Introduction

Acute kidney injury (AKI) can portend an incremental risk for short- and long-term complications including fatal outcome [1], nephron loss along with risk of chronic kidney disease (CKD) or end-stage kidney disease (ESKD) [2]. In intensive care unit (ICU) settings, up to 60% of patients develop AKI and up to 12% of patients require renal replacement therapy (RRT) [1]. Although the vast majority of AKI is multifactorial, with sepsis, shock and nephrotoxicity accounting for most episodes [1, 3], it is important to remember that critically ill patients may also present with specific types of AKI requiring targeted diagnostic work-up and treatment.

In this review, we will summarize uncertainties, current knowledge and recent advances regarding specific types of AKI, and outline available tools that may be used during the diagnostic work-up.

Diagnosis of AKI and Input of biomarkers

The Kidney Disease Improving Global Outcomes (KDIGO) guideline defines AKI as an abrupt decrease of kidney function over a period of 7 days or less based on a rise in serum creatinine or an episode of oliguria [4]. For patients whose renal function returns to baseline within 48 h, the Acute Disease Quality Initiative (ADQI) 16 conference proposed to use the term “rapid reversal” of AKI, while “persistent AKI” is characterized by a duration beyond 48 h [5]. In patients with persistent AKI, it is essential to determine its underlying aetiology. In addition, it is recommended to evaluate the haemodynamic and volume status, and to identify complications that may indicate the need for RRT. A period of renal dysfunction that persists for >7 days is categorized as Acute Kidney Disease (AKD) [4, 5], and the term Chronic Kidney Disease (CKD) is used for patients with kidney disease persistent after 90 days [4, 5].

The KDIGO definition for AKI is based on creatinine and oliguria, two imperfect markers. Other biomarkers may indicate kidney damage earlier than conventional biomarkers. At present, two kidney damage biomarkers are available for clinical use: neutrophil gelatinase-associated lipocalin (NGAL) can be measured in urine and blood, and the combination of urinary insulin-like growth factor-binding protein 7 (IGFBP7) and tissue inhibitor of metalloproteinases 2 (TIMP-2) can be measured using the NephroCheck® point of care device. Although the use of these biomarkers might allow earlier AKI detection [6,7,8], their performance is limited and their impact on relevant patient or kidney outcomes is uncertain, as is their usefulness for other endpoints such as use of RRT and long-term kidney outcomes [9].

Although the importance of differentiating functional volume-responsive AKI from intrinsic parenchymal diseases is classically emphasized, several studies have pointed out the frequent coexistence of intrinsic and pre-renal injury in critically ill patients [10,11,12,13,14]. This translates into poor performance of urinary indices and limited performance of kidney damage markers in predicting volume-responsive AKI [10,11,12,13,14]. Urine sediment analysis remains, however, mandatory to detect proteinuria, haematuria, leukocyturia or eosinophiluria, that may suggest specific causes of AKI.

Clinical imaging

Renal imaging is a mandatory component of AKI diagnostic work-up. Depending on the technique, it enables to evaluate organ morphology and provides insights in function, perfusion and possibly aetiology.

Ultrasonography

Renal ultrasonography (US) represents the first-line imaging modality in most cases of AKI [15, 16]. It is inexpensive, non-invasive, widely available, can be performed at the bedside and may influence further investigations and management.

Conventional (B-mode) imaging

Conventional (B-mode) is the basic examination mode. It generates grey-scale images based on the property of sound waves to reflect at interfaces of media of different densities. This mode permits the evaluation of longitudinal size and parenchyma echogenicity and demonstrates (or rules out) the presence of hydronephrosis or cysts.

Kidney size is typically measured in its long axis (bipolar length). Normal values range from 9 to 12 cm according to patient size and gender. Small kidney size is suggestive of underlying CKD, while enlarged kidneys might be observed in infiltrative diseases, renal vein thrombosis or acute rejection in transplanted kidneys.

Renal echogenicity is evaluated by comparison with that of adjacent tissues (liver or spleen). Decreased echogenicity can be physiologic but might be associated with pathological processes such as oedema. Hyperechogenicity almost always indicates diffuse kidney parenchymal pathology (infiltrative diseases, inflammatory states). Chronic kidney disease is often associated with increased brightness since fibrous tissues increase echogenicity.

