REVIEWS

Paradigms of acute kidney injury

in the intensive care setting
John A. Kellum1 and John R. Prowle2,3
Abstract | Acute kidney injury (AKI) is a heterogeneous clinical syndrome that has multiple
aetiologies, variable pathogenesis and diverse outcomes. However, these heterogeneities are not
reflected in current approaches to the diagnosis and, to some degree, treatment of AKI. For
example, congestive heart failure and dehydration can produce identical changes in serum
creatinine level and urine output (parameters that are used to define AKI); however, they differ
vastly in their physiological contexts and demand completely opposite treatments. AKI is often
still considered to be a homogeneous clinical entity, which implies a uniform pathogenesis and a
well-defined prognosis. As a consequence, efforts to find effective AKI treatments have been
hampered by a lack of clear clinical classifications for various types of AKI. In addition,
subclassification of AKI into subclinical phenotypes — for example, on the basis of protein
biomarkers and other in vitro diagnostics that take into account disease aetiology and underlying
pathogenesis — might be necessary to develop therapeutic approaches that effectively target
the widely differing pathomechanisms of AKI. In this Review, we discuss the major subtypes of AKI
that are associated with sepsis, major surgery, renal hypoperfusion and nephrotoxin exposure —
situations that are typically seen in the intensive care setting. We consider differences and
similarities in their phenotype, pathogenesis and outcomes and how this information might be
used to guide treatment.

1
Center for Critical Care
Nephrology, Department
of Critical Care Medicine,
University of Pittsburgh
School of Medicine and
University of Pittsburgh
Medical Center,
3347 Forbes Avenue,
Suite 220, Pittsburgh,
Pennsylvania 15213, USA.
2
Adult Critical Care Unit,
Department of Renal and
Transplant Medicine,
The Royal London Hospital,
Whitechapel Road,
London E1 1BB, UK.
3
William Harvey Research
Institute, Barts and the
London School of Medicine
and Dentistry, Queen Mary
University of London,
Charterhouse Square,
London EC1M 6BQ, UK.
Correspondence to J.A.K.
kellumja@upmc.edu
doi:10.1038/nrneph.2017.184
Published online 22 Jan 2018
Acute kidney injury (AKI) — a loose collection of
syndromes characterized by a sudden decrease in esti­
mated glomerular filtration rate (eGFR) — is estimated
to affect 2–3 people per 1,000 individuals in the USA1.
However, this approximation is likely to be an under­
estimate owing to the silent nature of AKI: the incidence
could be as high as 15 in 1,000 adults per year in the
UK2. Moreover, older individuals are disproportion­
ately affected by AKI: the rate of AKI in 2011 among
US Medicare patients aged 66–69 years was 14.9 per
1,000 individuals, increasing with age to 18.8 per 1,000
individuals aged 70–74, 26.4 per 1,000 individuals aged
75–79, 35.9 per 1,000 individuals aged 80–84, and
49.6 per 1,000 individuals aged ≥85 (REF. 3). The con­
sequences of AKI are no less substantial in older indi­
viduals. The relative risk of death attributable to AKI
is similar across age groups, despite lower overall sur­
vival and lower odds of renal function recovery among
older individuals4.
Evidence from the past few years also suggests that
even mild forms of AKI (for example, a 50% increase in
serum creatinine) are associated with an increased risk
of hospital mortality 5,6. Therefore, patients who develop
any degree of AKI have an increased risk of death that
is at least in part attributable to AKI itself, although the
reasons for this association are not fully understood.
Translation of these clear epidemiological observa­
tions to clinical practice has proved complex, however,
and current management remains based on minimizing
the risk of developing AKI and providing supportive
care7. This division between the increased recognition
of AKI and the lack of substantial developments in AKI
therapy has led to criticism of current concepts and defi­
nitions as being inadequate to individualize therapy and
improve outcomes8 (BOX 1). In the same way that can­
cer therapies were advanced initially by understanding
malignancy in general, but subsequently revolutionized
by characterization of differences among molecular
subtypes, a universal strategy is unlikely to be success­
ful in the treatment of AKI given the widely differing
underlying mechanisms; instead, targeted treatment will
probably be required to achieve consistent clinical bene­
fits. The concept that AKI is a syndrome of rising serum
creatinine levels and/or oliguria caused by multifactorial
systemic insults raises the question to what extent a com­
mon pathophysiological pathway exists following diverse

NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 217

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Key points Sepsis is characterized by systemic vasodilation,

• Acute kidney injury (AKI) is not a single disease but a loose collection of syndromes
characterized by an abrupt decrease in glomerular filtration rate
• Broad classifications of AKI — according to the dominant aetiology (such as sepsis or
nephrotoxicity) as opposed to pseudo-anatomical categories (for example, prerenal
or intrarenal) — demonstrate more consistent relationships with pathophysiology and
can improve therapeutic approaches
• Broad aetiology-based classification may still be insufficient, as heterogeneity could
exist at the molecular level; however, whether this heterogeneity leads to differences
in response to therapy is not yet known
• Protein biomarkers and other in vitro diagnostics are likely to have a role in both
characterizing clinical phenotypes of AKI and aiding the discovery of new
endophenotypes, with the ultimate goal of matching subsets of patients to effective
therapeutic approaches
injuries. Conversely, it is unclear to what extent differing
aetiologies of AKI constitute distinct paradigms (that is,
distinctive phenotypes caused by differing pathophysio­
logical mechanisms), which demand specific treatments
and confer unique prognoses. In this Review, we discuss
the differing causes and clinical contexts of AKI that
have been identified9, including a discussion of their
common and distinguishing features. We attempt to
identify paradigms that might inform our therapeutic
approach and draw parallels and contrasts between these
settings in terms of their epidemiology, pathophysiology
and outcomes. Given the wide range of aetiologies that
can cause AKI, we focus on AKI that typically occurs in
the intensive care setting.
Sepsis-associated AKI
Epidemiology. Sepsis is the leading condition associ­
ated with AKI in the hospital and intensive care unit
(ICU)5,9–11. Similar to AKI, sepsis is a disease of elderly
individuals and is substantially associated with risk
of death in the hospital12. Globally, sepsis is found in
almost 50% of patients with severe AKI in the ICU9,13.
An analysis of US health-care databases including
nearly 200,000 individuals across seven states identi­
Inflammatory mediators
fied AKI in 22% of patients with diagnostic codes for
A wide array of substances that severe sepsis, which was associated with 38.2% mortal­
mediate and modulate various ity 12. In Europe, a 51% rate of AKI has been reported in
aspects of inflammation. These patients with sepsis, with an associated ICU mortality of
substances include vasoactive
41%14. Thus, sepsis is likely to be at least a contributing
proteins (for example,
bradykinin, histamine), factor to AKI in many cases, and an understanding of
complement and molecules the mechanistic basis of septic AKI is essential for the
involved in coagulation (for development of improved therapeutic approaches for
example, plasmin, thrombin). individuals with septic AKI.
Cytokines (for example, TNF,
IL‑6, IL‑10), chemokines (for
example, IL‑8) and eicosanoids Pathophysiology. Sepsis is characterized by a systemic
(for example, leukotrienes, inflammatory response to infection. The pathogenesis
prostaglandins) are the major of septic AKI is still not fully understood; however, pro­
mediators of inflammation
found differences are apparent between experimental
during infection. Although
some molecules are often models of ischaemic AKI (see below) and related clini­
labelled pro-inflammatory cal settings. Post-mortem studies of patients with septic
(for example, IL‑6) and others AKI revealed heterogeneous tubular cellular injury with
anti-inflammatory (for example, apical cellular vacuolization but without the character­
IL‑10), the actions of each
mediator may be complex and
istic features of extensive tubular apoptosis or necrosis
context-dependent, making found in severe ischaemic injury 15. Similarly, necrosis is
simple classification difficult. uncommon in experimental septic AKI16.
profound alterations in the macrocirculation and
microcirculation with heterogeneity of regional blood
flow distribution17 and substantial decreases in func­
tional capillary density 18,19. In humans and large animal
­models, including sheep and pigs, cardiac output is usu­
ally increased during the initial stage of sepsis. Studies
in large animal models have also demonstrated that the
renal circulation participates in the systemic vasodila­
tation of sepsis, but despite an increase in renal blood
flow (RBF), oliguria occurs and AKI develops within
hours20,21. This dissociation between GFR and RBF can
be explained by alterations in the balance of efferent and
afferent glomerular arteriolar tone and decreased renal
perfusion pressure22 (FIG. 1). For example, a reduction in
GFR can be induced by a combination of systemic hypo­
tension and efferent arteriole vasodilation, which causes
the intraglomerular filtration pressure to drop below the
required threshold for effective ultrafiltration despite
normal glomerular perfusion. Such changes should be
reversible following systemic haemodynamic normal­
ization such as that induced by vasopressor therapy
(for example, noradrenaline). The reversibility of these
haemodynamic changes is evidenced by a comparison
of renal function, perfusion and histological changes
in sheep experiencing septic shock23, which showed no
change in RBF and renal oxygen consumption, with
histological alterations limited to minor focal mesan­
gial expansion, suggesting that mechanisms other than
RBF and oxygenation contribute to the development of
AKI in sepsis. For scenarios in which RBF does decrease
during sepsis, this change may be a consequence of AKI
rather than a cause. For example, a progressive increase
in renal vascular resistance occurred 18–22 hours after
induction of sepsis in a porcine sepsis model, but only
in animals with AKI. These haemodynamic changes
were preceded by substantial increases in plasma pro-­
inflammatory cytokines (such as IL‑6 and tumour
necrosis factor (TNF)) in the animals with AKI24, sug­
gesting that renal vasoconstriction occurs in response
to inflammatory injury rather than being the initiating
event of renal injury.
Similarly, in clinical sepsis, these early haemodynamic
changes with increased cardiac output and vasodilation
are likely to be accompanied by endothelial and renal
tubular cell inflammatory injuries that consequently
induce glomerular hypofiltration, which becomes refrac­
tory to global haemodynamic manipulation as afferent
vasoconstriction occurs in established AKI22.
Many mechanisms by which septic AKI causes
renal tubular injury have been proposed, including
ultrafiltration of circulating microbial toxins and the
release of inflammatory mediators that trigger tubular cell
stress and injury 25. Circulating toxins might act on the
endothelium, which triggers reduced microcirculatory
flow and interstitial infiltration of inflammatory cells26.
Concurrently with these inflammatory processes, local
ischaemia can still occur despite increased RBF due to
alterations in microcirculatory flow associated with
endothelial activation as well as the effects of locally
and systemically acting inflammatory mediators (such

218 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.

