Searching for Genes That Matter in Acute Kidney Injury: A Systematic Review

Abstract

Background and Objectives: Identifying patients who may develop acute kidney injury (AKI) remains challenging, as clinical determinants explain only a portion of individual risk. Another factor that likely affects risk is intrinsic genetic variability. Therefore, a systematic review of studies was performed that related the development or prognosis of AKI to genetic variation.

Design, setting, participants, and measurements: MEDLINE, EMBASE, HuGEnet, SCOPUS, and Web of Science were searched for articles from 1950 to Dec 2007. Two independent researchers screened articles using predetermined criteria. Studies were assessed for methodological quality via an aggregate scoring system.

Results: The 16 included studies were of cohort or case-cohort design and investigated 35 polymorphisms in 21 genes in association with AKI. Fifteen gene-gene interactions were also investigated in four separate studies. Study populations were primarily premature infants or adults who were critically ill or postcardiac bypass patients. Among the studies, five different definitions of AKI were used. Only one polymorphism, APO E e2/e3/e4, had greater than one study showing a significant impact (P < 0.05) on AKI incidence. The mean quality score of 5.8/10 (range four to nine), heterogeneity in the studies, and the dearth of studies precluded additional meta-analysis of the results.

Conclusions: Current association studies are unable to provide definitive evidence linking genetic variation to AKI. Future success will require a narrow consensus definition of AKI, rigorous epidemiologic techniques, and a shift from a priori hypothesis-driven to genome-wide association studies.

Acute kidney injury (AKI) is a complex disorder manifested by a rapid loss of renal function that results in retention of metabolic waste products (1). The incidence of AKI has steadily increased over the past decade, and while mortality has fallen with advances in renal replacement therapy (RRT), AKI still confers significant morbidity and mortality (2–4). In-hospital mortality for patients who develop AKI ranges from 20% to 28%; for those requiring RRT, mortality ranges from 28% to 33% (3,4). As such, the ability to identify high-risk patients and potentially prevent AKI becomes crucial.

The current literature maintains that a patient's risk for AKI depends on a combination of acute insults and chronic co-morbidities. Acute risk factors include volume depletion, exposure to nephrotoxic agents, surgery, and sepsis (5–8). Chronic risk factors include age, chronic kidney disease (CKD), diabetes, and congestive heart failure (8,9). However, models using these traditional risk factors remain inadequate (5,6,10–12). Two patients with identical clinical risk factors often react differently to the same insult; one may suffer no harm while the other may require RRT. Furthermore, we remain unable to predict who will progress to chronic dialysis and who will recover.

Consequently, there are likely to be clinically unobservable risk factors that contribute to one's susceptibility to AKI. Our understanding of epithelial, vascular, and immune responses in kidney injury makes it likely that genetic variability in regulatory elements of these responses plays a major role in determining one's risk of AKI. This has been seen in other complex diseases, such as the discovery of a single nucleotide polymorphism (SNP) in complement factor H as a risk factor for macular degeneration (13). In AKI, genetic variation in inflammatory cytokines such as Tumor Necrosis Factor-alpha (TNF-α) and Interleukin-6 (IL-6) have recently been proposed as risk factors (11,14–17). This is due to the role of inflammatory mediators in the pathophysiology of AKI, especially with ischemia and sepsis (1,18,19). Genetic variations of vasomotor regulatory proteins, such as angiotensin converting enzyme (ACE) and endothelial nitric oxide synthase (eNOS), have also been investigated given the importance of vascular reactivity in the pathogenesis of AKI (11,18,20–22).

However, it remains difficult to ascertain which polymorphisms are truly associated with AKI (23). False positive reports are common in genetic association studies, and the plausibility of an association is highly dependent on the quality of the studies involved (24). Therefore, we conducted this systematic review to evaluate the quality of published studies on genetic associations with AKI and to ascertain if the current evidence demonstrates any polymorphism to be conclusively associated with AKI.

Materials and Methods

Literature Selection

Studies were selected by searching MEDLINE, EMBASE, HuGEnet, SCOPUS, and Web of Science for articles listed from 1950 until December 2007. Terms such as “acute kidney injury” and “acute renal failure” were used in combination with “genetic variation” and “polymorphism” as the search criteria. The exact criteria are listed in Table 1.

Table 1.