Urinary tract obstruction represents a relatively easily reversible cause of AKI. Although less commonly encountered in ICU patients, it should be ruled out in all patients with AKI with suggestive history or lack of clear causes of AKI. On US, the collecting system of the kidney is not normally visible unless significant hydronephrosis is present. Hypovolemia, early obstruction and retroperitoneal tumours or fibrosis may lead to false negative results. False positive findings can be observed in pregnancy and in patients with diabetes insipidus, vesico-ureteral reflux, megacystis-megaureter syndrome, full bladder and urinary tract infection.

Although unrelated to AKI, cysts are often encountered on renal US examination. Benign cysts are well delineated with a clear content, while complex cysts contain septa or wall thickening. Solid masses are usually neoplastic. Specialist referral is required when such images are encountered.

Doppler-based resistive indices (RI)

Adding Doppler technology to grey-scale images enables visualization of intra-renal vasculature and computation of dynamic parameters. The most studied of those is the intra-renal resistive index (RI). RI can be calculated based on two parameters of flow (the peak systolic shift and minimum diastolic shift) which can be measured in the arcuate or interlobar arteries in pulsed wave Doppler mode [RI = (peak systolic velocity − end diastolic velocity)/peak velocity]. Renal RI is a simple, non-invasive tool easy to use at the patient bedside (Fig. 1). It has been shown to have an excellent inter-observer reproducibility but limited precision [17]. RI is considered normal if <0.70; however, the normal RI range is age-dependent [18].

Fig. 1
figure 1

Results of a renal colour-Doppler ultrasonography showing renal vascularization (a). RI measurement using pulsed wave Doppler (b) [18]

Unfortunately, the physiological and clinical significance of RI remain debated. Initially considered an indicator of renal vascular resistance and blood flow, changes in RI have been proposed to help titrate norepinephrine to better tailor mean arterial pressure in ICU patients [19]. However, in vivo and in vitro studies have suggested a correlation with vascular compliance (vascular distensibility) [18]. In addition, RI is largely influenced by extra-renal factors such as cardiac output, heart rate, PaCO2 and PaO2, and renal interstitial and intra-abdominal pressures [18]. As a result, the use of RI for AKI diagnosis is currently not recommended and further studies are required to establish its role in clinical practice [20].

Contrast-enhanced ultrasonography (CEUS)

CEUS is a newer ultrasound-based imaging modality that utilizes dedicated US contrast agents. US contrast agents are made of gas microbubbles, which act as potent ultrasound reflectors. Commercially available preparations have high stability and very small-sized microbubbles. They are injected into the peripheral vein and can be visualized in arteries and capillaries. As such, CEUS enables delicate visualization of organs’ macro- and microcirculation (Fig. 2). Cortical necrosis can be detected using CEUS. Methods have been proposed to use CEUS-derived techniques to measure organ perfusion. These methods have been validated in animal models and healthy volunteers [21]. To date, there are limited data on its possible utilization in critically ill patients. Studies have shown that CEUS was feasible and safe in such contexts [22] but with potentially conflicting results [23]. Further research is required to establish the role of CEUS in ICU.

Fig. 2
figure 2

Examples of renal contrast-enhanced ultrasound images: with direct application of the US probe on the kidney (a) or in a critically ill patient (b)

Computerized tomography (CT)

CT is increasingly used in the ICU. Despite the potential nephrotoxicity of iodinated contrast agents, in patients with septic AKI, contrast CT may be necessary to identify the source of sepsis. Non-contrast CT remains the gold standard for the diagnosis of urolithiasis and complicated pyelonephritis. It enables detection of underlying renal or abdominal abnormalities in patients with AKI (cysts, renal carcinoma, aortic aneurysms, etc.) and detection of hydronephrosis. Finally, last generation triphasic CT (functional CT), can allow assessment of GFR and renal blood flow, but the exact role in critically ill patients needs to be determined in future studies [24].

Magnetic resonance imaging (MRI)

MRI techniques allow detailed assessment of renal anatomy, tumors and infections. Some techniques enable quantification of renal perfusion and differentiation between cortical and medullar areas. Unfortunately, these techniques require the injection of gadolinium, which has been associated with the risk of nephrogenic systemic fibrosis in patients with altered renal function [25]. Non-contrast methods such as cine-contrast MRI or BOLD MRI have been proposed and tested in critically ill patients [26]. To date, given the cost, availability and length of the protocols, MRI does not play a role in the routine diagnostic workup of AKI outside research studies.

