Race, Class, and AKI

If you are poor in this country that is hazardous to your health, if you are black or brown, too, and poor, it can be downright deadly.

President Barack Obama, 2007 Iowa Brown & Black Presidential Forum

In the United States, 46.5 million Americans are poor and 42.5 million do not have health insurance coverage. Income, used as a proxy for socioeconomic status, is a persistent and pervasive determinant of one’s health. Individuals with lower socioeconomic status are more likely to have diabetes, hypertension, and obesity; they are more likely to die younger. For decades, black and Hispanic adults had poverty rates at least twice as white adults or Asians. Only one in four adults (black or white) living below the 200% poverty level have health insurance.^1,2

In this issue of JASN, Grams et al. examine the effect of race on the incidence of AKI-related hospitalizations.³ They used self-identified black and white participants from the Atherosclerosis Risk in Communities (ARIC) study. The ARIC study is a prospective community-based observational cohort, which randomly selected and recruited participants since its inception in 1989 with special emphasis on race and geographic location.⁴ Indeed, at first glance, the authors report significantly higher incidence of AKI along racial lines, with a 30% higher risk in blacks during 13 years of follow-up. As we follow the sequential multivariable models in the article, the risk increases to 50% with the inclusion of age and sex, which reflects the elevated risk in blacks despite younger age. Interestingly, major mediators such as diabetes, hypertension, and obesity, all of which are more prevalent in blacks, only partly account for the risk attributed to race. However, when social indicators are included in the analysis, the association between race and AKI is significantly weakened.

Health inequity because of social status is unfair. Therefore, it is imperative to disentangle social determinants of health from those that are biologic. The availability and the proper use of granular, high-quality socioeconomic data enabled Grams et al. to avoid reaching a confounded conclusion regarding race and AKI. In the absence of individual-level income, education, and insurance information, meticulously collected in the ARIC study, researchers can resort to area-based socioeconomic data inferred from zip code or census geocoding. However, these methods can mask significant variation on the individual level.⁵

Considering the well established genetic predisposition for kidney disease along population ancestry lines, we share the authors’ astonishment at the lack of an association between AKI and high-risk APOL1 allele variants.^6–8 Individuals carrying two copies of high-risk APOL1 alleles are prone for CKD progression regardless of the level of BP control or angiotensin-converting enzyme inhibitor use.⁹ APOL1 renal disease risk alleles, which are virtually absent in whites, are present in >3 million Americans of West African ancestry as a result of positive natural selection; these same APOL1 allele variants confer resistance to Trypanosoma brucei that causes sleeping sickness.¹⁰

Why is there no association between APOL1 and AKI in the Grams et al. study? The mechanism by which APOL1 triggers CKD progression is uncertain. FSGS has been the predominant pattern of injury, and is localized to podocytes. However, APOL1 has also been localized to proximal tubular and arteriolar epithelia, meeting the biologic plausibility criterion to play a role in AKI.¹¹ A major disadvantage of this study in detecting an association between APOL1 and AKI is the study’s reliance on billing codes to ascertain outcome, as acknowledged by the authors. The concern is not only the low sensitivity, but also the lack of specifics regarding cause, type, and timing of AKI. Because of the tremendous heterogeneity present in AKI (compared with CKD), an ill-defined phenotype at best is hazardous (nondifferential misclassification will reduce study power) or downright deadly (differential classification will invalidate the results). Are the majority of AKI incidences in the study community acquired or in-hospital incidences? Are they related to surgery, trauma, or multisystem organ failure? Are these acute, acute on chronic, or acute diagnosis of CKD? Phenotypic heterogeneity has a direct detrimental effect on study power, and this is particularly pertinent to this study.¹² Despite the large cohort size and long follow-up duration, 785 AKI events were abstracted; 196 occurred in blacks. Subsequently, the estimated number of participants with two copies of APOL1 high-risk alleles and an AKI event is only 24 patients. Did phenotype ascertainment based on billing codes vary by race? AKI spectrum, environmental factors, and billing practices are more likely to vary between, rather than within, communities. In the ARIC study, the majority of black participants were sampled and recruited from Mississippi and North Carolina, whereas Maryland and Minnesota predominantly contributed white participants. The unfortunate coupling of race and geography in the ARIC design phase will taint racial disparity studies because of geographic confounding.⁴

In conclusion, further research is warranted to decipher APOL1 biology, its role in well defined AKI phenotypes, and its possible modifying effect on renal function recovery after injury. However, the immediate issue at hand is the 50% higher incidence of AKI (regardless of cause) reported in blacks as a result of the downright deadly aggregation of diabetes, hypertension, obesity, low income, and lack of insurance.

Disclosures

None.