Search criteria

MEDLINE

Anuria, acute kidney failure, acute renal insufficiency, uremia, oliguria, creatinine, acute nephropathy, or acute kidney injury

AND

Gene frequency, genotype, phenotype, variation (genetics), gene expression, gene expression regulation, genes or nucleic acid regulatory sequences

EMBASE

Oliguria, creatinine blood level, creatinine, creatinine clearance, anuria, uremia, kidney tubule necrosis, acute kidney failure, acute kidney tubule necrosis, kidney dysfunction, acute nephropathy or acute kidney injury

AND

Allelism, genetic heterogeneity, genotype, human genetics, mutation, phenotype, population genetics, population genetic parameters, gene expression or gene

Web of Science, SCOPUS and HuGEnet

Acute renal insufficiency, acute nephropathy, acute kidney injury, acute tubular necrosis, acute kidney failure, kidney dysfunction, or acute renal failure

AND

Gene frequency, genetic frequency, genotype, phenotype, gene variation, genetic variation, gene expression, genetic expression, polymorphism, nucleic acid regulatory sequences, allelism, genetic heterogeneity, mutation or population genetics

Any original study that pertained to associations between AKI and human gene polymorphisms or variability was included. Exclusion criteria included the following: (1) studies with less than 20 subjects; (2) case reports and series; (3) animal studies; (4) outpatient studies; (5) studies where the etiology of AKI was hemolytic uremic syndrome/thrombotic thrombocytopenic purpura (HUS/TTP), ischemia/reperfusion of a kidney allograft, IgA nephropathy, or glomerulonephritis. The final criterion was included as we hoped to capture studies of AKI primarily from ischemia or nephrotoxic injury, which are the most common causes of AKI in hospitalized settings.

Titles and abstracts of articles found were screened, and articles of interest were selected for full article evaluation (Figure 1); bibliographies of relevant articles were also searched. Full articles were then reviewed by two researchers (JL and UP) independently and a final selection was made; disagreements were discussed and a consensus was reached. On one article pertaining to AKI in preeclampsia (25), discussion was inadequate to achieve consensus and a third researcher (CP) evaluated the study, which resulted in its exclusion.

Figure 1.

Selection of studies. AKI, acute kidney injury; HUS/TTP, hemolytic uremic syndrome/thrombotic thrombocytopenic purpura; IgA, immunoglobulin A nephropathy; GN, glomerulonephritis.

Data Abstraction

The following data was extracted from each study: first author, journal, year of publication, number of cases/controls, ethnicity, and the clinical setting in which AKI occurred. The gene polymorphisms or combinations thereof investigated by each study were recorded, and conclusions of the authors noted. Primary and secondary endpoints of each study were recorded as well. Study quality parameters were also collected. Each study then received an aggregate quality score using a system adapted from Clark et al. (26). Studies were scored in 10 categories (Table 2). Cohort studies were automatically given one point in the control group category to facilitate comparison; this was deemed acceptable, as cohort studies are better for studying high prevalence conditions such as AKI. Studies were scored as “good” if the score was 8 to 10, “fair” if the score was 5 to 7 and “poor” if the score was <4.

Table 2.

Scoring system for study quality used in this systematic review

Quality Criterion	Explanation	Scoring
Control Group	Was the control group equal or larger than the case group, and can it be replicated from the description given? Cohort studies were automatically given a point.	Yes = 1No = 0
Hardy-Weinberg Equilibrium	Were the case and control groups assessed for Hardy-Weinberg Equilibrium?	Yes = 1No = 0
Case Group	Can the case group be replicated from the description given, and was the disease state of interest adequately defined?	Yes = 1No = 0
Primer	Was the primer sequence used for genotyping or a reference to one provided?	Yes = 1No = 0
Reproducibility	Can the genotyping method be reproduced from the description given, and was the method validated via a second technique?	Yes = 1No = 0
Blinding	Were the researchers performing the genotyping blinded to the clinical status of the patient?	Blinded = 1Not blinded = 0
Power Calculation	Was a power calculation performed?	Yes = 1No = 0
Statistics	Were the major findings presented with well described tests of significance?	Yes = 1No = 0
Corrected Statistics	If a study examined two or more polymorphisms, were the statistics corrected for the increased risk of a false-positive finding?	Corrected = 1No correction = 0
Independent Replication	Was a second, confirmatory study performed?	Yes = 1No = 0

Results

The authors’ search returned 7273 unique articles, of which 43 were retrieved for full article review based upon titles and abstracts. Of these, 16 articles met the established eligibility criteria and were included in the analysis; initial inter-reviewer agreement on article selection was excellent (40/43). Overall, these 16 articles investigated 21 candidate genes and 35 separate polymorphisms in association with AKI. Four studies investigated the role of gene-gene interactions describing 15 different combinations and their association with AKI. Of the 21 genes studied, seven were involved in inflammatory pathways, five in oxidative stress or ischemia, four in vasomotor regulation, two in drug metabolism and one each in angiogenesis, coagulation regulation, and cholesterol metabolism.