Renal biopsy in the intensive care setting

Given that the common heterogeneity in causes of AKI, histological diagnosis is often not considered, and indeed renal biopsy, despite being an essential tool in non-critical care nephrology, is rarely performed. However, as outlined above, there are specific situations where histopathological diagnosis does deserve consideration. In spite of the improved AKI recognition and assessment of severity provided by recently developed scoring systems such as KDIGO, recent evidence demonstrates a lack of correlation between KDIGO classification and actual AKI aetiology [27]. These studies emphasize that kidney biopsy should be considered more frequently than currently employed in order to better understand AKI and avoid missing specific causes of AKI and treatment opportunities to reduce the risk of subsequent CKD.

Indications in the ICU

Although the common causes of AKI within the ICU can be diagnosed clinically without histological confirmation, the presumptive diagnosis risks ignoring treatable conditions such as acute interstitial nephritis (AIN). Occasionally, AKI may be due to a specific multi-systemic disorder or conditions fulfilling standard indications for renal biopsy such as nephritic and/or nephrotic syndrome, or rapidly progressive AKI of unclear aetiology. Although renal biopsy should probably be discussed after initial diagnostic work-up in patients with AKI without obvious aetiologies or in patients with renal parenchymal disease, clinicians must ensure that the benefits of renal biopsy outweigh the risks. Thus, renal biopsy is usually not done if the expected histological findings are assumed neither to guide a specific treatment nor to determine a change in treatment. Main indications for renal biopsy in the critically ill patient include persistent loss of kidney function, proteinuria greater than 3 g/day or a clinical picture suggestive of systemic disease [28].

Precautions and contraindications

Careful patient selection as well as the use of real-time US has minimized the risks associated with percutaneous renal biopsy, but it does carry a morbidity and mortality risk [29]. Significant complications include haemorrhage, infection and arteriovenous fistula formation occurring in 3–13% of cases and a mortality risk of up to 0.2% [30]. Although controlling blood pressure for preventing complications has not been firmly demonstrated, maintaining blood pressure under 140/90 mmHg on the day of renal biopsy is a classical objective. Haemostatic abnormalities and particularly a decreased platelet count have been reported to be associated with higher complication rates [31], but whether there is a role for routinely measuring bleeding time and administering desmopressin prior to renal biopsy remains a matter of debate. In patients with coagulation disorders or taking anticoagulants or antiplatelet treatment, we suggest managing patients according to national guidelines on perioperative management of patients at high risk of bleeding. The “classic” contraindications to renal biopsy are outlined in Table 1.

Table 1 Usual contra-indications to percutaneous biopsy

Technique

Renal biopsy can be performed at the bedside under real-time US guidance in a patient placed in the prone position. It should only be done by clinicians that are expert in performing the procedure. In patients on mechanical ventilation, a ventilator pause can be used to stop kidney movement with respiration. Non-percutaneous renal biopsy techniques may be considered where there are absolute contraindications to the percutaneous technique. These include open renal biopsy [32], laparascopic approach [33] and transjugular approach, the latter carrying the risk of inadequate tissue sampling [34]. This is relevant in that renal biopsy findings must be interpreted in the clinical context and in tandem with laboratory features, as prognostication based upon renal pathology alone is affected by the sample size and may be less accurate in biopsies with few glomeruli (i.e. ≤5).

Renal biopsy on the ICU: experience

Only two studies, both from France, have examined the impact of renal biopsy in the critically ill. Augusto et al. preformed a multicentre retrospective analysis over a 10-year period from ten ICUs [31]. Among 77 patients who underwent a renal biopsy (88% on native kidneys), 50% of non-transplanted patients had a specific diagnosis including glomerulonephritis (22%), acute vascular nephritis (10%), AIN (3%), and deposit disease (3%) [31]. The occurrence of AKI before hospital admission, the presence of clinical extra-renal signs, and the absence of classical contributory factors of acute tubular necrosis were associated with renal biopsy yielding a specific diagnosis [31]. In this study, one-fifth of the non-transplanted patients had their treatment modified and one-sixth had a specific treatment stopped. Similarly, Philipponnet et al. described a retrospective multicenter study over 10 years in five French ICUs. Of the 54 kidney biopsies preformed, acute tubular necrosis was found in 46%, glomerulonephritis in 25%, acute vascular nephritis in 20%, AIN in 10%, and deposit disease in 5% [28]. Of note, in most cases, interstitial nephritis had not been suspected prior to renal biopsy, and renal biopsy results contributed to the management in over 70% of cases [28].