Damage-associated
molecular patterns
(DAMPs; also known as
danger-associated molecular
patterns). Endogenous
molecules often released from
damaged tissues or dying cells.
They can be detected by
various receptors on cells,
including TLRs and receptor for
advanced glycosylation end
products (RAGE; also known as
AGER). Typical DAMPs include
high mobility group protein B1
(HMGB1), RNA, DNA and
myoglobin.
Pathogen-associated
molecular patterns
(PAMPs). Similar to DAMPs,
PAMPs are molecules released
from pathogens. Many of the
same receptors that recognize
DAMPs also recognize PAMPs,
leading to similar cellular
responses when either are
encountered. Typical PAMPs
include lipopolysaccharide
(LPS; also known as endotoxin),
lipoteichoic acid from
Gram-positive bacteria,
peptidoglycan and nucleic acid
variants normally associated
with viruses.
NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 219
REVIEWS
as TNF and transforming growth factor‑β (TGFβ)). induces angio­poietin 2 expression and release, reduces
These alterations cause redistribution of perfusion angiopoietin 1 receptor activation and leads to vascular
away from the renal medulla to the cortex, leading to leakage33. Renal tubular responses to this inflammatory
medullary hypoxaemia27–30, which implies intrarenal milieu — including autophagy, mitochondrial dys­
shunting 31. The inflammatory processes involved in function, loss of cell polarity and, to a limited degree,
septic AKI might be mediated by Toll-like receptors necroptosis and apoptosis34 — lead to sustained AKI
(TLRs) expressed on tubular and endothelial cells that and potentially to fibrosis and development of chronic
recognize damage-associated molecular patterns (DAMPs) kidney disease (CKD).
and pathogen-associated molecular patterns (PAMPs), Finally, the relationship between inflammatory
such as lipopolysaccharide, as well as receptors for responses and AKI in sepsis might not be unidirec­
cytokines and other inflammatory mediators32. Receptor tional. One study reported that in 40% of ICU patients
activation then triggers local release of additional with sepsis and AKI, sepsis occurred after the onset of
pro-inflammatory mediators and the recruitment of AKI; however, this number is likely to be an underesti­
peritubular inflammatory infiltrating cells. Sepsis also mate because AKI diagnosis based on serum creatinine
Box 1 | Definitions and concepts of acute kidney injury
The standardization of definitions and diagnostic criteria into three stages based on serum creatinine level and urine
output122,123 (see the table) has been crucial in advancing the field of acute kidney injury (AKI). These stages serve as the
basis for AKI diagnosis, biomarker discovery and validation118 and aid epidemiological studies as well as patient
stratification in clinical trials. Furthermore, the staging specified by these definitions has been validated in terms of their
associations with the short-term and long-term risks of death and end-stage renal disease3,4. However, this
standardization obscures the reality that AKI is a collection of extremely heterogeneous clinical syndromes. For example,
utilizing the Kidney Disease: Improving Global Outcomes (KDIGO) definition, AKI may manifest as changes in only serum
creatinine (12% of patients), only urine output (38%) or both (50%)124. Importantly, these varying manifestations are
strongly associated with different outcomes: hospital mortality within these groups is 9.2% (only serum creatinine),
7.5% (only urine output) and 26.7% (both), and a similar pattern exists for 1‑year survival124.
Taking these differences into account, understanding of the underlying processes that cause the physiological changes
leading to AKI is of paramount importance. Creatinine excretion decreases during a sustained decrease in glomerular
filtration rate (GFR), which causes serum creatinine concentration to increase over time until creatinine excretion once
again equals production at the reduced GFR. Similarly, a sustained period of low urine output (oliguria) can be caused by
either an intact urinary-concentrating mechanism that works at increased capacity or a system that has failed because of
injury, in which GFR is so low that urine output cannot be maintained. In this light, more severe oliguria can be seen as an
additional marker of AKI severity through the indication of a substantial decline in GFR that may be difficult to appreciate
from early changes in serum creatinine in the dynamic setting of acute illness. For example, in the AWARE study,
a substantial number of AKI cases in children were missed when urine output was not considered125. Certain aetiologies
for AKI seem more likely to manifest as oliguric (such as sepsis and cardiorenal syndrome) or non-oliguric (such as
nephrotoxicity); however, this observation may be due to the systemic physiological milieu in these conditions (for
example, vasodilatory shock in sepsis). Thus, although both urine output and serum creatinine can indirectly give insight
into renal excretory function and are important as indicators of organ dysfunction and the potential need for supportive
therapy, in isolation, they give little information about the underlying renal pathology.
In addition, certain clinical conditions are usually termed ‘AKI’, whereas other conditions that may meet the
biochemical definition are generally not categorized as AKI because they constitute specific diagnoses. The term ‘AKI’
generally excludes specific disease states that cause acute renal dysfunction due to transient urinary obstruction unless
actual cellular damage has occurred as a result. Similarly, intrinsic renal conditions such as glomerulonephritis or
interstitial nephritis are usually thought of as specific diseases that are part of the differential diagnosis for AKI, but they
are uncommon causes. Instead, the term ‘AKI’ often refers to a syndrome of renal dysfunction arising in the context of a
systemic illness or toxin exposure, usually pathologically manifested as renal tubular cell dysfunction or injury, utilizing
other terms for intrinsic renal diagnosis such as glomerulonephritis. Thus, the ‘modern concept of AKI’ is a disorder
caused by a variety of haemodynamic, toxic and inflammatory insults, often occurring in combination, that result in a
disease spectrum ranging from a transient reduction in GFR — with, at most, mild tubular dysfunction — to severe
tubular injury, resulting in sustained anuria for days to weeks. Thus, in some aspects, this concept recapitulates the
historic categories of ‘prerenal failure’ and so‑called acute tubular necrosis, which are not truly distinct clinical entities
but extremes in a spectrum of the severity and nature of injury in the AKI syndrome. However, this observation does not
mean that diagnostic evaluation should focus on distinguishing transient or reversible AKI from established injury as this
approach historically has not proved useful owing to broad clinical overlap.
AKI KDIGO stage Serum creatinine Urine output
1 1.5–1.9 × baseline in 7 days or ≥0.3 mg/dl (26.5 μmol/l) in 48 h <0.5 ml/kg per h for 6–12 h
2 2.0–2.9 × baseline <0.5 ml/kg per h for ≥12 h
3 ≥3 × baseline or ≥4.0 mg/dl (35.4 μmol/l) or initiation of RRT <0.3 ml/kg per h for ≥24 h
or eGFR <35 ml/min/1.73 m2 in patients aged <18 years or anuria for ≥12 h
eGFR, estimated GFR; RRT, renal replacement therapy.
©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

is often delayed35. Another study found that bacterial
invasion and lung injury were more severe in mice with
experimental AKI and pneumonia than in mice
with pneumonia alone. Moreover, neutrophils isolated
from mice with AKI had impaired function in vitro
compared with that of wild-type mice36. Together with
the emer­ging literature on the various pathophysio­
logical effects of AKI11, these results provide evidence
supporting an effect of AKI on the immune system and
on the risk of sepsis. These findings are also reminiscent
of the more well-established literature on the effects of
CKD on innate immune pathways and, taken together,
support a key role for the kidney in the regulation of
immune function37–39.
Outcomes. Because of a lack of studies, the long-term
effects of septic AKI, or indeed the long-term effects of
sepsis in general, are not well understood40. However, the
inflammation associated with septic AKI is coupled to
long-term outcomes, as elevated plasma concentrations

Macrocirculation
Anaesthesia Major surgery Infection Kidney

Blood and Antimicrobial drugs

Redistribution of
fluid losses perfusion away
Venodilation from renal medula
Cardiopulmonary Tissue injury
bypass
Myocardial Obstruction
depression to venous
Hypovolaemia outflow
Arteriodilation Elevated
intrarenal
Capillary pressure
leakage Innate and acquired
Systemic
immune responses
hypotension
Extrinsic
renal
compression

Circulating mediators
↓ Renal
Large-volume perfusion
fluid resuscitation Sympathetic nervous pressure
system activation

Interstitial oedema Efferent

and fluid overflow arteriole
Afferent arteriole constriction dilation

Tubuloglomerular
feedback
Peritubular Proximal
Interstitium capillary tubule Lumen Intraglomerular
hypotension
↑ Tubular
Endothelial pressure
activation and Filtered
injury mediators
↓ GFR
Loss of Impaired urinary
endothelial concentrating Tubular
cell–cell Rouleaux capacity cell injury
junctions formation

↑ Capillary Tubular cell injury

permeability • De-differentiation
• Loss of cell–cell junctions Nephron
Leukocyte– Cytokines • Cell shedding
endothelial Chemokines Oxidative
stress • Vacuolation
adhesion and • Loss of polarity • Baseline O2 supply
interaction • Loss of transmembrane channels dependence
Larger O2 • Mitochondrial dysfunction • Oxidative stress
Coagulation Transendothelial diffusion • G1 cell-cycle arrest
Impaired migration distances • Apoptosis
blood flow

Renal microcirculation

Potential for rapid reversibility Red blood cell Inflammatory cell DAMPs Cytokines Cell adhesion
molecule
Likely to be associated with Dendritic cell TLRs PAMPs Toxins
sustained AKI Platelets

Nature Reviews | Nephrology

220 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph
©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