Of the 16 included studies, 14 described genetic associations with AKI incidence (Table 3a) while two analyzed AKI outcomes (Table 3b). All studies had a cohort or case-cohort design; 11 were prospective studies. All study populations were of mixed gender; 12 studies investigated adults, while four studied neonates. Only nine of 16 studies reported the ethnicity of their populations. The clinical setting where AKI was studied fell primarily into two groups: postcardiopulmonary bypass (seven studies), and critically ill with or without sepsis (six studies). Of the remaining studies, two investigated all hospitalized patients with AKI, while one study investigated patients after hematopoietic cell transplant. AKI in all studies was caused by ischemia/reperfusion, sepsis, or nephrotoxic agents.

Table 3a.

Characteristics of studies examining genetic risk factors for acute kidney injury (AKI) incidence

Article (author, year)	Type of study	Site	Gene Polymorphism(s) studied	# of patients	# of AKI cases	Study Population	Clinical Setting	Ethnicity	% Caucasian	Definition of AKI or clinical variable studied	Quality Score
Banyasz 2006	Retrospective case cohort	SC	VEGF −2578 C/A, +405 G/C, −460 T/C	128	41	VLBW infants	Critically ill	NR		Modi 1999b	5
Chew 2000	Prospective cohort	SC	APO E e2/e3/e4	564	a	Adults	Post-CPB	NR		Delta Cr	7
Fekete 2003	Retrospective case cohort	SC	HSP72 + 1267 A/G HSP73 + 190 G/C	120	37	VLBW infants	Critically ill	Hungarian Ethnicity	100	Modi 1999	5
Gordon 2004	Prospective cohort	MC	TNF-a −238 G/A, −308 G/A LTA + 365 C/G, +249G/A TNFRSF1A + 1135 C/T, +36 A/G,−609 G/T TNFRSF1B + 1663 A/G, +676 T/G	213		Adults	Critically ill with sepsis	Caucasian	100	SOFA renal score	9
Guadino 2002	Prospective cohort	SC	IL-6 −174 G/C	111		Adults	Post-CPB	NR		Delta Cr	6
Isbir 2007	Prospective case cohort	NR	ACE I/D APO E e2/e3/e4 AGTR1 + 1166 A/C	248	54	Adults	Post-CPB	NR		Bellomo 2004 (RIFLE)c	4
Luo 2004	Prospective cohort	SC	Haptoglobin phenotype (Hp2-2 vs (Hp2−1 and Hp1−1))	148	27	Adults	Post-CPB	Chinese, Malay Indian	0	Nash 2002d	4
MacKensen 2004	Prospective cohort	NR	APO E e2/e3/e4	130		Adults	Post-CPB	Mixed	NR	Delta Cr	7
Nobilis 2001	Retrospective case cohort	SC	ACE I/D AGTR1 + 1166 A/C	110	42	VLBW infants	Critically ill	NR		Modi 1999	4
Stafford-Smith 2005	Prospective cohort	SC	ACE I/D Angiotensinogen + 842 T/C AGTR 1 + 1166 A/C eNOS + 894 G/T IL-6 −174 G/C, −572 G/C, −597 G/A TNF-a + 488 G/A, +376 G/A, −308 G/A APO E + 448 T/C (APOE e4) APO E + 586 C/T (APOE e2)	1671		Adults	Post-CPB	Caucasian, African American	88	Delta Cr	8
Sirgo 2004	Prospective cohort	SC	PAI-1 4G/5G	150	11	Adults	Post-CPB	Caucasian	100	Doubling of Cr	4
Treszl 2002	Retrospective case cohort	SC	TNF-a −308 G/A IL-1b + 3954 C/T IL-6 −174 G/C IL-10 −1082 G/A	92	36	VLBW infants	Critically ill with sepsis	NR		Modi 1999	7
Wattanathum 2005	Prospective cohort	SC	IL-10 haplotype (−592 C/A, −1082 A/G,+3367 G/A)	158		Adults	Critically ill with sepsis	Caucasian	100	# of days free of renal dysfunction	5
Woodahl 2007	Retrospective cohort	SC	ABCB1 + 1236C/T, +2677 G/T/A +3435 C/T,+1199 G/A	121	48	Adults	Post-HCT	NR		Doubling of Cr	6

^aStudies that looked at continuous outcome variables did not define cases.

^bModi 1999 definition of AKI: serum Cr > 120 umol and/or serum urea >9 mmol/L, and diuresis of 1.0 ml urine/kg/h.