Interestingly, in both studies, biopsy findings confirmed clinical suspicion, suggesting a selection bias which may have contributed to the high diagnostic and therapeutic contribution. Patient inclusion in these cohorts was highly selective with biopsies performed in less than 1% of patients with AKI on ICU, and as such no recommendations can be drawn as to who may require biopsy. It must be noted, however, that in both these studies complications related to biopsy were more frequently observed in ICU patients when compared to the reported incidences in nephrology patients, including a high risk of severe and even fatal bleeding [28, 31].

These studies demonstrate that in highly selected cases renal biopsy may carry meaningful and clinically relevant information at a cost of significant risk of severe complications. Most patients underwent a percutaneous procedure, and it remains unclear whether complications could have been prevented by using a transjugular approach. Similarly, it is also possible that the development of new biomarkers or techniques to image renal injury will reduce the need for renal biopsies [35].

Acute kidney injury in specific settings and diagnosing unexpected aetiologies

Several situations or settings deserve particular attention. In addition, some patients may present with AKI without associated manifestations or obvious predisposing factors. In these patients, a basic diagnostic work-up should be performed to avoid missing specific types of AKI (Fig. 3). This also applies to AKI in transplant recipients (Table 2). Additionally, several ICU conditions, including prolonged and complex surgery, comatose states or status epilepticus, and exposure to some drugs such as cocaine and envenomation may predispose to rhabdomyolysis [36]. To ensure prompt implementation of preventive measures, measuring creatinine kinase levels should be part of the routine testing in such high-risk patients. Last, intensivists should be aware of differential diagnosis of AKI in specific conditions such bile acid-induced cholemic nephropathy in patients with advanced liver disease and AKI unresponsive to vasopressin analog [37].

Fig. 3
figure 3

First-line diagnostic work-up in patients with AKI without obvious predisposing factor

Table 2 Complications not to be missed in renal transplant recipients with AKI

Acute glomerulonephritis—systemic diseases: vasculitis

Acute glomerulonephritis or rapidly progressive glomerulonephritis (RPGN) are rare forms of renal diseases manifesting by a sub-acute and progressive rise in serum creatinine. In the majority of cases, RPGN is encountered in the context of systemic diseases with multiple organ involvement. The pathological lesion of RPGN is extracapillary glomerular proliferation (crescents). The usual classification is based on immunofluorescence examination of the glomerulus [38]. Goodpasture’s disease is due to antibodies directed against the glomerular basement membrane (GBM). It typically presents as RPGN and intra-alveolar haemorrhage. Immunofluorescence examination shows linear deposition of IgG along the GBM.

Immune complex diseases are characterized by granular deposits of various immunoglobulins and complement fractions. Several diseases share this presentation, including systemic lupus erythematosus, Henoch–Schönlein purpura, cryoglobulinemia and post-infectious glomerulonephritis (in particular endocarditis). Lastly, pauci-immune glomerulonephritis is characterized by the absence of immune deposits on immunofluorescence and is associated with antineutrophil cytoplasmic antibodies (ANCA)-positive vasculitides.

Although RGPN is uncommon in ICU patients, its exact prevalence remains unknown. Outside the ICU setting, a study showed that acute glomerulonephritis was diagnosed in 5.6% of 748 patients admitted over a year in a nephrology unit [39]. Frequency may rise up to 31% in patients aged 65 years or older [40]. Among ICU patients undergoing a renal biopsy, RGPN was observed in 25% of cases [31]. Finally, in a recent study on 363 patients admitted to the ICU with rheumatic disease, one-third of them were admitted for disease activity and less than 10% for RGPN [41]. In this subset of patients, ANCA-associated vasculitis were the most frequent diseases [41]. Studies on ANCA-associated vasculitis in ICU patients showed a higher incidence of granulomatosis with polyangiitis than microscopic polyangiitis [42].