of inflammatory and apoptosis biomarkers (such as IL‑8)
have been associated with nonrecovery of renal function
and death in critically ill patients receiving renal replace­
ment therapy (RRT)41. These data suggest that sepsis and
AKI have long-term effects on inflammation, coagula­
tion and possibly other systems that might be critical in
determining long-term outcomes as well as in identifying
potential targets for therapy. However, septic AKI might
represent a different phenotype from that of other forms
of AKI that have inflammation and bioenergetic failure
as major pathophysiological features (for example, AKI
secondary to trauma or major surgery). In patients with
sepsis-associated AKI, renal recovery — even partial
recovery — seems to confer significant bene­fits such that
1‑year survival is indistinguishable from that of patients
with sepsis but without AKI42. However, this seemingly
good prognosis for patients with renal recovery may dis­
guise substantial underlying kidney damage that could
take years to manifest. Specifically, it will be important to
determine whether septic AKI increases the risk of CKD
years later, even in patients who recover renal function
prior to hospital discharge. For example, whether these
patients are left with interstitial fibrosis and tubular atro­
phy or demonstrate apparent recovery from AKI due to
hyperfiltration in remaining nephrons, which ultimately
leads to glomerulosclerosis, should be investigated. If so,
angiotensin-converting enzyme (ACE) or angiotensin
receptor inhibition might be effective at preventing pro­
gression of CKD, as has been shown in animal models
of AKI and CKD43–45. Determining whether septic AKI
increases the risk of cardiovascular events and/or infec­
tion even in the absence of CKD is also important, as
◀ Figure 1 | Principal pathogenic mechanisms of acute kidney injury in the context of
sepsis or major surgery. In both settings, alterations in systemic and glomerular
haemodynamics — arising from the vasodilatory effects of systemic inflammatory
mediators or anaesthesia, decreases in actual or effective in plasma volume and
myocardial depression — may cause an early decrease in glomerular filtration rate (GFR)
with potential for rapid reversibility (dotted arrows). Depending on the clinical context,
these changes can be associated with high, normal or low cardiac output. However, both
sepsis and tissue injury in surgery result in the generation of damage-associated
molecular patterns (DAMPs) — including high mobility group protein B1 (HMGB1),
histones, cell-free RNA and DNA, ribonuclear proteins, S100 proteins and heat-shock
proteins — or pathogen-associated molecular patterns (PAMPs) — prototypically
lipopolysaccharide but also including a vast array of other bacterial-specific or
viral-specific molecules. These ‘molecular patterns’ are recognized by Toll-like receptors
(TLRs), triggering systemic and local release of pro-inflammatory mediators, such as IL‑1,
IL‑2, IL‑6, IL‑8 IL‑12, IL‑18, transforming growth factor‑β (TGFβ), tumour necrosis factor
(TNF), IFNγ, C‑C chemokine ligand 2 (CCL2) and CCL5. Furthermore, antimicrobial drugs
that are used for the treatment of infections can contribute to nephrotoxicity (FIG. 3).
The effects of DAMPs and PAMPs are combined with those of specific immune
responses and the generation of other inflammatory mediators, including products of
coagulation and complement activation, leading to the recruitment of a peritubular
inflammatory cell infiltrate (neutrophils, monocytes, macrophages and lymphocytes).
Importantly, these circulating toxins and mediators may act directly on the endothelium
or may be filtered in the glomerulus, where they interact with tubular cells and cause
injury. The renal inflammatory response can cause injury to the endothelium,
microcirculation and tubular cells. If this damage is severe, it will lead to sustained
clinical acute kidney injury (AKI) that is unresponsive to the correction of systemic
haemodynamic abnormalities. Finally, during critical illness, the effects of fluid overload,
elevated venous pressure and development of abdominal compartment syndrome can
also contribute to the persistence of AKI by an increase in tubular pressure and reduction
of glomerular perfusion.
is whether known or novel biomarkers can predict the
subsequent risk of cardiovascular events or recurrent
infection after an episode of AKI with or without evi­
dence of renal recovery. This level of understanding is
necessary so that interventions (such as statin treatment
or immune stimulation) can be developed for patients
at risk of long-term sequelae, as opposed to waiting for
these events to occur.
Surgery-associated AKI
Epidemiology. Surgery and trauma have long been
recognized as a leading cause of AKI in hospitalized
patients46,47. According to crude estimates, 310 million
surgical procedures were undertaken worldwide in 2012,
an increase of one-third in 8 years48. In the USA, 1.5
million patients develop a medical complication within
30 days after surgery, and at least 150,000 patients die
each year 49. Importantly, AKI occurring after surgery is
strongly associated with risk of death50–55.
Employing modern consensus definitions of AKI
(BOX 1), the incidence of AKI among patients under­
going surgery varies by surgical setting: ~18% of patients
undergoing cardiac surgery develop AKI56 compared
with ~13% of patients undergoing major abdominal
surgery; however, the incidence is ≥50% in procedures
such as liver transplantation or emergency aortic sur­
gery 55. Although most surgery-associated AKI is mild,
these complications remain consistently associated
with greater short-term and long-term risks of death,
­especially in cases of severe AKI46,57.
Pathophysiology. Multiple aetiological factors predis­
pose to perioperative AKI; depending on the surgi­
cal context, these include haemodynamic alterations,
exposure to exogenous and endogenous nephrotoxins,
ischaemia–reperfusion injury, renal artery embolization,
activation of neurohormonal responses to hypotension
and tissue injury (the sympathetic nervous system and
the renin–angiotensin–aldosterone system), inflam­
mation and oxidative stress. However, as these factors
coexist and influence each other throughout the peri­
operative period, determining the extent to which
individual components contribute to the development
of AKI is difficult. Furthermore, patients requiring sur­
gery often have multiple comorbidities (for example,
CKD, diabetes, hypertension and heart disease) that
may impair renal autoregulation, chronically activate
neurohormonal responses and predispose to tubular
injury. The prevalence of comorbid conditions might be
particularly high in specific high-risk surgical contexts
such as among patients undergoing cardiac and vascular
surgery 55. Importantly, pre-existing CKD is an impor­
tant risk factor for postoperative mortality 58 and has
been consistently identified as the strongest risk ­factor
for perioperative AKI51,52,59.
Multiple factors and procedures can result in haemo­
dynamic instability during surgery, including the dila­
tory properties of anaesthetic agents, blood and fluid
losses, poor cardiovascular reserve and artificial circu­
lation during cardiopulmonary bypass (CPB) (FIG. 1).
Importantly, appropriate management of intraoperative

NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 221

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

haemodynamic changes, and preoperative and postoper­
ative optimization of intravascular volume, cardiac out­
put and oxygen delivery is associated with a substantial
reduction in the incidence and severity of periopera­
tive AKI among high-risk patients60,61. In retrospective
analyses of noncardiac surgical populations comprising
~5,000 patients, the presence of intraoperative hypo­
tension was predictive of adverse cardiac-related and
renal-related outcomes62,63, with the risk increasing in
association with the duration and severity of hypo­
tension for individuals with a mean arterial pressure
<65 mmHg (REF. 64). Overall, the relationship between
haemodynamics and AKI during surgery suggests a role
for altered renal perfusion in the development of post­
operative AKI. In some circumstances, such as in aortic
surgery, this alteration may be unmistakably ischaemic,
whereas in other cases, changes in GFR may be rapidly
reversible, with a correction of haemodynamics in the
absence of other sources of renal injury. For example,
experimental reduction of RBF to 10% of baseline levels
causes an abrupt loss of glomerular filtration but is com­
pletely and rapidly reversible with restoration of normal
RBF in the absence of other renal insults65. Conversely,
in a porcine animal model of aortic cross-clamping, RBF
and microvascular oxygenation were initially compara­
ble to those of control animals but rapidly fell over the
next 4 hours after release of a suprarenal clamp66. This
observation suggests that evolving reperfusion injury
in the context of surgery compromises renal micro­
circulation and macrocirculation rather than having an
­immediate effect of ischaemia.
Both cardiac and noncardiac surgical proce­
dures result in a systemic immune and inflammatory
response to injury as well as the release and recognition
of DAMPs67. Such responses are well recognized in the
context of CPB procedures, in which these responses are
thought to be triggered by contact between blood and
the artificial surfaces of the CPB68. Although the exact
contribution of the upregulated inflammatory response
to kidney injury in humans undergoing CPB is not
known, it is likely to be substantial69. In a multicentre
cohort study including 960 individuals, increased post­
operative concentrations of both pro-inflammatory and
anti-inflammatory cytokines (IL‑6 and IL‑10, respec­
tively) were associated with an increased risk of AKI after
cardiac surgery, demonstrating a correlation between
systemic inflammatory responses and AKI70. It should
be noted that this relationship may also be bidirectional,
given the importance of the kidney in clearing inflam­
matory mediators from the circulation71. Sustained sys­
temic inflammatory responses are also common after
major abdominal surgery, and increased IL‑6 production
and upregulation of TLR4 and TLR5 expression in the
postoperative period have been proposed as potential
biomarkers for increased AKI risk72. These responses are
also associated with the d ­ evelopment of AKI in other
settings such as sepsis.
Finally, patients undergoing surgery are exposed
to multiple potential nephrotoxins, such as antibiot­
ics, nonsteroidal anti-inflammatory drugs (NSAIDs),
ACE inhibitors, angiotensin receptor blockers and
intra­venous radiocontrast agents. Some of these com­
pounds predominantly affect renal haemodynamics,
whereas others are directly nephrotoxic (see below). In
addition, release of endogenous nephrotoxins — such as
myoglobin in trauma and haemoglobin or free haem in
CPB — may also cause AKI.
Outcomes. Despite great heterogeneity in its patho­
physiology, perioperative AKI of any severity remains
associated with an ongoing increased risk of death,
even many years after surgery. For example, patients
with RIFLE (risk, injury, failure, loss, and end stage)
stage F after cardiothoracic surgery have a significantly
lower survival over 10 years than patients without AKI
(adjusted HR 2.14, 95% CI 1.73–2.66)73. Importantly —
and unlike what has been observed in sepsis-associated
AKI — even in cases of complete resolution, periopera­
tive AKI is associated with increased risks of CKD and
death in the years after surgery 52,73–78. After cardiac sur­
gery in particular, perioperative AKI is also associated
with higher risks of CKD78, end-stage renal disease78
and death79,80. Importantly, as serum creatinine levels
tend to fall during hospitalization for surgery, an overt
decline in renal function may not be evident at the point
of hospital discharge in patients who have apparently
recovered from perioperative AKI. This decline might
become apparent only at follow-up54, which partially
explains the poor prognosis and emphasizes the need
for ­long-term follow‑up.
AKI associated with renal hypoperfusion
Epidemiology. Aetiologies of AKI that are compatible
with sustained renal hypoperfusion — such as major
obstetric haemorrhage and crush injury with prolonged
extraction — as well as exposure to potent exogenous
nephrotoxins or similar insults, including incompatible
blood transfusion, were commonly reported in the 1940s
and 1950s81. These observations led to the concept that
tubular necrosis is the dominant pathology of AKI, which
translated into widespread popularity that continues to
this day, even in the use of animal models of AKI caused
by renal ischaemia–reperfusion injury or exposure to
potent nephrotoxins. However, AKI that is directly
attributable to a gross reduction in RBF is less common in
contemporary, integrated, medical settings in the devel­
oped world. Nevertheless, severe reductions in RBF can
still occur, most commonly in specific contexts, such as
following abdominal aortic dissection, sustained cardiac
arrest or exsanguinating trauma. It should be noted that
even in cardiac arrest, in which RBF can be very low for
a long period, development of AKI is not universal and
is most strongly associated with persistent cardiogenic
shock in the period after arrest 82. Therefore, it seems that
hypoperfusion is more commonly a contributing factor
rather than the sole driver of AKI in many clinical set­
tings. However, certain individuals have an inherently
high risk of developing AKI following mild to moderate
reductions in kidney perfusion owing to the presence of
comorbidities (such as chronic hyper­tension and diabe­
tes) and certain medications (for example, NSAIDs and
calcineurin inhibitors)82. Conversely, even in situations