^cBellomo 2004 (RIFLE) definition of AKI: increase in serum creatinine of 50% or greater.

^dNash 2002 definition of AKI: delta Cr > 0.5 mg/dl from baseline of 1.9 mg/dl or less, 1.0 mg/dl from baseline of 2.0 to 4.9 mg/dl and 1.5 mg/dl for patients with baseline >5.0 mg/dl. NR, not reported; SC, single center; MC, multicenter; VLBW, very low birth weight; CPB, cardiopulmonary bypass; Delta Cr, change in serum creatinine in mg/dl; HCT, hematopoetic cell transplantation.

Table 3b.

Characteristics of studies examining genetic risk factors for AKI outcome

Article (author, year)	Type of study	Site	Gene Polymorphism(s) studied	# of patients	# of AKI cases	Study Population	Clinical Setting	Ethnicity	% Caucasian	Definition of AKI or clinical variable studied	Quality Score
Perianayagam 2007	Prospective cohort	MC	NADPH Oxidase p22phox + 242C/T Catalase −262 C/T	200	200	Adults	Hospitalized	Mixed	90	Dialysis or mortality	7
Jaber 2004	Prospective cohort	MC	TNF-a −308 G/A IL-10 −1082 A/G	61	61	Adults	Hospitalized	Mixed	93	Mortality and recovery of renal function	5

MC, multicenter.

The definition of AKI varied greatly in the included studies, as five different definitions of AKI were used (9,27,28). In addition, four studies opted to report only the change in creatinine. Two studies used hard endpoints of dialysis or mortality. The quality of studies was generally mediocre; mean quality score was 5.8/10 (range 4 to 9). Four studies were scored as “poor,” 10 were “mediocre,” while only two were of “good” quality. Primarily, studies were lacking in the areas of reproducibility, blinding, power calculations, corrected statistics, and independent replication (Figure 2) (26).

Figure 2.

Percentage of all studies reviewed that achieved each of the 10 quality criteria.* *Based on Clark MF, Baudouin SV: A systematic of the quality of genetic association studies in human sepsis (26). Please see Table 2 for full descriptions of each criteria.

Table 4 describes the specific polymorphisms studied and the authors’ conclusions. In summary, nine polymorphisms (NADPH Oxidase p22phox + 242C/T; Haptoglobin Hp2-2, Hp2-1 or HP1-1; Heat Shock Protein 72 (HSP72) +1267A/G; Apolipoprotein E (APO E) e2/e3/e4; Angiotensin Converting Enzyme (ACE) I/D; TNF-α −308G/A; IL-6 −174G/C; Interleukin-10 (IL-10) −1082G/A; Vascular Endothelial Growth Factor (VEGF) −2578C/A) were found to have significant associations in individual studies (14,16,20,29–33). Only one polymorphism, APO E e2/e3/e4, had an association with AKI demonstrated in multiple studies.

Table 4.

