AKI: Desperate need for a diagnostic
AKI is a major risk to human health and wellbeing given the range of different causative factors and the potential for it to progress to long-term renal dysfunction. It is defined as the rapid deterioration in kidney function over hours or days. However, the damage associated with the condition is frequently reversible if the injury is identified quickly and patients are treated. There is therefore significant incentive to develop diagnostics that can quickly identify AKI in at-risk individuals.
There are a wide range of causes for AKI. The single largest contributing comorbidity to these rates of AKI is sepsis, which is responsible for 26% to 50% of all cases of AKI.1 The aetiology of sepsis-induced AKI is complex and includes both direct damage to the kidneys and a reduction in blood flow (hypoperfusion). Additionally, conditions that lead to hypoperfusion (low blood flow) can lead to AKI: as many as 30% of patients undergoing cardiac surgery have complications associated with AKI.2 AKI is responsible for a fivefold increase in mortality associated with these procedures. Other major surgeries also carry a risk of AKI, albeit at lower rates. Other conditions that can cause a severe drop in blood pressure or fluid loss and thus hypoperfusion can cause AKI, such as bleeding, diarrhoea, overdose of NSAIDs, allergic reactions and shock associated with trauma, although this list is by no means exhaustive.
Given the range of conditions that can lead to AKI, the condition is relatively common. Based on sCr measurements (more on this test below), the rate of AKI in the US is over 0.5% per year in the general population.3 Approximately 2% of hospital inpatients and 40% of those in intensive care have AKI.4 It is difficult to separate the prognosis of AKI from the underlying disorders, but AKI significantly increases the risk of death in a stage specific manner: odds ratio of 2.2 for stage 1, 6.1 for stage 2 and 8.6 for stage 3.5 Among patients with AKI severe enough to require renal replacement, mortality has been observed as high as 60%.6 Moreover, there is increasing evidence that even after resolution of AKI, the event is correlated with increased risk of developing chronic kidney disease.7
AKI is classically diagnosed and staged based on the concentration of creatinine in serum and urine output. Both measurements are proxies for the glomerular filtration rate (GFR), or the rate at which the kidney can process liquid. Creatinine is the metabolic product of creatine degradation in muscle that is typically filtered from the blood by the kidney. Given that its production is relatively constant, an increase in serum levels can be indicative of renal dysfunction. The Kidney Disease International Global Organization (KDIGO) provides the criteria for staging AKI (Exhibit 3).
Exhibit 3: Staging of AKI based on KDIGO criteria
Stage |
sCr |
Urine output |
1 |
1.5–1.9× baseline or ≥0.3 mg/dl above baseline |
<0.5 ml/kg/hr for 6–12 hr |
2 |
2.0–2.9× baseline |
<0.5 ml/kg/hr for >12 hr |
3 |
≥3.0× baseline, ≥4.0 mg/dl, or initiation of renal-replacement therapy |
<0.3 ml/kg/hr for ≥24 hr or anuria for ≥12 hr |
Despite its widespread use, there are significant limitations in the use of sCr as a tool to diagnose AKI. The primary limitation is that changes in GFR are indicative of kidney damage, therefore some injury and loss of function has already occurred by the time a change is measurable. This is exacerbated because creatinine must build up in the serum and it can take significant time for changes in GFR to manifest as measurable changes in creatinine. The kinetics of sCr are very complex and depend on a wide range of variables (which gives more credence to it being a poor biomarker), but modelling predicts that a healthy person would need 34 hours for their sCr levels to reach steady state after losing 50% renal function, and twice as long or more for patients with prior renal impairment. Renal impairment of 27% or less is undetectable even as stage 1 AKI within 48 hours in an otherwise healthy person. Because of this, sCr measurements are typically taken over several days to provide adequate time to detect changes from baseline. This substantially increases the burden on providers and increases the probability that marginal cases of AKI will go undetected under a reasonable timeframe.
Baseline rates of creatinine can differ significantly between individuals and even within the same individual due to a range of factors. Therefore, to be used as a biomarker, multiple measurements are required to establish a baseline and changes from this baseline. Patients of unknown status may already have elevated creatinine when they are initially tested, confounding the detection of issues. Moreover, the clearance of other substances such as medication can significantly affect the rate of creatinine clearance.
A problem in evaluating the utility of sCr as a diagnostic tool is that historically AKI has been defined in terms of sCr and little corroborative evidence has been available. One study, however, used biopsy data from deceased kidney donors to retrospectively evaluate the performance of sCr as a diagnostic. Biopsies from these patients were examined for evidence of acute tubular injury (ATI) and compared to sCr measurements. It should be noted that ATI is a subtype of AKI and is the leading cause of AKI in a hospital setting (approximately 50%). The results from this study suggest sCr is a very poor indicator of ATI (Exhibit 4). The area under the curve (AUC) for sCr to identify any grade of ATI was 0.52. This value increased marginally to 0.58 when the test was evaluated for the detection of severe ATI. AUC is a measure of the strength of a diagnostic irrespective of the particular cut-off value used for diagnosis, where 1.00 is a perfect test and 0.50 indicates no diagnostic value. So, in this case, sCr performed poorly; in fact we are unaware of any other diagnostic tool in routine use that shows such a low AUC.
