of haematology. 2011;152(4): 485–491. [DOI] [PubMed] [Google Scholar]
- 19.Ataga KI, Derebail VK, Caughey M, et al. Albuminuria Is Associated with Endothelial Dysfunction and Elevated Plasma Endothelin-1 in Sickle Cell Anemia. PLoS One. 2016;11(9): e0162652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Asnani MR, Lynch O, Reid ME. Determining glomerular filtration rate in homozygous sickle cell disease: utility of serum creatinine based estimating equations. PLoS One. 2013;8(7): e69922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Yee MEM, Lane PA, Archer DR, Joiner CH, Eckman JR, Guasch A. Estimation of glomerular filtration rate using serum cystatin C and creatinine in adults with sickle cell anemia. Am J Hematol. 2017;92(10): E598–E599. [DOI] [PubMed] [Google Scholar]
- 22.Babitt JL, Lin HY. Mechanisms of anemia in CKD. J Am Soc Nephrol. 2012;23(10): 1631–1634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Saraf SL, Zhang X, Shah B, et al. Genetic variants and cell-free hemoglobin processing in sickle cell nephropathy. Haematologica. 2015;100(10): 1275–1284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Chapter 2: Definition, identification, and prediction of CKD progression. Kidney International Supplements.3(1): 63–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ataga KI, Brittain JE, Moore D, et al. Urinary albumin excretion is associated with pulmonary hypertension in sickle cell disease: potential role of soluble fms-like tyrosine kinase-1. European journal of haematology. 2010;85(3): 257–263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Han E, Lee YH, Lee BW, Kang ES, Cha BS. Pre-sarcopenia is associated with renal hyperfiltration independent of obesity or insulin resistance: Nationwide Surveys (KNHANES 2008–2011). Medicine (Baltimore). 2017;96(26): e7165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Aoki RY, Saad ST. Enalapril reduces the albuminuria of patients with sickle cell disease. Am J Med. 1995;98(5): 432–435. [DOI] [PubMed] [Google Scholar]
- 28.Foucan L, Bourhis V, Bangou J, Merault L, Etienne-Julan M, Salmi RL. A randomized trial of captopril for microalbuminuria in normotensive adults with sickle cell anemia. Am J Med. 1998;104(4): 339–342. [DOI] [PubMed] [Google Scholar]
- 29.McKie KT, Hanevold CD, Hernandez C, Waller JL, Ortiz L, McKie KM. Prevalence, prevention, and treatment of microalbuminuria and proteinuria in children with sickle cell disease. J Pediatr Hematol Oncol. 2007;29(3): 140–144. [DOI] [PubMed] [Google Scholar]
- 30.Fitzhugh CD, Wigfall DR, Ware RE. Enalapril and hydroxyurea therapy for children with sickle nephropathy. Pediatr Blood Cancer. 2005;45(7): 982–985. [DOI] [PubMed] [Google Scholar]
- 31.Laurin LP, Nachman PH, Desai PC, Ataga KI, Derebail VK. Hydroxyurea is associated with lower prevalence of albuminuria in adults with sickle cell disease. Nephrol Dial Transplant. 2014;29(6): 1211–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Bartolucci P, Habibi A, Stehle T, et al. Six Months of Hydroxyurea Reduces Albuminuria in Patients with Sickle Cell Disease. J Am Soc Nephrol. 2016;27(6): 1847–1853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32a.Effect of renin-angiotensin-aldosterone system blocking agents on progression of glomerulopathy in sickle cell disease. Thrower A, Ciccone EJ, Maitra P, Derebail VK, Cai J, Ataga KI. Br J Haematol. 2018. Nov 21. doi: 10.1111/bjh.15651. [Epub ahead of print] [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.