Gene characteristics

Gene	Polymorphism	Chromosome	Functional significance	Type of polymorphism	Study	Significant association with AKI	Comparison	Association Variable	Magnitude of association	Significance of association
APO E	e2/e3/e4	19q13	Cholesterol metabolism	SNP	Isbir 2007	Yes	e4 vs. non-e4	AKI	OR = 0.18b	P = 0.002
	e2/e3/e4				MacKensen 2004	No	e4 vs. non-e4	Delta Cr		P = 0.82
	e2/e3/e4				Chew 2000	Yes	e4 vs. e2, e3	Delta Cr		P = 0.038, P = 0.015
	e2/e3/e4				Stafford-Smith 2005	No	e4 vs. non-e4	Delta Cr		P = 0.009a,c
	e2/e3/e4				Stafford-Smith 2005	No	e2 vs. non-e2	Delta Cr		P = 0.32d
NADPH Oxidase p22phox	+242C/T	16q24	Oxidative stress	SNP	Perianayagam 2007	Yes	CT/TT versus CC	Dialysis or mortality	OR = 2.11	P = 0.01
Catalase	−262 C/T	11p13	Oxidative stress	SNP	Perianayagam 2007	No	CT/TT vs. CC	Dialysis or mortality	OR = 1.05	P = 0.86
ACE	I/D	17q23	Vasomotor regulation	Intron deletion	Isbir 2007	Yes	ID/DD vs. II	AKI	OR = 2.37b	P = 0.021
	I/D				Stafford-Smith 2005	No	II/ID vs. DD	Delta Cr		P = 0.004a
	I/D				Nobilis 2001	No	I vs. D allele	AKI		NS
AGT	+842 T/C	1q42	Vasomotor regulation	SNP	Stafford-Smith 2005	No	CC/CT vs. TT	Delta Cr		P < 0.0001a,c; 0.99d
AGTR1	+1166 A/C	3q21-25	Vasomotor regulation	SNP	Isbir 2007	No	AA vs. AC/CC	AKI	OR = 1.09b	P > 0.05
	+1166 A/C				Stafford-Smith 2005	No	CC/CA vs. AA	Delta Cr		P = 0.84c
	+1166 A/C				Nobilis 2001	No	C vs. A allele	AKI		NS
	+1166 A/C				Li 2007	No	AA vs. AC/CC	AKI	OR = 0.34b	P = 0.116
eNOS	+894 G/T	7q36	Vasomotor regulation	SNP	Stafford-Smith 2005	No	TT/TG vs. GG	Delta Cr		P = 0.17c; 0.04a,d
ABCB1	+1236 C/T	7q21	Drug metabolism	SNP	Woodahl 2007	No	TT vs. CC/CT	AKI	OR = 1.9	P = 0.21
	+2677 G/T/A				Woodahl 2007	No	TT vs. GG/GT	AKI	OR = 1.6	P = 0.34
	+2677 G/T/A				Woodahl 2007	No	TT vs. TA/GA	AKI	OR = 2.5	P = 0.33
	+3435 C/T				Woodahl 2007	No	TT vs. CT/CC	AKI	OR = 1.1	P = 0.82
	+1199 G/A				Woodahl 2007	No	GG vs. GA/AA	AKI	OR = 3.2	P = 0.14
CYP3A5	a1/a3	7q21	Drug metabolism	Splice variant	Woodahl 2007	No	a3/a3 vs. a1/a3 + a1/a1	AKI	OR = 0.8	P = 0.68
HSP72	+1267 A/G	6p21	Ischemia tolerance	SNP	Fekete 2003	Yes	GG vs. GA/AA	AKI	OR = 3.17	P < 0.01
HSP73	+190 G/C	11q24	Ischemia tolerance	SNP	Fekete 2003	No	G vs. C allele	AKI		NS
Haptoglobin	Hp2-2, Hp2-1, or Hp1-1	16q22	Fe metabolism, Anti-oxidant	Phenotype	Luo 2004	Yes	Hp2-2 vs. Hp1-1/2-1	AKI	OR = 5.4	P = 0.03
TNF-α	−308 G/A	6p21	Proinflammatory	SNP	Jaber 2004	Yes	AA/AG vs. GG	Mortality	HR = 2.47	P = 0.04
	−308 G/A				Stafford-Smith 2005	No	AA/AG vs. GG	Delta Cr		P = 0.17c
	−308 G/A				Gordon 2004	No	A vs. G allele	Renal SOFA score		NS
	−308 G/A				Treszl 2002	No	A vs. G allele	AKI		NS
	−238 G/A				Gordon 2004	No	A vs. G allele	Renal SOFA score		NS
	+376 G/A				Stafford-Smith 2005	No	AA/AG vs. GG	Delta Cr		NS
	+488 G/A				Stafford-Smith 2005	No	AA/AG vs. GG	Delta Cr		NS
LTA	+365 C/G	6p21	Proinflammatory	SNP	Gordon 2004	No	C vs. G allele	Renal SOFA score		NS
	+249 G/A				Gordon 2004	No	G vs. A allele	Renal SOFA score		NS
IL-1b	+3954 C/T	2q14	Proinflammatory	SNP	Treszl 2002	No	T vs. C allele	AKI		NS
IL-6	−174 G/C	7p21	Inflammation modulator	SNP	Stafford-Smith 2005	No	GG vs. CC/CG	Delta Cr		NS
	−174 G/C				Guadino 2002	Yes	GG vs. CC/CG	Delta Cr		P < 0.0001
	−174 G/C				Treszl 2002	No	C vs. G allele	AKI		NS
	−572 G/C				Stafford-Smith 2005	No	GG vs. CC/CG	Delta Cr		P < 0.0001a,c
	−597 G/A				Stafford-Smith 2005	No	AA/AG vs. GG	Delta Cr		NS
TNFRSF1A	+1135 C/T	12p13	Anti-inflammatory	SNP	Gordon 2004	No	C vs. T allele	Renal SOFA score		NS
	+36 A/G				Gordon 2004	No	A vs. G allele	Renal SOFA score		NS
	-609 G/T				Gordon 2004	No	G vs. T allele	Renal SOFA score		NS
TNFRSF1B	+1663 A/G	1p36	Anti-inflammatory	SNP	Gordon 2004	No	A vs. G allele	Renal SOFA score		NS
	+676 T/G				Gordon 2004	No	T vs. G allele	Renal SOFA score		NS
IL-10	−1082 G/A	1q31-32	Anti-inflammatory	SNP	Jaber 2004	Yes	GG/GA vs. AA	Mortality	HR = 0.36	P = 0.03
	−1082 G/A				Treszl 2002	No	G vs. A allele	AKI		NS
VEGF	−2578 C/A	6p12	Angiogenesis	SNP	Banyasz 2006	Yes	AA vs. AC/CC	AKI	OR = 0.2	P = 0.021
	−460 T/C				Banyasz 2006	No	CC vs. CT/TT	AKI		NS
	+405 G/C				Banyasz 2006	No	CC vs. GC/GG	AKI		NS
PAI-1	-675 4G/5G	7q21	Coagulation activation	SNP	Sirgo 2004	No	4G/4G vs. 4G/5G + 5G/5G	AKI		NS