222 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

in which systemic ischaemia is not present, microcircula­
tory abnormalities may cause local ischaemia at the tissue
level. Although these abnormalities may be an impor­
tant mediator of injury, they would not be expected to
be responsive to vasopressor or fluid therapy targeted
at the macrocirculation as the ischaemic insult occurs at
the local tissue level rather than the whole-­organ level.
Thus, despite ischaemia being a historically dominant
paradigm of AKI, its epidemiology is difficult to define.
For example, one epidemiological study attributed ~20%
of cases of RRT-requiring AKI in the ICU to cardiogenic
shock or hypovolaemia82, both of which are likely to be
associated with renal hypoperfusion but with differing
underlying causes and approaches to treatment.
Pathophysiology. Various pathophysiological states
and  medications can contribute to a reduction of
RBF and have been described as causing generalized
or localized ischaemia in the kidney, leading to AKI83.
These contributing factors include intravascular volume
depletion and hypotension through internal or external
fluid losses from the circulation (for example gastrointes­
tinal bleeding, dermal losses and haemorrhage), condi­
tions associated with chronic neurohormonal activation,
systemic hypotension and fluid retention (for example,
congestive heart failure and cirrhosis) and medications
that cause renal vasoconstriction (for example, NSAIDs,
radiocontrast agents and calcineurin inhibitors) (FIG. 2).
Furthermore, endothelial and vascular injury after
ischaemia as well as the role of innate and adaptive
immune responses in microcirculatory dysfunction after
reperfusion are important in the pathogenesis of ischae­
mic AKI83,84. Thus, the microvascular and inflammatory
responses to reperfusion are likely to be key determinants
of the clinical course of ischaemic AKI. Of note, owing to
the anatomy of the nephron microcirculation, the renal
medulla and corticomedullary junction receive poorly
oxygenated blood in the context of renal hypoperfusion
and are therefore much less well perfused than the renal
cortex. As a consequence, these areas are susceptible to
multiple sources of renal injury (inflammatory, oxida­
tive, ischaemic or toxic insults), particularly in zones of
the nephron (for example, in the proximal tubule) that
require high oxygen consumption for active ion trans­
port. Thus, even without overt renal ischaemia, base­
line oxygen supply levels might render the renal tubules
susceptible to non-ischaemic renal injury and to subtle
­disruptions of the renal microcirculation83.
Distinguishing whether a reduction in RBF is a cause
or a consequence of already established AKI is also chal­
lenging. Various studies have demonstrated that, once
established, all types of AKI may be associated with a
reduction in RBF, in many cases with normal systemic
blood pressure and cardiac output 85. This observation
suggests that intrarenal factors specifically increase
renal vascular resistance in response to AKI, resulting in
decreased RBF. Such renal vasoconstriction is mediated
by the tubuloglomerular feedback mechanism and is
therefore preferentially pre-glomerular, causing a dispro­
portionate decrease in glomerular filtration by ­reducing
both intraglomerular hydrostatic pressure and flow.
By contrast, in mild forms of AKI, changes in glo­
merular haemodynamics are reversible and may be
associated with increased serum creatinine levels and
oliguria in the absence of severe tubular injury. Such
reversible AKI encompasses severe volume depletion
or low cardiac output states that cause renal hypoperfu­
sion. Together, these different states have been consid­
ered under the umbrella term ‘prerenal’ AKI; however,
considerable overlap exists in terms of risk factors and
aetiologies with tubular injury, which may often evolve
in concert with acute haemodynamic changes that con­
tribute to oliguria and/or a fall in GFR. Importantly,
cardiac dysfunction can reduce renal perfusion either
through decreased forward flow or by increasing back­
pressure and venous congestion. Right-sided heart fail­
ure and raised venous pressure may actually be more
important as a cause of cardiorenal syndrome than
left-sided heart failure, especially in settings such as
cardiac surgery. As such, high venous pressures have
been shown to predict deterioration of renal function
in hospitalized patients with advanced chronic cardiac
failure; however, cardiac output and left ventricular
function did not 86,87. This deterioration presumably
occurs because the venous outflow pressure from the
kidney is normally low but increases with right-sided
heart failure. Furthermore, because the kidney is in a
poorly compliant capsule, it is more sensitive to venous
congestion. Pericardial disease, cirrhosis with ascites
and even abdominal compartment syndrome88 can
produce similar results because they all increase back­
pressure on the kidney. Even in the absence of these
disease manifestations, fluid overload itself can com­
promise renal (and other tissue) perfusion89. Finally,
loss of glomerular perfusion pressure can also occur in
vasodilatory states with systemic hypotension — such
as early septic shock and chronic liver disease — and
through exposure to vasoactive nephrotoxins, poten­
tially resulting in reversible, haemodynamically medi­
ated AKI despite preserved or elevated cardiac output.
Taken together, reversible AKI can arise in very diverse
settings, and in practice, the dominant underlying
mechanism is often apparent only as the clinical course
of AKI progresses.
Outcomes. In contrast to sepsis and surgery, less informa­
tion on clinical outcomes in settings of hypoperfusion-­
associated AKI is available. Patient outcomes are likely
to be strongly determined by the underlying conditions,
with advanced chronic heart or liver failure conferring
a poor prognosis. However, in the specific context of
myocardial infarction-related cardiogenic shock, RRT-
requiring AKI has been shown to predict substantially
elevated short-term mortality as well as long-term risks
of chronic dialysis and death90.
Nephrotoxic AKI
Epidemiology. The true contribution of nephro­toxicity
to the incidence of AKI is difficult to determine given
the inevitable indication bias in which the use of a
drug is associated with the presence of an underlying
disease state. In the ICU, some forms of nephrotoxic

NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 223

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Macrocirculation Heart Kidney

Redistribution of
perfusion away
Haemorrhage or from renal medula
hypovolaemia

Acute heart failure Obstruction

Low cardiac Central PE to venous
output Tamponade outflow
Cardiac arrest Elevated
Capillary Systemic intrarenal
leakage hypotension pressure

Venodilation Hypoperfusion
Arteriodilation ↓ Renal perfusion
pressure
Direct ischaemic Reperfusion
tissue injury injury

Activation of
sympathetic
system

Circulating inflammatory
mediators and toxins Afferent
arteriole ↓ GFR
constriction

Peritubular Proximal Efferent

Interstitium capillary tubule Lumen arteriole Intraglomerular
dilation hypotension
↑ Tubular
Endothelial pressure
activation and Tubuloglomerular
injury feedback
Filtered
Loss of mediators Impaired urinary
endothelial Tubular cell
concentrating
cell–cell Rouleaux injury
capacity
junctions formation

↑ Capillary Tubular cell injury

permeability • De-differentiation
• Loss of cell–cell junctions Nephron
Leukocyte– Cytokines • Cell shedding
endothelial Chemokines Oxidative
stress • Vacuolation
adhesion and • Loss of polarity • Baseline O2 supply
interaction • Loss of transmembrane channels dependence
Larger O2 • Mitochondrial dysfunction • Oxidative stress
Coagulation Transendothelial diffusion
migration • G1 cell-cycle arrest
Impaired distances • Apoptosis
blood flow

Renal microcirculation

Potential for rapid reversibility Red blood cell Inflammatory cell DAMPs Cytokines Cell adhesion
molecule
Likely to be associated with Dendritic cell TLRs Toxins Platelets
sustained AKI

Figure 2 | Principal pathogenic mechanisms of acute kidney injury in the context of global hypoperfusion.
Nature Low
Reviews | Nephrology
cardiac output and systemic perfusion may arise in the context of major haemorrhage, profound hypovolaemia or an
intrinsic or obstructive cause of acute heart failure or cardiac arrest. On the one hand, acute haemodynamic changes can
mediate reversible falls in glomerular filtration rate (GFR) after profound global hypoperfusion (dotted arrows). On the
other hand, both the direct effects of ischaemia and reperfusion on the kidney and the local and systemic effects of
damage-associated molecular patterns (DAMPs) and other inflammatory mediators produced in response to injury
and reperfusion may drive sustained acute kidney injury (AKI; solid arrows). The ongoing inflammatory nature of the
renal injury after the initial insult shares features with the pathogenesis of other forms of AKI (FIGS 1,3) and explains
the persistence of injury, even if systemic perfusion is restored, and may also be associated with a second
systemic vasodilatory response as commonly seen after resuscitated cardiac arrest. PE, pulmonary embolus;
TLRs, Toll-like receptors.

224 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

exposure are unavoidable, as >20% of the 100 drugs
most frequently administered to adult patients are clas­
sified as nephrotoxic91. Reports suggest that nephro­
toxicity contributes to ~8–60% of AKI among patients
in hospital92–96, whereas in the ICU, drug nephro­toxi­
city has been reported to contribute to 19% of RRT-
requiring AKI cases9. Nephrotoxic exposure is likely
to be increasing 95 owing to the more aggressive med­
ical management of complex conditions with nephro­
toxic drugs, in p ­ articular in oncology and with bone
marrow transplantation97.
Pathophysiology. To achieve adequate glomerular ultra­
filtration, the renal vascular bed receives a dispropor­
tionately large blood flow fraction of the resting cardiac
output (20–25%) compared with most other organs.
Consequently, the renal vasculature, nephrons and the
interstitium are exposed to large quantities of blood-
borne toxins (FIG. 3). Furthermore, tubular epithelial
cells are especially prone to toxic injury, as they concen­
trate filtered toxins along the nephron and express spe­
cific transporters that result in high intracellular uptake
of toxins and their metabolites. The biological effects of
nephrotoxins are highly varied. Injury to the kidney
may occur secondary to renal haemodynamic altera­
tions, tubular cell damage by direct toxicity or osmotic
injury, obstruction of the tubular lumen due to precip­
itation, direct or indirect (immunological) interstitial
inflammation and/or vascular injury triggering throm­
botic microangiopathy 98. Although haemodynamic-­
mediated alterations in GFR may be rapidly reversible
by withdrawal of the relevant toxin, long-term exposure
to renal vasoconstrictors can result in chronic ischaemic
nephropathy. Importantly, some toxins are associ­ated
with multiple forms of injury, and patients are fre­
quently exposed to multiple nephrotoxins at the same
time. In addition to drug-related causes, haemoglobin
or myoglobin, released to the bloodstream through
haemolysis or myocyte injury, may act as endogenous
nephro­toxins in the kidney by mechanisms similar to
those of exogenous nephrotoxins through causing oxi­
dative and inflammatory stress as well as having vaso­
active effects (for example, by depletion of endothelial
nitric oxide).
As with hypoperfusion, nephrotoxic renal injury
may recapitulate the AKI pathophysiology seen in other
settings. For example, the pathophysiology of cisplatin-­
induced AKI involves multiple mechanisms, including
proximal tubular injury, oxidative stress and vascu­
lar injury in the kidney 99. Proximal tubular cell injury
encompasses several different mechanisms in common
with other forms of AKI, such as apoptosis, autophagy,
cell-cycle arrest and mitochondrial dysfunction. In addi­
tion, the pro-inflammatory nature of cisplatin-induced
AKI has been well documented100,101: the chemotherapy
drug induces the phosphorylation and subsequent trans­
location of nuclear factor‑κB, leading to transcriptional
upregulation of inflammatory mediators, including TNF,
which, in turn, induces the expression of other inflam­
matory cytokines and recruitment of ­inflammatory cells
into the kidney tissue102.
Outcomes. The long-term outcomes of nephrotoxic AKI
are not well established. Most cases of nephro­toxicity are
acute, non-oliguric and resolve with discontinuation of
the causal drug. However, for some drugs, or when other
AKI risk factors coexist, mixed patterns of injury result
in heterogeneity of recovery. For example, in cisplatin-­
mediated nephrotoxicity, the GFR decline tends to be
reversible, whereas tubular dysfunction may persist 103.
A retrospective cohort study of 100 children with AKI
in the context of high nephrotoxic exposure reported
evidence of CKD 6 months after AKI in 70% of cases104.
Thus, in common with AKI in other settings, devel­
opment of CKD might be a common and i­mportant
­consequence of nephrotoxic AKI.
Understanding paradigms of AKI
The scientific foundation for acute kidney injury. We
have described AKI as a clinical syndrome that arises in
several prototypical acute settings, including sepsis, fol­
lowing major surgery, in states of low cardiac output and
nephrotoxin exposure and often in combination with
pre-morbid chronic risk factors. Within these settings,
a number of common themes emerge with patterns of
inflammatory, ischaemic and nephrotoxic renal injury
that can occur with all of these exposures. Considerable
overlap exists even within these categories of injury; for
instance, inflammatory mechanisms can mediate rep­
erfusion injury after ischaemia or certain nephrotoxins,
whereas abnormalities of microvascular flow can result
in local ischaemia in the settings of sepsis or toxic injury
(FIGS 1–3). The extent of overlap is often difficult to pre­
dict at the point of clinical presentation. Furthermore,
in almost all settings, changes in GFR can be caused by
renal or systemic haemodynamic abnormalities and can
either be reversible or result in tubular cell injury.
Importantly, the apparent return of GFR to normal
levels may not guarantee resolution of AKI105, as the
remaining nephrons may increase their function to
maintain GFR while masking nephron loss and tubular
atrophy. Thus, clinical assessment of pathological versus
physiological reductions in GFR may be impossible.
As described in this Review, strong clinical reasons
cast doubt on the notion that all AKI syndromes are cre­
ated equal in terms of their aetiology, pathogenesis and
prognosis. Even among patients with septic AKI, some
seem to have a more ‘pure’ form of disease, whereas dis­
ease progression in others is complicated by nephrotoxic
exposures and complex underlying illnesses. Therefore,
a rational clinical approach to an AKI diagnosis requires
assessment of the underlying mechanisms to help estab­
lish a prognosis and identify potentially beneficial inter­
ventions. At present, this approach relies on clinical
assessment and serial measurement of serum creatinine
and urine output until either recovery has occurred or
nonrecovery has been established — usually requiring
several days or more105. However, novel diagnostics
might one day provide greater insights into the likely
­short-term, medium-term and long-term courses of AKI.
Hence, well-defined animal models, human
epidemio­logical studies and biological samples are
required to establish a better scientific foundation for

NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 225

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Macrocirculation Kidney
Nephrotoxins • Ciclosporin
• NSAIDs

• ACE inhibitors Chronic renal

• ARBs ischaemia
• IL-2
• Proton-pump inhibitors
• Gentamicin • β-Lactams
• Cisplatin • Allopurinol
Arteriodilation • Hydroxyethyl starch • Aristolochic acids
• Radiocontrast agents Allergic
• Myoglobin interstitial
• Haem and/or iron nephritis
Systemic ↓ Renal perfusion
hypotension pressure

• Mitomycin • ACE inhibitors

• Ciclosporin • Ciclosporin • ARBs
• NSAIDs • IL-2
• Radiocontrast • Aciclovir
agents • Foscarnet
• Myoglobin • Methotrexate
• Haem and/or iron • Myoglobin

Sympathetic nervous Efferent ↓ GFR

system activation arteriole
dilation
↑ Tubular
Peritubular Proximal Afferent pressure
Interstitium capillary tubule Lumen arteriole Intraglomerular
constriction hypotension
Endothelial Tubular
activation and obstruction
injury Tubuloglomerular
Filtered feedback
Loss of toxins
endothelial Impaired urinary Tubular cell
cell–cell Rouleaux concentrating injury
junctions formation capacity

↑ Capillary Tubular cell injury

permeability • De-differentiation
• Loss of cell–cell junctions Nephron
Leukocyte– Cytokines • Cell shedding
Chemokines Oxidative
endothelial • Vacuolation
stress • Baseline O2 supply
adhesion and • Loss of polarity
interaction • Loss of transmembrane channels dependence
Larger O2 • Mitochondrial dysfunction • Oxidative stress
Transendothelial
Coagulation migration diffusion • G1 cell-cycle arrest
Impaired distances • Apoptosis
flow
Renal microcirculation

Potential for rapid reversibility Red blood cell Inflammatory cell DAMPs Cytokines Cell adhesion
molecule
Likely to be associated with Dendritic cell TLRs Toxins Platelets
sustained AKI

Figure 3 | Principal pathogenic mechanisms of acute kidney injury in the context of exogenous and
Nature Reviews | Nephrology
endogenous nephrotoxins, including typical examples. Many potential pathways for nephrotoxicity exist, and
individual nephrotoxins may exploit more than one mechanism. Changes in systemic and/or glomerular haemodynamics
may mediate changes in the glomerular filtration rate (GFR) related to nephrotoxin exposure; however, direct tubular
cellular injury from filtered toxins or (in some cases) direct toxic endothelial injury (including thrombotic microangiopathy)
is more likely to lead to acute kidney injury (AKI) that is sustained even with withdrawal of the toxic exposure. The
inflammatory response to tissue injury may be important in potentiating and sustaining the initial toxic insult, with
mechanisms in common with other AKI settings (FIGS 1,2). In addition to direct cellular injury, tubular obstruction by
precipitates of substances filtered and concentrated along the nephron is an important mechanism of nephrotoxicity.
Finally, chronic ischaemia from renal vasoconstrictor exposure and drug-induced allergic tubulointerstitial nephritis may
present subacutely with worsening renal function due to nephrotoxin exposure with distinct pathogenesis from the
typical AKI mechanisms discussed elsewhere in this Review. Importantly, acute and chronic nephrotoxin exposure often
involves multiple agents and occurs in the context of other AKI aetiologies further overlapping with mechanisms
illustrated in FIGS 1,2. ACE, angiotensin-converting enzyme; ARBs, angiotensin receptor blockers; DAMPs,
damage-associated molecular patterns; NSAIDs, nonsteroidal anti-inflammatory drugs; TLRs, Toll-like receptors.

226 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

Table 1 | Selected clinical and preclinical studies exploring acute kidney injury endotypes
Model systems Technique Findings Potential clinical benefit Refs
Comparison of ischaemia– Microarray analysis Spatiotemporal differences in RNA • Improved diagnosis 108
reperfusion-induced and mercuric expression levels, such as haem oxygenase 1. • Identification of drug
chloride-induced rat models of AKI targets
Comparison of ischaemia– Chromatin Upregulated expression of some genes (such Identification of potential 109
reperfusion-induced and immunoprecipitation as those encoding TNF and NGAL) in both drug targets
LPS-induced mouse models of AKI models with some insult-specific expression
patterns.
Porcine model of septic AKI Whole-kidney gene Differences in gene expression between • Identification of 110
expression analysed by sepsis AKI and sepsis without AKI. potential drug targets
quantitative real-time PCR Inflammation, metabolism and apoptotic • Guiding mechanistic
molecular responses in early septic AKI. research
• Dehydration versus ischaemia– • RNA sequencing of Activated genes included different, • Identification of 112
reperfusion injury in mice specific kidney regions functionally unrelated signal transduction potential drug targets
• Humans with brief (<72 h) isolated by laser pathways and were expressed in different • Guiding mechanistic
or long-term (>7 day) AKI microdissection regions of the kidney. Brief and long-term research
• RNA sequencing of AKI in humans also showed different RNA
human urine expression.
NGAL luciferase reporter assays NGAL expression via LPS resulted in kidney NGAL expression, Clarify specificity of NGAL 113
in mice exposed to dehydration luciferase reporter whereas a 50% increase in creatinine with for intrinsic AKI
versus LPS dehydration did not.
Multicentre study of 744 patients 340 candidate molecules Urinary TIMP2 and IGFBP7 were topic Discovery and validation 117,
with a mixed aetiology of AKI examined in urine and candidates in discovery and validation of biomarkers 118
plasma for stage 2–3 AKI within 12 h. TIMP2 was
somewhat better for septic AKI according
to the AUC.
AKI, acute kidney injury; AUC, area under the curve; IGFBP7, insulin-like growth factor-binding protein 7; LPS, lipopolysaccharide; NGAL, neutrophil gelatinase-
associated lipocalin. TIMP2, tissue inhibitor metalloproteinases 2; TNF, tumour necrosis factor.

phenotyping AKI. Although current animal models fail genes differentially expressed according to the specific
to recapitulate the multiple renal insults and comorbid type of renal injury 109. Not only might transcriptional
diseases that are present in most clinical cases of AKI, changes reflect the nature of the injury but also the elici­
understanding the common features and differences ted host response. For example, in a porcine model of
among the various causes of AKI that may coexist within septic AKI, only 50% of animals developed AKI despite
an individual may be established by studying phenotypes similar initial haemodynamic responses to experimental
of pure forms of AKI. Progress in AKI research has also sepsis; however, development of AKI was preceded by
been substantially limited by a lack of available patientthe differential expression of genes with inflammatory,
biopsy tissue, as emphasized by multiple investigators106.
metabolic and apoptotic functions110. Beyond the level
Furthermore, as AKI contributes to the development of of gene expression, microRNAs (mi­RNAs) play a cru­
CKD107, patient biopsy samples could represent a con­ cial part in the post-transcriptional regulation of almost
siderable asset to the scientific communities trying to all biological cell functions. In this regard, alterations in
understand this relationship. ­levels of circulating tissue mi­RNAs have been demon­
strated in several experimental and clinical AKI settings,
Endophenotyping of acute kidney injury. Apart from with some mi­RNAs commonly detected across many
describing the paradigms of AKI by aetiology, several forms of AKI and others u ­ pregulated or downregulated
investigators have also sought to define AKI subtypes in specific circumstances111.
by examining patterns of gene expression in response to RNA sequencing (RNAseq) of anatomically dis­
kidney injury. These subtypes are sometimes referred tinct domains of the kidney from ischaemic AKI and
to as ‘endophenotypes’ or simply ‘endotypes’ to distin­ volume-depletion mouse models, which cause a simi­
guish them from phenotypes that are directly observ­ lar rise in serum creatinine levels, revealed distinct and
able without the aid of molecular diagnostics. In a largely nonoverlapping sets of genes that are expressed
comparison of ischaemia–reperfusion-induced and in response to these differing renal insults112. Similar
nephrotoxin (mercuric chloride)-induced rat models differences were seen among urinary proteomic pro­
of kidney injury, spatiotemporal differences of renal files of selected cases of human AKI with rapidly revers­
transcriptomes were reported108 (TABLE 1). Similarly, in ible (<72 hours) or sustained (≥7 day) time courses112.
another comparison study of ischaemia–reperfusion-­ In murine models, ischaemic AKI was accompanied by
induced and lipopolysaccharide-­induced AKI mouse expression of genes related to cytokines and inflamma­
models, transcriptional and epigenetic heterogeneity tory mediators, the cell cycle, coagulation and TLR sig­
was observed, with some genes (for example, those nalling, whereas volume depletion was associated with
encoding TNF and neutrophil gelatinase-associated a smaller set of activated genes associated with bioener­
lipocalin (NGAL)) commonly upregulated but other getics and cellular transport 112. This study suggests that

NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 227

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Recovery Good health Furthermore, patients with increased AKI biomarkers

Delayed recovery
CVD and other
AKI Nonrecovery complications
CKD
Death (early) AKI AKD AKI Death
Time
Days to weeks Weeks to months Months to years
Figure 4 | Potential patient outcomes following acute
kidney injury. As a result of anNature
episodeReviews | Nephrology
of acute kidney
injury (AKI), patients may recover, be discharged without
recovery of renal function or die. Patients who seem to
recover may also later develop chronic kidney disease
(CKD) or cardiovascular disease (CVD) (dashed lines).
Although information exists on in‑hospital events and final
outcomes, the pathways leading to these outcomes are
virtually unknown. Owing to the high complexity, not all
relationships are displayed. AKD, acute kidney disease.
although volume depletion and ischaemia–­reperfusion
both involve renal hypoperfusion, they represent very
different clinical entities, with confirmation of the
importance of inflammatory pathways in ischaemia–
reperfusion leading to sustained renal impairment.
This differing pathology might not be evident from
a snapshot assessment of serum creatinine level but
will become evident as the clinical course of illness
progresses over time. However, distinguishing severe
dehydration (for example, through diarrhoea and/or
vomiting) from situations in which the renal circulation
has been stopped and restarted (for example, in vascular
surgery or cardiac arrest) is not difficult in clinical prac­
tice. Instead, a better understanding of likely outcomes is
needed for the majority of clinical cases, in which multi­
ple, but less severe, renal insults coexist. These cases are
poorly reflected by current animal models.
Importantly and as mentioned earlier, profound
hypovolemic states are unlikely to arise in patients in
modern ICUs, where volume overload is the prevailing
clinical picture. Although it is true that extravascular
volume may not reflect circulating blood volume, the
development of AKI in the ICU is far more likely to be a
result of sepsis, nephrotoxins and/or major surgery than
hypovolaemia. Disparities between physiological reduc­
tions in GFR and established tubular injury could be
revealed by novel AKI biomarkers. In an Ngal reporter
mouse, ischaemia–reperfusion, cisplatin and endotoxin
insults induced renal NGAL expression, whereas volume
depletion that was sufficient to induce a 50% increase
in serum creatinine level did not 113. In clinical practice,
urinary detection of NGAL in 1,635 unselected patients
predicted reversible versus nonrapidly reversible AKI
(with 81% specificity and 68% sensitivity at a thresh­
old of 104 ng/ml) at emergency room presentation114.
(NGAL or kidney injury molecule 1 (KIM1; also known
as HAVCR1)) but without significantly elevated serum
creatinine levels remained at increased risk of a com­
posite of dialysis or death compared with those without
increased NGAL, KIM1 or creatinine levels114.
Indeed, the concept of ‘subclinical AKI’ has been sug­
gested as a way to understand kidney injury detected by
biomarkers in cases that are clinically ‘silent’ in terms of
creatinine level and urine output115. However, NGAL has
proved less discriminating in the development of septic-­
associated or adult cardiac-surgery-associated AKI than
in other types of AKI, possibly because subclinical AKI
is common in these settings or because neutrophils
themselves may be a source of NGAL in the setting of
systemic inflammation116. Conversely, tissue inhibitor
of metalloproteinases 2 (TIMP2) and insulin-like growth
factor-binding protein 7 (IGFBP7), which are involved
in G1 cell-cycle arrest, were shown to predict the devel­
opment of severe AKI in both adults and children after
ICU admission in the context of sepsis, surgery, trauma
and nephrotoxic exposure117,118. As G1 cell-cycle arrest
is a common response to tubular injury, these mark­
ers might reflect the presence of injury rather than the
underlying aetiology. However, some differences exist
in terms of the ability of these biomarkers to identify
different AKI insults, with TIMP2 and IGFBP7 being
better discriminants of AKI in the contexts of sepsis and
surgery, respectively 117,119.
Biomarker signatures might also improve disease
prognosis. Although uncertainty remains regarding the
determinants of long-term outcomes in AKI120 (FIG. 4),
various studies have reported that plasma biomarker
alterations correlate with recovery (or lack of it), sug­
gesting targets for therapy and/or methods to identify
patients for clinical trials. Measuring a panel of urinary
markers (NGAL, matrix metalloproteinase protein 9
(MMP9), IL‑18, hepatocyte growth factor (HGF) and
cystatin C) at 14 days after onset of RRT-requiring AKI41
proved very predictive of renal recovery within 60 days
from the start of RRT (area under the curve = 0.88–0.94
in a model that included clinical variables), although
no individual marker performed well in isolation,
which may indicate that differing phenotypes of recov­
ery from AKI were captured by the broader model.
When concentrations of TIMP2 and IGBP7 exceeded
2.0 (ng/ml)2 × 1,000, the risk of death or dialysis at
9 months was increased (HR 2.11, 95% CI 1.37–3.23),
which suggests a connection with both underlying renal
injury and its recovery as well as potential utility in tar­
geting patients for interventions aimed at long-term
outcomes from AKI121.
Conclusions
The various common settings of AKI seem to involve
inflammatory, ischaemic and direct toxin-mediated
injury, which may induce rapidly reversible or persis­
tent renal dysfunction depending on the nature, con­
text and severity of the injury. Furthermore, resolution
of dysfunction may not guarantee resolution of injury.
Whereas these paradigms can be distinct in animal

228 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

models, they are often blurred in clinical practice.
Furthermore, various AKI paradigms, although distinct
from each other, seem to be heterogeneous at the sub­
clinical level, and multiple endophenotypes may exist
within each paradigm. Although sepsis is the leading
cause of AKI, our understanding of the distinguish­
ing features of its pathophysiology remains limited25.
Overall, despite some advances in biomarker develop­
ment, the scientific foundation for distinct phenotypes
of clinical AKI is weak, and early tissue analysis of
human AKI has not been carried out on a large scale.
Our current understanding is that long-term outcomes
of AKI are linked to the development of CKD; however,
this relationship is complex. Although recent evidence
suggests that most deaths after ICU discharge occur
within 90 days42, even in patients who seem to recover
renal function, CKD may develop, conferring ongoing
risks of death, cardiovascular disease and recurrent AKI
in later years. Although we believe that some progress
has been made in terms of targeting individual AKI
syndromes, clinical phenotyping may not be sufficiently
specific, and efforts are needed to develop classifications
that can guide therapy — both at the bedside today and
for the development of new therapies tomorrow.
Finally, a crucial need exists for longitudinal epi­
demiological studies and samples (blood, urine and tis­
sue) that would enable the testing of multiple hypotheses
as to the nature of AKI and its pathophysiology and out­
comes. Only then can new interventions be developed
— existing interventions have failed owing to our lack
of understanding of the diverse underlying pathophysi­
ology of AKI as well as its common and contrasting
features that are poorly encompassed within current
functional definitions of AKI.