Patient and renal outcomes depend on rapid initiation of a specific treatment. The combination of a low incidence, an acute presentation mimicking more frequent pathologies, and the need for emergent specific treatment make RPGN diagnosis particularly challenging in ICU patients. RPGN are classically characterized by development of AKI or AKD over a period of days to a few weeks, arterial hypertension, proteinuria and haematuria. Hypertension may be absent especially in ICU patients, and proteinuria and haematuria are often non-specific. [43]. Acute respiratory failure with lung consolidation and rising serum creatinine (the so-called “pulmonary-renal syndrome”) may occasionally be due to RGPN and intra-alveolar haemorrhage associated with vasculitis [44]. Non-renal symptoms evocative of systemic diseases including weight loss, symptoms lasting over several weeks, haemoptysis with anemia, arthritis, skin involvement, sinusitis and peripheral neurologic disease are important clues that should alert the ICU physician [31, 44]. The diagnosis of RPGN relies on examination of a renal biopsy, which in this case is a medical emergency. However, alternate diagnostic approaches based on immunological blood assays may be preferred when available. An evocative clinical picture should prompt the measurement of ANCA, anti-GBM antibodies, cryoglobulins, anti-nuclear and anti-DNA antibodies and complement fractions C3, C4 and CH50. It must be noted, however,that these auto-antibodies should be interpreted cautiously since low titres can be observed in inflammatory conditions.

In ICU patients with suspected vasculitis, urgent treatment is required. In case of probable or confirmed Goodpasture disease or ANCA-associated vasculitis, treatment with steroids, cyclophosphamide and plasma exchange is often required. Although the efficacy of anti-CD20 antibodies (i.e. Rituximab) has been demonstrated, its clinical usefulness in ICU patients remains uncertain [45, 46].

Acute interstitial nephritis

Acute interstitial nephritis (AIN) accounts for 10–25% of biopsy-proven AKI and should be considered one of the frequently missed AKI diagnoses in the ICU [28, 47, 48]. It is a histological diagnosis characterized by infiltration of inflammatory cells within the renal interstitium with sparing of the glomeruli and blood vessels. AIN can evolve to interstitial fibrosis within 7–10 days [28, 47, 48]. The most frequent causes of AIN are drugs, primary renal infections such as acute bacterial pyelonephritis, inflammatory disorders such as systemic lupus, Sjögren’s syndrome and sarcoidosis, or neoplastic diseases. Any drug can cause AIN, but antibiotics, non-steroidal anti-inflammatory drugs and proton-pump inhibitors are the most common medications associated with AIN. Drug-induced AIN can manifest at any stage after starting the medication and is not usually dose-related [49].

In critically ill patients with AIN, classical symptoms are usually absent. The presence of a rash, eosinophilia or eosinophiluria in a patient with AKI may be suggestive of AIN, and although insufficient to confirm diagnosis, should raise ICU physician awareness. Renal biopsy is the only definitive method of establishing the diagnosis. New diagnostic tests are urgently needed to improve AIN diagnosis, especially in critically ill patients [47].

Management of AIN consists of withdrawal of offending medications and treatment of an underlying inflammatory disease. Although the use of corticosteroids remains controversial, delayed initiation of steroids has been associated with incomplete renal recovery. Therefore, it is reasonable to use corticosteroids when there is no improvement within 3–7 days after discontinuation of the offending drug [47, 48].

AKI in haematological patients

Critically ill cancer patients seem at higher risk of developing AKI [50, 51]. Although most of the AKI episodes are multifactorial, specific causes may deserve urgent therapies. Thus, cancer patients may develop obstructive diseases and are more prone to nephron loss due to nephrectomy, nephrotoxic agents or previous AKI episodes [52].

Tumor lysis syndrome (TLS) may arise as result of rapid destruction of malignant cells leading to the release of intracellular ions, proteins, and metabolites into the extra-cellular space [53]. Although TLS usually occurs in patients with extensive, rapidly growing, chemosensitive malignancies, TLS has also been described in patients with low-grade malignancy treated with targeted therapies. The diagnostic criteria of TLS have been recently standardized (Table S1) [54]. Accordingly, a constellation of metabolic disturbances (hyperkalaemia, hyperphosphataemia, hypocalcaemia, and hyperuricaemia) defines laboratory TLS in high-risk patients, while clinical manifestations (cardiac, renal or neurological manifestations of TLS) in patients with laboratory TLS defines clinical TLS [54]. TLS may occur rapidly. Frequent (up to every 6 h) and routine search for metabolic disturbances, along with preventative measures, including hydration and treatment with urate oxidase, are mandatory [54].