^aNot significant after adjustment for multiple comparisons;

^bcalculated from published data;

^cfor Caucasian subgroup;

^dfor African-American subgroup.

NR, not reported; SC, single centre; MC, multicenter; VLBW, very low birth weight; HCT, hematopoietic cell transplantation; Cr, creatinine; NS, not significant. Italics represent results from multivariable analysis.

Apolipoprotein E

Isbir et al. (20) found that in patients undergoing coronary artery bypass grafting (CABG), carriers of the APO E e4 allele had a decreased risk of AKI compared with non-APO E e4 patients (unadjusted odds ratio [OR] = 0.18, P = 0.002). Chew et al. (30) found the same result in a similar population, as those with the e4 allele had a smaller postoperative change in creatinine compared with those with e3 and e2 alleles (P = 0.015 versus e3, P = 0.038 versus e2), even after adjustment for preoperative creatinine, age, bypass time, hypertension, diabetes, and ejection fraction. Mackensen et al. (34), examining a similar group of 130 CABG patients, also found this allele to have a protective effect but only after adjustment for ascending aortic atheroma burden; without adjustment, no association was found. Finally, a large study of 1671 cardiac surgical patients by Stafford-Smith et al. (11) found no significant association between APO E alleles and the degree of change of postoperative creatinine in Caucasian or African American populations.

Oxidative Stress Genes

In a study of 200 patients with mixed-cause AKI, Perianayagam et al. (33) examined the association of polymorphisms in NADPH Oxidase p22phox and Catalase, two enzymes involved with the regulation of reactive oxygen species, with dialysis and mortality. Individuals with the T allele in the NADPH Oxidase p22phox gene possessed a greater risk of dialysis or mortality (unadjusted OR = 2.11). This remained true after adjusting for race, gender, age, APACHE II score and CKD. No association was demonstrated between the Catalase gene and AKI. This study was unique in that it is one of two studies that examined associations with firm outcomes, i.e., dialysis or mortality (16,33).

Vasomotor Regulation Genes

For the ACE I/D polymorphism, three studies were identified, of which only one found significant associations with AKI. Isbir et al. found that patients with the ACE D allele exhibited increased risk of AKI following CABG (unadjusted OR = 2.37, P = 0.021). However, Stafford-Smith and colleagues also examined the ACE I/D polymorphism and found no association with AKI (11). Another study of 110 very low birth weight (VLBW) infants also found no association (21).

No significant associations were found with other genes involved with vasomotor regulation, including Angiotensinogen (AGT), Angiotensin Receptor 1 (AGTR1), and eNOS.

Inflammatory and Anti-Inflammatory Genes

Six different studies examined a total of 15 polymorphisms in seven genes involved in inflammatory and anti-inflammatory pathways. Jaber et al. (16), in a study of 61 patients with AKI requiring hemodialysis, found that high producers of TNF-α (−308 A-allele carriers) possessed an increased risk of death after adjustment for APACHE II score (adjusted hazard ratio [HR] = 2.5, P = 0.04). However, three other studies searched for an association between this polymorphism and AKI incidence, but found none (11,15,17).

The same study by Jaber et al. (16) also found that IL-10 intermediate/high producers (−1082 G-allele carriers) had a decreased risk of death after adjustment for the multiple organ failure score (adjusted HR = 0.36, P = 0.03). Treszl et al. also investigated this polymorphism in VLBW infants but found no association with AKI incidence (17).

