Development of a Genetic Risk Score for Albuminuria in Pediatric Patients with Sickle Cell Anemia

Introduction

Sickle cell anemia (SCA) causes progressive, age-dependent, kidney damage. Albuminuria is an early clinical biomarker for chronic kidney disease (CKD) and is associated with progressive glomerular injury. Identifying patients at risk for albuminuria is critical for early institution of renoprotective interventions to prevent progressive kidney injury. Developing a genetic risk score (GRS) to predict albuminuria is one approach to identifying high risk patients. Our prior data demonstrates that Apolipoprotein-L1 (APOL1 G1/G2) is associated with early development of albuminuria in children with SCA (PMID:30890594). However, this gene variant does not account for all kidney disease progression. Multiple other gene variants that modify SCA-severity have been associated with CKD in adults with SCA. We hypothesized that developing a genetic risk score (GRS) will allow for identification of pediatric patients at increased risk for development and progression of kidney disease. We aimed to develop a novel pediatric GRS that can identify pediatric SCA patients at high risk for albuminuria.

Methods

Gene allele variants in APOL1 G1/G2, HMOX1, BCL11A and α-thalassemia (α^-3.7 allele), were identified in 324 African American patients with SCA (HbSS or HbS⁰-thalassemia genotypes) enrolled in the Sickle Cell Clinical Research and Intervention Program (NCT02098863), a lifespan, longitudinal cohort study. The first urine albumin to creatinine ratio (ACR) collected after birth was analyzed as a continuous variable and dichotomized by a threshold of ACR ≥30mg/g creatinine as albuminuria. We defined a GRS³ based on established risk alleles in three genes: APOL1 (G1/G2), BCL11A (rs1427407), and α^-3.7(PMID:27658436). High risk was defined as patients with APOL1 (G1/G1, G2/G2 or G1/G2), without α^-3.7 and with wildtype BCL11A (G/G); low-risk was defined as patients negative for APOL1 G1/G2 and with α^-3.7 (either α^-3.7/αα or α^-3.7/α^-3.7) and the BCL11A T allele (either G/T or T/T). All other combinations were defined as intermediate risk. By adding HMOX1 (rs743811), C allele (TT/TC vs CC), we further refined GRS³ as GRS⁴ using the summation of the number of high-risk alleles of kidney outcomes for four variants and created four risk categories. The generalized linear regression model was used to associate additive dose-effect models for GRS³ and GRS⁴ with continuous ACR and albuminuria, adjusted for age, sex, hydroxyurea treatment, 5 principal components, and, if significant, the age-genotype interaction. ACR was log-transformed to approximate normality. As we included variants with known associations, we used a one-sided p-value for testing associations of individual variants, GRS³, and two-sided p-values for GRS⁴. P-values <0.05 were considered significant.

Results

In 324 individuals with SCA, those with albuminuria (34, 10.5%) appeared to be older than those without albuminuria (290, 89.5%) (mean 9.8 years [standard deviation ±4.4 years] vs. 8.6 [± 4.0], p=0.053). There were no significant differences between these groups by sex (female 58% vs. 49%, p=0.37), proportion on hydroxyurea (62% vs. 59%, p=0.85) or fetal hemoglobin (15.6 [± 10.61] vs. 14.8% [± 10.0], p= 0.81). All four individual genetic variants were associated with the first ACR (all p ≤ 0.028) and albuminuria (all p≤ 0.03). GRS³ was significantly associated with the first ACR (estimate=1.4, standard error [SE]=0.5, p=0.0035) and albuminuria (odds ratio [OR]=3.0, 95% CI: 1.2-7.5, p= 0.008). After adding HMOX1 to GRS³, GRS⁴ was more significantly associated with the first ACR (estimate=0.9, SE=0.3, p=0.0005) and albuminuria (OR=2.0, 95% CI: 1.2-3.4, p=0.003).

Summary

We developed a GRS³ based on three genetic variants associated with albuminuria in pediatric patients with SCA and stratified patients into different risk categories. Addition of a fourth genetic variant to create GRS⁴improved the model. Our findings will need to be validated in larger longitudinal studies to further examine the association between GRS⁴ and development of albuminuria and chronic kidney disease. Genetic prediction strategies using GRS has the potential to identify high risk SCA patients for whom renoprotective measures could prevent the development of albuminuria.