In patients with myeloma, cast nephropathy remains the most common cause of renal injury, and reversibility may be achieved in up to 60% of these patients with immediate initiation of appropriate therapy [55, 56]. Cast nephropathy is caused by the precipitation of monoclonal light chains which bind Tamm Horsfall protein causing tubular obstruction. High amounts of Bence Jones proteinuria, extreme κ/λ light chain ratios or high serum levels of free light chains are suggestive of cast nephropathy [57, 58]. In some patients, renal biopsy may be avoided since cast nephropathy is the most common pathology, especially if urine albumin excretion is low (<25%) [59]. Early detection of cast nephropathy, combined with urgent chemotherapy, may improve the chances of renal recovery [55]. Bortezomib-based combinations are considered the most effective therapy for patients with AKI due to cast nephropathy, and should start as soon as possible [55]. The role of plasmapheresis is not established, and renal replacement therapy with high cut-off membranes (with cut-off around 50 kDa) might decrease free light chain concentrations [60]. Two prospective studies have recently shown conflicting results on the potential benefits of such an approach in patients with cast nephropathy [61, 62], and further analysis is required to identify which patients may get the maximum benefit from this demanding approach.

In patients with haemolytic anemia and thrombocytopenia, thrombotic microangiopathy (TMA) should be ruled out. Although rare, the combination of microangiopathic anaemia [non-immune haemolytic anaemia due to mechanical damage of red blood cells and resulting in high plasma lactate dehydrogenase (LDH) levels] with peripheral thrombocytopenia suggests this syndrome [63, 64]. The presence of damaged red blood cells (schistocytes) on peripheral blood smears confirms the microangiopathy. Differential diagnoses include auto-immune cytopenia and infectious causes of microangiopathic anaemia (malaria, babesiosis). Various organ dysfunctions may occur related to small vessel obstruction [63, 64]. Although overlap may occur, renal injury is more frequent in patients with haemolytic uremic syndrome (HUS), while neurological dysfunction is more common in patients with thrombotic thrombocytopenic purpura (TTP, also known as Moschowitz syndrome) [63, 64]. Three main syndromes should be searched for, including diarrhoea-associated HUS, usually associated with Shiga toxin-producing E. coli [63, 64], atypical HUS (aHUS) which is usually complement-mediated and associated with congenital abnormalities of the alternative complement pathway regulation [63,64,65] and TTP which is related to congenital or acquired ADAMTS13 deficiency [62, 63]. In patients suspected of either aHUS and TTP, looking for congenital complement abnormalities, measuring ADAMTS13 activity and looking for anti-ADAMTS13 antibodies should be the rule [62, 63].

AKI in the obstetric setting

In the developed world, the incidence of AKI during pregnancy, currently estimated around 1/20,000 pregnancies, has substantially decreased due to better prenatal care and reduction of illegal abortion. The diagnosis of obstetric AKI may be masked by the physiological increase of GFR (up to 40%). Obstruction by the gravid uterus may cause obstructive AKI, the diagnosis of which is complicated by physiological hydronephrosis and hydroureter. AKI during pregnancy is often multifactorial with both pregnancy-related and other potential causes. Timing may help in the differential diagnosis (Figure S1) [65]. In the early phase, sepsis (pyelonephritis, septic abortion) and hypovolemia (hyperemesis gravidarum, haemorrhage) are the most common causes, whereas in the later phase haemorrhage (abruptio placentae, postpartum haemorrhage), sepsis and typical pregnancy-related complications predominate. Preeclampsia, a poorly understood systemic disorder due to dysregulation of angiogenic factors, characterized by hypertension and proteinuria, or its variant Hemolysis with Elevated Liver tests and Low Platelets (HELLP), typically develop during the third trimester. The TMA include aHUS, that more commonly begins peri- and post-partum, and TTP, which develops in the second and third trimester. Pregnancy predisposes to TTP due to a decrease of ADAMTS-13 activity and increase of von Willebrand factor concentrations, but may also be a triggering factor for aHUS (due to dysregulation of the alternative complement pathway). Acute fatty liver of pregnancy (AFLP), caused by foetal and maternal congenital defects in beta-oxidation of fatty acids, is a rare but life-threatening cause of pregnancy-related AKI that mostly occurs close to term. Other potential causes of peripartum AKI include amniotic fluid embolism or cardiomyopathy.

