Genetic variants associated with sepsis-associated acute kidney injury.

BACKGROUND

Kidney dysfunction is a common complication in septic patients. Studies have identified numerous risk factors for sepsis-associated acute kidney injury (S-AKI), yet there is wide variability in the incidence even among patients with similar risk factors, suggesting the presence of additional uncharacterized risk factors, including genetic differences. The expansion of biobanks, advances in genotyping, and standardized diagnostic criteria have enabled large-scale, hypothesis-generating studies into the genetic mechanisms underlying S-AKI. We hypothesize that the genetic pathway behind S-AKI has overlapping mechanisms with key differences based upon the specific subtype of acute kidney injury (AKI).

METHODS

To test this hypothesis, we performed a genome-wide association study (GWAS) of S-AKI in three logistic regression models. Model 1, controlled for 1) age, 2) sex, 3) genotyping chip, and 4) the first five principal components. In Model 2, pre-sepsis baseline serum creatinine was added to the variables in Model 1. Finally, in Model 3, we controlled for the full range of patient, clinical, and ICU-related risk factors. Each of the 3-models were repeated in a pre-specified sensitivity analysis of higher severity S-AKI, defined as KDIGO Stage 2 or 3. We then compare associated variants and genes from our GWAS with previously published AKI sub-types and model other factors associated with S-AKI in our dataset.

FINDINGS

3,348 qualifying Sepsis-3 patients have been genotyped in our dataset. Of these patients, 383 (11.4%) developed Stage 1, 2, or 3 AKI (primary outcome) and 181 (5.4%) developed Stage 2 or 3 AKI (sensitivity analysis). The median age was 61 years (interquartile range (IQR): 51,69), 42% were female, and the increase in SOFA score (between 48-hours before to 24-hours after the onset of suspected infection) was 2 (2-3). No variants exceeded our threshold for genome-wide significance (P<5x10-8), however, a total of 13 variants exceeded the suggestive (P<1x10-6) threshold. Notably, rs184516290 (chr1:199814965:G:A), near the NR5A2 gene, chr1:199805801:T:TA, also near the NR5A2 gene, and rs117313146 (chr15:31999784:G:C), near the CHRNA7 gene, were associated with S-AKI at the suggestive level in all three models presented. Variants in the suppressor of fused homolog (SUFU) gene, previously shown to be correlated with renal function in bacteremic patients, consistently exceeded the P<0.05 threshold in our models.

CONCLUSIONS

While failing to identify any novel association for S-AKI at the level of genome-wide significance, our study did suggest multiple variants in previously characterized pathways for S-AKI including CHRNA7, NR5A2, and SUFU. We failed to replicate associations from multiple prior studies which may result from differences in how the phenotype was defined or, alternatively, limited genetic contribution and low heritability.