The IL-6 −174G/C polymorphism was investigated in three studies, of which only Guadino et al. (14) found a significant association. This study found that in patients undergoing CABG, IL-6 −174GG, carriers had significantly higher elevations in perioperative creatinine versus non-GG carriers (P < 0.0001). However, Stafford-Smith et al. (11) found no such association in a similar population, nor did Treszl et al. (17) in VLBW infants.

Other Genes

Luo et al. (32), in a population of 148 CABG patients, found that Haptoglobin 2-2 phenotype was associated with an increased risk of AKI (OR = 5.4, P = 0.03). In the HSP72 gene, Fekete et al. (31) found that VLBW infants homozygous for the G allele were at increased risk for AKI (OR = 3.17, P < 0.01). This same group also found that for the VEGF −2578 C/A polymorphism, VLBW infants homozygous for the A allele were protected against AKI (OR = 0.2, P = 0.021) (29).

None of the remaining polymorphisms investigated were found to be significantly associated with AKI (35,36).

Gene-Gene Interactions

There were four studies that investigated gene-gene interactions and their association with AKI. The combinations studied were mostly those that augmented inflammatory or down-regulated anti-inflammatory pathways (Table 5). Wattanathum et al. (37) found that the CGG haplotype involving three separate polymorphisms in the IL-10 gene (−592 C/A, +734 A/G and + 3367 G/A) was associated with a greater degree of AKI in patients with sepsis from pneumonia. They postulated that this genotype is associated with lower anti-inflammatory IL-10 production, thus causing increased renal dysfunction.

Jaber et al. (16) investigated whether the combination of pro-inflammatory alleles from the TNF-α −308G/A and IL-10 −1082 G/A polymorphisms, namely the TNF-α −308 AA and IL-10 −1082 AA/AG genotypes, were associated with an increased risk for dialysis or death. They found that patients with these genotypes had an elevated risk for dialysis or death after adjustment for APACHE II score (adjusted HR = 5.17, P = 0.005). Treszl et al. (17) also investigated the TNF-α −308G/A polymorphism, but in combination with the IL-6 −174G/C polymorphism. VLBW infants who had a combination of the IL-6 −74C allele and the TNF-α −308A allele were at increased risk for developing AKI (OR = 6.07, P < 0.01).

Finally, Stafford-Smith et al. (11) investigated the interaction between AGT + 842T/C and IL-6 −572G/C polymorphisms and found the combination of C-alleles in both genes to be significantly associated with AKI (P < 0.0001) in Caucasians. This study also found many interactions that trended toward significance (Table 5); however, only this association remained significant after adjustment for multiple comparisons (adjusted P = 0.032).

Table 5.

Gene combinations

Study	Gene polymorphism combinations	Significant association with AKI	Magnitude of association	Significance of association
Wattanathum 2005	IL-10 −592 C/A, +734 A/G and + 3367 G/A	Yes		P = 0.024
Stafford-Smith 2005	eNOS + 894 G/T and AGTR1 + 1166 A/C	No		P = 0.006a,c
	IL-6 −572 G/C and TNF-α −308 G/A	No		P = 0.05a,c
	AGT + 842 T/C and IL-6 −572 G/C	Yes		P ≤ 0.0001c
	APO E e4 and AGT + 842 T/C	No		P = 0.03a,c
	AGTR1 + 1166 A/C and APO E e4	No		P = 0.02a,c
	eNOS + 894 G/T and ACE I/D	No		P = 0.006a,d
	AGT + 842 T/C and APO E e2	No		P = 0.03a,d
Jaber 2004	IL-10 −1082 G/A and TNF-α −308 G/A	Yes	HR = 5.72	P = 0.004
Treszl 2002	TNF-α −308 G/A and IL-6 −174 G/C	Yes	OR = 6.07b	P < 0.01
	TNF-α −308 G/A and IL-1b + 3954 C/T	No		NS
	TNF-α −308 G/A and IL-10 −1082 G/A	No		NS
	IL-1b + 3954 C/T and IL-6 −174 G/C	No		NS
	IL-1b + 3954 C/T and IL-10 −1082 G/A	No		NS
	IL-10 −1082 G/A and IL-6 −174 G/C	No		NS

^aNot significant after adjustment for multiple comparisons;

^bcalculated from published data;

^cfor Caucasian subgroup;

^dfor African-American subgroup.

Italics represent results of multivariable analysis.

Discussion

While there have been several descriptive reviews on this topic (38,39), to the authors’ knowledge this is the first systematic review of genetic determinants of AKI. In summary, it was found that there is no single polymorphism that can be conclusively described as a risk factor in AKI. The general dearth of studies, the lack of confirmatory studies, and their overall mediocre quality led to this conclusion.