1. Kellum, J. A., Bellomo, R. & Ronco, C. Kidney Attack.
JAMA 307, 2265–2266 (2012).
2. Bedford, M., Stevens, P. E., Wheeler, T. W. K. &
Farmer, C. K. T. What is the real impact of acute
kidney injury? BMC Nephrol. 15, 95 (2014).
3. United States Renal Data System. 2013 Annual Data
Report Vol. 1 Ch. 6 (USRDS, 2013).
4. Kane-Gill, S. L. et al. Risk factors for acute
kidney injury in older adults with critical illness:
a retrospective cohort study. Am. J. Kidney Dis. 65,
860–869 (2015).
5. Murugan, R. et al. Acute kidney injury in non-severe
pneumonia is associated with an increased immune
response and lower survival. Kidney Int. 77,
527–535 (2010).
6. Sileanu, F. E. et al. AKI in low-risk versus high-risk
patients in intensive care. Clin. J. Am. Soc. Nephrol.
10, 187–196 (2015).
7. Pickkers, P. et al. The intensive care medicine agenda
on acute kidney injury. Intensive Care Med. 8,
R204–R212 (2017).
8. Barasch, J., Zager, R. & Bonventre, J. V. Acute kidney
injury: a problem of definition. Lancet 389, 779–781
(2017).
9. Uchino, S. et al. Acute renal failure in critically ill
patients: a multinational, multicenter study. JAMA
294, 813–818 (2005).
10. Bagshaw, S. M. et al. A multi-centre evaluation of the
RIFLE criteria for early acute kidney injury in critically
ill patients. Nephrol. Dial. Transplant 23, 1203–1210
(2007).
11. Singbartl, K. & Kellum, J. A. AKI in the ICU: definition,
epidemiology, risk stratification, and outcomes.
Kidney Int. 81, 819–825 (2012).
12. Angus, D. C. et al. Epidemiology of severe sepsis in
the United States: analysis of incidence, outcome,
and associated costs of care. Crit. Care Med. 29,
1303–1310 (2001).
13. Bouchard, J. et al. A prospective international
multicenter study of AKI in the intensive care
unit. Clin. J. Am. Soc. Nephrol. 10, 1324–1331
(2015).
14. Vincent, J.‑L. et al. Sepsis in European intensive
care units: results of the SOAP study. Crit. Care Med.
34, 344–353 (2006).
15. Takasu, O. et al. Mechanisms of cardiac and renal
dysfunction in patients dying of sepsis. Am. J. Respir.
Crit. Care Med. 187, 509–517 (2013).
16. Langenberg, C., Gobe, G., Hood, S., May, C. N. &
Bellomo, R. Renal histopathology during
experimental septic acute kidney injury and recovery.
Crit. Care Med. 42, e58–e67 (2014).
17. De Backer, D., Creteur, J., Preiser, J.‑C., Dubois, M.‑J.
& Vincent, J.‑L. Microvascular blood flow is altered in
patients with sepsis. Am. J. Respir. Crit. Care Med.
166, 98–104 (2002).
18. Spronk, P. E. et al. Nitroglycerin in septic shock after
intravascular volume resuscitation. Lancet 360,
1395–1396 (2002).
19. Donati, A. et al. The aPC treatment improves
microcirculation in severe sepsis/septic shock
syndrome. BMC Anesthesiol. 13, 25 (2013).
20. Langenberg, C., Wan, L., Egi, M., May, C. N. &
Bellomo, R. Renal blood flow in experimental septic
acute renal failure. Kidney Int. 69, 1996–2002
(2006).
21. Di Giantomasso, D., May, C. N. & Bellomo, R. Vital
organ blood flow during hyperdynamic sepsis. Chest
124, 1053–1059 (2003).
22. Prowle, J. R. & Bellomo, R. Sepsis-associated acute
kidney injury: macrohemodynamic and
microhemodynamic alterations in the renal circulation.
Semin. Nephrol. 35, 64–74 (2015).
23. Maiden, M. J. et al. Structure and function of the
kidney in septic shock. a prospective controlled
Exp. Study. Am. J. Respir. Crit. Care Med. 194,
692–700 (2016).
24. Benes, J. et al. Searching for mechanisms that matter
in early septic acute kidney injury: an experimental
study. Crit. Care 15, R256 (2011).
25. Gomez, H. et al. A unified theory of sepsis-induced
acute kidney injury: inflammation, microcirculatory
dysfunction, bioenergetics, and the tubular cell
adaptation to injury. Shock, 41, 3–11 (2014).
26. Verma, S. K. & Molitoris, B. A. Renal endothelial injury
and microvascular dysfunction in acute kidney injury.
Semin. Nephrol. 35, 96–107 (2015).
27. Post, E. H., Kellum, J. A., Bellomo, R. & Vincent, J.‑L.
Renal perfusion in sepsis: from macro- to
microcirculation. Kidney Int. 91, 45–60 (2017).
28. Calzavacca, P., Evans, R. G., Bailey, M., Bellomo, R. &
May, C. N. Cortical and medullary tissue perfusion and
oxygenation in experimental septic acute kidney injury.
Crit. Care Med. 43, e431–e439 (2015).
29. Lankadeva, Y. R. et al. Intrarenal and urinary
oxygenation during norepinephrine resuscitation in
ovine septic acute kidney injury. Kidney Int. 90,
100–108 (2016).
30. Bezemer, R. et al. Real-time assessment of renal
cortical microvascular perfusion heterogeneities using
near-infrared laser speckle imaging. Opt. Express 18,
15054–15061 (2010).
31. Casellas, D. & Mimran, A. Aglomerular pathways in
intrarenal microvasculature of aged rats. Am. J. Anat.
156, 293–299 (1979).
32. Dellepiane, S., Marengo, M. & Cantaluppi, V.
Detrimental cross-talk between sepsis and acute kidney
injury: new pathogenic mechanisms, early biomarkers
and targeted therapies. Crit. Care 20, 61 (2016).
33. Ghosh, C. C. et al. Gene control of tyrosine kinase TIE2
and vascular manifestations of infections. Proc. Natl
Acad. Sci. USA 113, 2472–2477 (2016).
34. Mariano, F. et al. Circulating plasma factors induce
tubular and glomerular alterations in septic burns
patients. Crit. Care 12, R42 (2008).
35. Mehta, R. L. et al. Sepsis as a cause and consequence
of acute kidney injury: program to improve care in
acute renal disease. Intensive Care Med. 37, 241–248
(2010).
36. Singbartl, K. et al. Differential effects of kidney–lung
cross-talk during acute kidney injury and bacterial
pneumonia. Kidney Int. 80, 633–644 (2011).
37. Johnston, O., Zalunardo, N., Rose, C. & Gill, J. S.
Prevention of sepsis during the transition to dialysis
may improve the survival of transplant failure patients.
J. Am. Soc. Nephrol. 18, 1331–1337 (2007).
38. Sarnak, M. J., Sarnak, M. J., Jaber, B. L. & Jaber, B. L.
Mortality caused by sepsis in patients with end-stage
renal disease compared with the general population.
Kidney Int. 58, 1758–1764 (2000).
39. Vanholder, R. & Ringoir, S. Infectious morbidity and
defects of phagocytic function in end-stage renal
disease: a review. J. Am. Soc. Nephrol. 3, 1541–1554
(1993).
40. Winters, B. D. et al. Long-term mortality and quality of
life in sepsis: a systematic review. Crit. Care Med. 38,
1276–1283 (2010).
41. Murugan, R. et al. Plasma inflammatory and apoptosis
markers are associated with dialysis dependence and
death among critically ill patients receiving renal
replacement therapy. Nephrol. Dial. Transplant 29,
1854–1864 (2014).
42. Kellum, J. A. et al. The effects of alternative
resuscitation strategies on acute kidney injury in
patients with septic shock. Am. J. Respir. Crit. Care
Med. 193, 281–287 (2016).
43. Brunner, F. P., Thiel, G., Hermle, M., Bock, H. A. &
Mihatsch, M. J. Long-term enalapril and verapamil
in rats with reduced renal mass. Kidney Int. 36,
969–977 (1989).
44. Ikoma, M., Kawamura, T., Kakinuma, Y., Fogo, A. &
Ichikawa, I. Cause of variable therapeutic efficiency of
angiotensin converting enzyme inhibitor on glomerular
lesions. Kidney Int. 40, 195–202 (1991).
45. Cheng, S.‑Y. et al. Losartan reduces ensuing chronic
kidney disease and mortality after acute kidney injury.
Sci. Rep. 6, 34265 (2016).
46. Carmichael, P. & Carmichael, A. R. Acute renal failure in
the surgical setting. ANZ J. Surg. 73, 144–153 (2003).
47. Shusterman, N. et al. Risk factors and outcome of
hospital-acquired acute renal failure. Clinical
epidemiologic study. Am. J. Med. 83, 65–71 (1987).
48. Weiser, T. G. et al. Estimate of the global volume of
surgery in 2012: an assessment supporting improved
health outcomes. Lancet 385 (Suppl. 2), S11 (2015).
49. Weiser, T. G. et al. An estimation of the global volume
of surgery: a modelling strategy based on available
data. Lancet 372, 139–144 (2008).
50. Kork, F. et al. Minor postoperative increases of
creatinine are associated with higher mortality and
longer hospital length of stay in surgical patients.
Anesthesiology 123, 1301–1311 (2015).
51. Korenkevych, D. et al. The pattern of longitudinal
change in serum creatinine and 90‑day mortality
after major surgery. Ann. Surg. 263, 1219–1227
(2016).
52. Grams, M. E. et al. Acute kidney injury after major
surgery: a retrospective analysis of veterans health
administration data. Am. J. Kidney Dis. 67, 872–880
(2016).
53. Grams, M. E. et al. Candidate surrogate end points for
ESRD after AKI. J. Am. Soc. Nephrol. 27, 2851–2859
(2016).
54. O’Connor, M. E. et al. Acute kidney injury and mortality
1 year after major non-cardiac surgery. Br. J. Surg.
104, 868–876 (2017).