Differentiating HUS/TTP from HELLP and AFLP is challenging but vital because of the therapeutic consequences [66,67,68]. An adequate differential diagnosis requires careful history including timing of onset, clinical examination, drug review, dipstick and microscopic urine examination, proteinuria, haemoglobin level, platelet counts, peripheral blood smears, LDH, coagulation tests, bilirubin, transaminases, blood glucose levels, ADAMTS-13 activity, complement factors and ultrasound (Table S2).

Acute kidney injury in low and middle income countries

In low and middle income countries (LMIC), early preventive and therapeutic measures are the key way to decrease morbidity, mortality and cost (Table 3). Early detection of AKI is impaired by limited diagnostic assets and poor understanding of the condition [69]. Such limited understanding—to a large extent determined by inadequate reporting and education—limits awareness and early recognition and delays the implementation of adequate management [70]. In LMIC, the common lack of access to specialized nephrology care requires that AKI be understood and recognized at all levels of the healthcare system [71]. A practical and easily accessible educational strategy focused on providers at the forefront of healthcare delivery (including primary physicians, nurses and community healthcare providers) is indispensable to achieve this goal [72].

Table 3 The contrasting characteristics of AKI around the world

In the community, in areas where infectious diseases such as severe malaria, leptospirosis or dengue are endemic and associated with high rates of AKI, a febrile patient should elicit concern for renal injury [73]. Similarly, in patients with severe volume depletion due to gastrointestinal loss, volume resuscitation is central to care and to prevent renal injury—preferably before the onset of persistent oliguria [72]. In some areas of the world, exposure to snake venom represents a frequent cause of AKI. Administration of herbs by traditional healers has been associated with nephrotoxicity and must be considered when confronted with AKI of unclear aetiology [74]. Increased availability and use of over-the-counter medications such as NSAIDs significantly contribute to a rising incidence of AKI.

The development of AKI as a maternal and neonatal complication is especially important in the LMIC environment [74], where failure to recognize renal injury frequently leads to significant consequences for mother and child. In this area, successful efforts to improve early recognition have clearly demonstrated benefit—especially by reducing some of the more dreaded consequences such as cortical necrosis [75].

Conversely, in the LMIC hospital and ICU settings, AKI recognition faces challenges akin to those seen in the developed world; among hospitalized patients, AKI related to exposure to nephrotoxic medications, antibiotics, intravascular administration of iodinated radiocontrast and surgical procedures is very common. Especially in the LMIC setting, additional testing and urinary microscopy are necessary to identify the underlying etiology. The performance of basic urine microscopy focusing on the presence of erythrocytes, leukocytes, eosinophils and casts in the sediment is invaluable to assess the initial presentation of the patient with AKI [76]. Training in microscopic urine examination and availability of basic examination equipment for such testing should be promoted as a key, low-resource test for detection of AKI in LMIC. Point of care testing (POCT) for creatinine measurements can be performed by non-laboratory-trained individuals, thus eliminating delays in testing and reporting of results [77]. However, it requires the implementation of a quality assurance program that ensures accurate and reliable results. Last, kidney biopsies in patients with AKI are more common in LMIC than in HIC and, thus, there is a greater appreciation of the relative incidence of multiple aetiologies and the value of a renal biopsy to guide management [69].

Conclusion and future directions

Identifying and diagnosing specific causes of AKI in critically ill patients remains challenging. A high degree of suspicion must be the rule and a systematic diagnostic work-up should be undertaken in every AKI presenting without an obvious predisposing factor or following an unusual course. Although data regarding specific causes of AKI remain limited, increased recognition of the nephrotoxic contribution to AKI [78], and of the infrequent finding of true “acute tubular necrosis” in critically ill patients, will result in improved AKI management [79,80,81].

It may be time to search more closely for specific causes, assess more carefully the prevalence of aetiologies even when typical symptoms are misleadingly absent—such as acute post-infectious glomerulonephritis or AIN—and, ultimately, reshape an old-fashioned and probably outdated AKI diagnostic paradigm.