Of the 35 individual polymorphisms whose association with AKI has been studied, only APO E e2/e3/e4 had a significant association in more than one study (20,30). Recent evidence points to a regulatory role of APO E in inflammatory responses and may thus be affecting a given patient's susceptibility to AKI (18,40). However, the positive association found in two studies (20,30) was contradicted by the results from the largest, highest quality study by Stafford-Smith et al. (11). In this case, having to correct for multiple comparisons decreased this study's power to find an effect, for while the p-values found by this study were <0.05, they were not statistically significant after adjustment. While a future, higher-powered study may find a role of APO E in AKI, current evidence is inadequate to make such a claim.

Of the gene-gene interactions studied, the association between the TNF-α −308 AA and IL-10 −1082 AA/AG genotypes with dialysis or mortality demonstrated by Jaber et al. is the most promising (16). While this association has yet to be replicated, the individual polymorphisms had small but significant associations with poor outcomes, and the combined gene-gene interaction was associated with a much higher risk than either polymorphism individually. This group also demonstrated phenotypic differences in ex vivo production of TNF-α and IL-10 between genotypic groups, which lend biologic plausibility to this linkage.

These results however do not belie the major finding of this review that, thus far, findings in this field have been inconsistent and contradictory. There is significant interstudy heterogeneity of results, as seen here with APO E. Causes for such heterogeneity include the large number of comparisons performed, which significantly increases the number of associations found by chance (41). The case of APO E clearly shows this, for while Stafford-Smith et al. demonstrated p-values <0.05 in a number of genes, statistical significance was not achieved due to the large number of comparisons performed (11). Population stratification from ethnic admixture, variable linkage disequilibrium, and population specific gene-gene or gene-environment interactions are also potential sources of heterogeneity. Therefore, it is recommended for authors to independently verify their results before publication (42). This is especially necessary given that initial studies tend to show more impressive associations than subsequent research, a problem that is particularly endemic in genetic association research (24).

The lack of a gold standard outcome has also contributed to study heterogeneity (43). The authors found five different definitions of AKI in the studies found by their search; such variability increases the number of spurious associations and makes interstudy comparison difficult. Use of well-defined criteria, such as the RIFLE or AKIN criteria, will decrease heterogeneity and facilitate comparisons in future investigations; (27,44) specifically, using stage 2 of the AKIN criteria would allow for the creation of case groups that are both highly specific and have a well-defined phenotype. Also, as evidence accumulates for the role of novel biomarkers such as Neutrophil Gelatinase-associated Lipocalin (NGAL) in AKI, these biomarkers can be used to further increase specificity (45). Studying outcomes such as dialysis or mortality would also alleviate this problem; however, due to the rarity of such events, an adequately powered study examining such outcomes may not be feasible.

The authors’ study has limitations in several regards. This systematic review is vulnerable to several types of bias, the first being publication bias (46). The literature has an inherent bias toward the publication of studies that find positive results (47). No attempt was made to retrieve unpublished data, as it is unknown whether this would have corrected the bias or worsened it. Therefore, despite using systematic methods, this may be a nonrepresentative sample of existing genetic association studies of AKI.

Another potential limitation is analysis reporting bias caused by researchers who report only a portion of their analyses (48). This is especially salient in the reporting of gene-gene interactions, where many combinations may have been analyzed for association but never reported.

In conclusion, the present understanding of AKI suggests that genetic heterogeneity in pathways that regulate vascular and inflammatory responses to injury provide a plausible explanation for individual variability in susceptibility to AKI. Continued efforts in this field are important, as finding genetic risk factors will allow clinicians to identify patients at risk and implement preventive therapies. Identification of culprit genes may also elucidate the true pathophysiology of AKI. Although some genes show promise, existing candidate gene studies have thus far been unable to find conclusive evidence to confirm any association.

An alternative approach will be to use genome-wide association (GWA) studies. This technique has been harnessed to successfully investigate genetic vulnerabilities in complex diseases such as macular degeneration and diabetes (13,49). GWA studies have the advantages of searching for candidate polymorphisms unhindered by previous hypotheses and greater power in demonstrating the effects of gene-gene interactions or high-risk haplotypes (50). This is especially important for complex disorders, where increased genetic risk may be from several different polymorphisms acting together. A pitfall is that GWA studies have unprecedented potential for false-positive results given the immense number of statistical tests performed; this problem can be somewhat alleviated by multistage designs and stringent requirements for statistical significance (50). Nevertheless, the success of future work will likely depend on harnessing this revolutionary technique and the application of a true consensus definition of AKI.

Disclosures

None.