NATURE REVIEWS | NEPHROLOGY VOLUME 14 | APRIL 2018 | 229

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

55. O’Connor, M. E., Kirwan, C. J., Pearse, R. M. &
Prowle, J. R. Incidence and associations of acute kidney
injury after major abdominal surgery. Intensive Care
Med. 42, 521–530 (2016).
56. Thiele, R. H., Isbell, J. M. & Rosner, M. H. AKI
associated with cardiac surgery. Clin. J. Am. Soc.
Nephrol. 10, 500–514 (2015).
57. Chertow, G. M. et al. Preoperative renal risk
stratification. Circulation 95, 878–884 (1997).
58. Prowle, J. R. et al. Preoperative renal dysfunction and
mortality after non-cardiac surgery. Br. J. Surg. 103,
1316–1325 (2016).
59. Kheterpal, S. et al. Development and validation of an
acute kidney injury risk index for patients undergoing
general surgery: results from a national data set.
Anesthesiology 110, 505–515 (2009).
60. Brienza, N., Giglio, M. T., Marucci, M. & Fiore, T.
Does perioperative hemodynamic optimization protect
renal function in surgical patients? A meta-analytic
study. Crit. Care Med. 37, 2079–2090 (2009).
61. Prowle, J. R., Chua, H.‑R., Bagshaw, S. M. &
Bellomo, R. Clinical review: volume of fluid
resuscitation and the incidence of acute kidney injury
— a systematic review. Crit. Care 16, 230 (2012).
62. Sun, L. Y., Wijeysundera, D. N., Tait, G. A. &
Beattie, W. S. Association of intraoperative
hypotension with acute kidney injury after elective
noncardiac surgery. Anesthesiology 123, 515–523
(2015).
63. Walsh, M. et al. Relationship between intraoperative
mean arterial pressure and clinical outcomes after
noncardiac surgery: toward an empirical definition of
hypotension. Anesthesiology 119, 507–515 (2013).
64. Salmasi, V. et al. Relationship between intraoperative
hypotension, defined by either reduction from baseline
or absolute thresholds, and acute kidney and
myocardial injury after noncardiac surgery:
a retrospective cohort analysis. Anesthesiology 126,
47–65 (2017).
65. Saotome, T., Ishikawa, K., May, C. N., Birchall, I. E. &
Bellomo, R. The impact of experimental hypoperfusion
on subsequent kidney function. Intensive Care Med.
36, 533–540 (2010).
66. Siegemund, M. et al. Aortic cross-clamping and
reperfusion in pigs reduces microvascular oxygenation
by altered systemic and regional blood flow
distribution. Anesth. Analg. 111, 345–353 (2010).
67. Akira, S. & Takeda, K. Toll-like receptor signalling.
Nat. Rev. Immunol. 4, 499–511 (2004).
68. Cremer, J. et al. Systemic inflammatory response
syndrome after cardiac operations. Ann. Thorac. Surg.
61, 1714–1720 (1996).
69. Rosner, M. H., Portilla, D. & Okusa, M. D. Cardiac
surgery as a cause of acute kidney injury: pathogenesis
and potential therapies. J. Intensive Care Med. 23,
3–18 (2008).
70. Zhang, W. R. et al. Plasma IL‑6 and IL‑10
concentrations predict AKI and long-term mortality in
adults after cardiac surgery. J. Am. Soc. Nephrol. 26,
3123–3132 (2015).
71. Andres-Hernando, A. et al. Cytokine production
increases and cytokine clearance decreases in mice
with bilateral nephrectomy. Nephrol. Dial. Transplant
27, 4339–4347 (2012).
72. Lahiri, R. et al. Systemic inflammatory response
syndrome after major abdominal surgery predicted by
early upregulation of TLR4 and TLR5. Ann. Surg. 263,
1028–1037 (2016).
73. Hobson, C. E. et al. Acute kidney injury is associated
with increased long-term mortality after cardiothoracic
surgery. Circulation 119, 2444–2453 (2009).
74. Bihorac, A. et al. National surgical quality
improvement program underestimates the risk
associated with mild and moderate postoperative
acute kidney injury. Crit. Care Med. 41, 2570–2583
(2013).
75. van Kuijk, J.‑P. et al. Temporary perioperative decline
of renal function is an independent predictor for
chronic kidney disease. Clin. J. Am. Soc. Nephrol. 5,
1198–1204 (2010).
76. Bihorac, A. et al. Incidence, clinical predictors,
genomics, and outcome of acute kidney injury among
trauma patients. Ann. Surg. 252, 158–165 (2010).
77. Bihorac, A. et al. Long-term risk of mortality and acute
kidney injury during hospitalization after major surgery.
Ann. Surg. 249, 851–858 (2009).
78. Ishani, A. et al. The magnitude of acute serum
creatinine increase after cardiac surgery and the risk
of chronic kidney disease, progression of kidney
disease, and death. Arch. Intern. Med. 171, 226–233
(2011).
79. Leacche, M. et al. Outcomes in patients with normal
serum creatinine and with artificial renal support for
acute renal failure developing after coronary artery
bypass grafting. Am. J. Cardiol. 93, 353–356 (2004).
80. Thakar, C. V. et al. Predicting acute renal failure after
cardiac surgery: validation and re‑definition of a risk-
stratification algorithm. Hemodial. Int. 7, 143–147
(2003).
81. Oliver, J., MacDowell, M. & Tracy, A. The pathogenesis
of acute renal failure associated with traumatic and
toxic injury; renal ischemia, nephrotoxic damage and
the ischemic episode. J. Clin. Invest. 30, 1307–1439
(1951).
82. Sandroni, C. et al. Acute kidney injury after cardiac
arrest: a systematic review and meta-analysis of clinical
studies. Minerva Anestesiol. 82, 989–999 (2016).
83. Bonventre, J. V. & Yang, L. Cellular pathophysiology
of ischemic acute kidney injury. J. Clin. Invest. 121,
4210–4221 (2011).
84. Sharfuddin, A. A. & Molitoris, B. A. Pathophysiology
of ischemic acute kidney injury. Nat. Rev. Nephrol. 7,
189–200 (2011).
85. Prowle, J. R., Ishikawa, K., May, C. N. & Bellomo, R.
Renal blood flow during acute renal failure in man.
Blood Purif. 28, 216–225 (2009).
86. Mullens, W. et al. Importance of venous congestion
for worsening of renal function in advanced
decompensated heart failure. J. Am. Coll. Cardiol. 53,
589–596 (2009).
87. Damman, K. et al. Increased central venous pressure is
associated with impaired renal function and mortality
in a broad spectrum of patients with cardiovascular
disease. J. Am. Coll. Cardiol. 53, 582–588 (2009).
88. Dalfino, L., Tullo, L., Donadio, I., Malcangi, V. &
Brienza, N. Intra-abdominal hypertension and acute
renal failure in critically ill patients. Intensive Care
Med. 34, 707–713 (2008).
89. O’Connor, M. E. & Prowle, J. R. Fluid Overload.
Crit. Care Clin. 31, 803–821 (2015).
90. Lauridsen, M. D. et al. Acute kidney injury treated with
renal replacement therapy and 5‑year mortality after
myocardial infarction-related cardiogenic shock:
a nationwide population-based cohort study. Crit. Care
19, 452 (2015).
91. Taber, S. S. & Mueller, B. A. Drug-associated renal
dysfunction. Crit. Care Clin. 22, 357–374 (2006).
92. Hou, S. H., Bushinsky, D. A., Wish, J. B., Cohen, J. J. &
Harrington, J. T. Hospital-acquired renal insufficiency:
a prospective study. Am. J. Med. 74, 243–248 (1983).
93. Hui-Stickle, S., Brewer, E. D. & Goldstein, S. L. Pediatric
ARF epidemiology at a tertiary care center from 1999
to 2001. Am. J. Kidney Dis. 45, 96–101 (2005).
94. Liaño, F., Junco, E., Pascual, J., Madero, R. & Verde, E.
The spectrum of acute renal failure in the intensive care
unit compared with that seen in other settings. The
Madrid Acute Renal Fail. Study Group. Kidney Int.
Suppl. 66, S16–S24 (1998).
95. Nash, K., Hafeez, A. & Hou, S. Hospital-acquired renal
insufficiency. Am. J. Kidney Dis. 39, 930–936 (2002).
96. Schetz, M., Dasta, J., Goldstein, S. & Golper, T. Drug-
induced acute kidney injury. Curr. Opin. Crit. Care 11,
555–565 (2005).
97. Rosner, M. H. & Perazella, M. A. Acute Kidney Injury in
Patients with Cancer. N. Engl. J. Med. 376, 1770–1781
(2017).
98. de Broe, M. E. & Porter, G. A. Clinical Nephrotoxins
(eds De Broe, M. E., Porter, G. A., Bennett, W. M. &
Deray, G.) (Springer, 2008).
99. Ozkok, A. & Edelstein, C. L. Pathophysiology of
cisplatin-induced acute kidney injury. Biomed. Res. Int.
2014, 967826 (2014).
100. Faubel, S. et al. Cisplatin-induced acute renal failure is
associated with an increase in the cytokines interleukin
(IL)-1beta, IL‑18, IL‑6, and neutrophil infiltration in the
kidney. J. Pharmacol. Exp. Ther. 322, 8–15 (2007).
101. Ramesh, G. & Reeves, W. B. TNF-alpha mediates
chemokine and cytokine expression and renal injury in
cisplatin nephrotoxicity. J. Clin. Invest. 110, 835–842
(2002).
102. Sánchez-González, P. D., López-Hernández, F. J.,
López-Novoa, J. M. & Morales, A. I. An integrative view
of the pathophysiological events leading to cisplatin
nephrotoxicity. Crit. Rev. Toxicol. 41, 803–821 (2011).
103. Skinner, R. et al. Cisplatin dose rate as a risk factor
for nephrotoxicity in children. Br. J. Cancer 77,
1677–1682 (1998).
104. Menon, S., Kirkendall, E. S., Nguyen, H. &
Goldstein, S. L. Acute kidney injury associated with
high nephrotoxic medication exposure leads to chronic
kidney disease after 6 months. J. Pediatr. 165,
522–527 (2014).
105. Kellum, J. A., Sileanu, F. E., Bihorac, A., Hoste, E. A. J.
& Chawla, L. S. Recovery after acute kidney injury.
Am. J. Respir. Crit. Care Med. 195, 784–791 (2017).
106. Molitoris, B. A. et al. Design of clinical trials in AKI:
a report from an NIDDK workshop. Trials of patients
with sepsis selected hospital settings. Clin. J. Am. Soc.
Nephrol. 7, 856–860 (2012).
107. Chawla, L. S., Eggers, P. W., Star, R. A. & Kimmel, P. L.
Acute kidney injury and chronic kidney disease as
interconnected syndromes. N. Engl. J. Med. 371,
58–66 (2014).
108. Yuen, P. S. T., Jo, S.‑K., Holly, M. K., Hu, X. & Star, R. A.
Ischemic and nephrotoxic acute renal failure are
distinguished by their broad transcriptomic responses.
Physiol. Genom. 25, 375–386 (2006).
109. Mar, D. et al. Heterogeneity of epigenetic changes at
ischemia/reperfusion- and endotoxin-induced acute
kidney injury genes. Kidney Int. 88, 734–744 (2015).
110. Matejovic, M. et al. Molecular differences in
susceptibility of the kidney to sepsis-induced kidney
injury. BMC Nephrol. 18, 183 (2017).
111. Fan, P.‑C., Chen, C.‑C., Chen, Y.‑C., Chang, Y.‑S. &
Chu, P.‑H. MicroRNAs in acute kidney injury.
Hum. Genom. 10, 29 (2016).
112. Xu, K. et al. Unique transcriptional programs identify
subtypes of AKI. J. Am. Soc. Nephrol. 28, 1729–1740
(2017).
113. Paragas, N. et al. The Ngal reporter mouse detects the
response of the kidney to injury in real time. Nat. Med.
17, 216–222 (2011).
114. Nickolas, T. L. et al. Diagnostic and prognostic
stratification in the emergency department using
urinary biomarkers of nephron damage: a multicenter
prospective cohort study. J. Am. Coll. Cardiol. 59,
246–255 (2012).
115. Haase, M., Kellum, J. A. & Ronco, C. Subclinical AKI —
an emerging syndrome with important consequences.
Nat. Rev. Nephrol. 8, 735–739 (2012).
116. Borregaard, N. & Cowland, J. B. Granules of the
human neutrophilic polymorphonuclear leukocyte.
Blood 89, 3503–3521 (1997).
117. Kashani, K. et al. Discovery and validation of cell cycle
arrest biomarkers in human acute kidney injury.
Crit. Care 17, R25 (2013).
118. Bihorac, A. et al. Validation of cell-cycle arrest
biomarkers for acute. Am. J. Respir. Crit. Care Med.
189, 932–939 (2014).
119. Honore, P. M. et al. Urinary tissue inhibitor of
metalloproteinase‑2 and insulin-like growth factor-
binding protein 7 for risk stratification of acute kidney
injury in patients with sepsis. Crit. Care Med. 44,
1851–1860 (2016).
120. Murugan, R. & Kellum, J. A. Acute kidney injury:
what’s the prognosis? Nat. Rev. Nephrol. 7, 209–217
(2011).
121. Koyner, J. L. et al. Tissue inhibitor metalloproteinase‑2
(TIMP‑2)⋅IGF-binding protein‑7 (IGFBP7) levels are
associated with adverse long-term outcomes in
patients with AKI. J. Am. Soc. Nephrol. 26,
1747–1754 (2015).
122. Bellomo, R. et al. Acute renal failure — definition,
outcome measures, animal models, fluid therapy and
information technology needs: the Second
International Consensus Conference of the Acute
Dialysis Quality Initiative (ADQI) Group. Crit. Care 8,
R204–R212 (2004).
123. Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Work Group. KDIGO clinical
practice guideline for acute kidney injury. Kidney Int.
Suppl. 2, 1–138 (2012).
123. Kellum, J. A. et al. Classifying AKI by urine output
versus serum creatinine level. J. Am. Soc. Nephrol. 26,
2231–2238 (2015).
124. Kaddourah, A., Basu, R. K., Bagshaw, S. M.,
Goldstein, S. L. & AWARE Investigators. Epidemiology
of acute kidney injury in critically ill children and young
adults. N. Engl. J. Med. 376, 11–20 (2017).
Author contributions
J.A.K. and J.R.P. researched the data, discussed the article’s
content, wrote the text and reviewed or edited the article
before submission.
Competing interests statement
J.A.K. has received consulting fees and grant support from
Astute Medical and Bioporto. J.R.P. declares no competing
interests.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

230 | APRIL 2018 | VOLUME 14 www.nature.com/nrneph

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.