Exhibit 4: Performance of sCr versus histology of ATI
|
Severe ATI versus no ATI |
Any ATI versus no ATI |
Sample size |
483 |
581 |
AUC |
0.58 |
0.52 |
sCr criteria |
Sensitivity |
Specificity |
Sensitivity |
Specificity |
Stage I AKI or higher |
51% |
61% |
42% |
61% |
Stage II AKI or higher |
26% |
84% |
20% |
84% |
As an alternative to sCr, urine output can be used. However, monitoring of urine is unwieldy in clinical practice and is generally limited to patients with a catheter. Moreover, this measurement is rendered ineffective when patients are taking diuretic medications. There are not as much clinical data to support urine output, because retrospective data are generally unavailable and clinical studies have had mixed results.
NGAL: A better alternative
Due to the limitations of sCr, there has been an effort by scientists to identify other biomarkers with improved performance. The most concerted effort has been focused on the investigation of NGAL. It was first identified as a marker for AKI by researchers in 2003 and has subsequently been the subject of multiple studies. Perhaps the clearest benefit of NGAL over sCr is evident in the time course of its elevation following kidney injury (Exhibit 5). NGAL is elevated within hours of the damage that results in AKI, as opposed to sCr, which requires a prolonged period of impaired GFR. Patients undergoing cardiopulmonary bypass (a procedure known to induce AKI in some cases) had NGAL peak two hours after surgery, whereas diagnosis with sCr took 2-3 days. Moreover, there is increasing evidence that patients that are identified by NGAL carry an increased risk for adverse events such as need for replacement therapy and death, even when they are sCr negative. NGAL is elevated before major loss of function, which should enable earlier intervention to halt progressive deterioration.
Exhibit 5: Time course of biomarker elevation in AKI
|
|
Source: BioPorto Diagnostics
|
NGAL is a member of the lipocalin family, a class of proteins that bind hydrophobic molecules. It binds specifically to siderophores, proteins that bind to iron and aid in its metabolism. A range of different tissues, including but not limited to the kidney, secrete NGAL in response to cellular damage. Due to this, an increase in serum concentrations of the protein is associated with a range of indications including infection, inflammatory disorders, cancer and obesity. NGAL can be isolated from either the urine or plasma, with differing results. Urine NGAL provides a more direct readout of protein released in the kidney, but is affected by urine production and complicated by common conditions such as urinary tract infection. Plasma NGAL provides a less variable baseline but can be complicated by injury or inflammation in other tissues. Whether urine or plasma NGAL is a better indicator is an unsettled question, although BioPorto is advancing a plasma-based test, initially citing better reproducibility.
Two products have previously been developed for use as a NGAL diagnostic test in the clinical setting, although neither was cleared in the US for diagnosis of AKI. Abbott developed a urine NGAL-based test for use with its Architect clinical chemistry platform. It was submitted to the FDA in 2010 but did not receive clearance. Alere also developed a point-of-care plasma NGAL based test for use with its Triage MeterPro platform. However, it was not submitted for FDA clearance. Alere was subsequently purchased by Abbott in 2017 and as the Triage NGAL diagnostic test was one of its many products, it was likely not a motivating factor. BioPorto launched its first NGAL-based bioassay in 2006. This was the first commercially available NGAL ELISA kit available worldwide, although limited to research purposes. In 2010 the company subsequently launched a new version of the assay prepared for use in a high-throughput clinical chemistry analyser, branded the NGAL Test.
Previous noteworthy NGAL studies
The NGAL biomarker has been investigated in a large number of clinical studies, in urine and in plasma and by BioPorto and others, and significant variability between results has been observed. A recent meta-analysis was performed that examined the capacity of biomarkers to predict the initiation of renal replacement therapy, an intervention to limit the damage from AKI. Urine NGAL biomarker was evaluated in 12 studies and showed a pooled AUC of 0.720 and plasma NGAL was evaluated in 16 studies with a pooled AUC of 0.787. AUCs in the meta-analysis for NGAL ranged from 0.260 to 0.884. A consistent factor that has been cited as a source of this variability has been the differing response of patients with sepsis. As mentioned above, sepsis is the most common cause of AKI in intensive care patients, but the systemic inflammation associated with the condition results in the release of NGAL from neutrophils. Although the biomarker retains utility in sepsis patients, the AUC for predicting renal replacement therapy can drop significantly, to 0.700 in one targeted study. NGAL may be a better biomarker in patient populations outside the ICU , where there are lower rates of confounding factors such as sepsis. For instance, one study using BioPorto’s NGAL antibodies (although not performed by the company) that examined 635 patients presenting in an emergency department (instead of an ICU) showed dramatically better statistics: 90.0% sensitivity, 99.5% specificity and an AUC of 0.948.
Another issue that has limited the interpretation of NGAL studies is that frequently the readout used to evaluate the test is the presence of AKI, as evaluated under the standard diagnostic criteria, namely, sCr. This is a problem intrinsic to this field, and has been highlighted in research. Even a perfect test (100% sensitivity and specificity) will have substantially lower apparent statistics when measured against an imperfect gold standard. The fact that NGAL can identify patients at increased risk of major intervention or death that are sCr negative is also supportive of this fact.11
A significant factor that can be difficult to capture in these statistical studies is the improvement in care NGAL can provide. In particular, the ability to identify patients sooner and before significant loss of function can translate into improved outcomes and the associated reduction in costs. One study modelled these factors and estimated costs savings in the range of $408–522 per patient admitted to an emergency department. The models in the study were using real world outcomes data from two emergency departments in New York tested for urine NGAL and sCr.
Earlier BioPorto clinical data
BioPorto has released relatively little data detailing the performance of the NGAL Test from its earlier clinical studies. The most information is from the company’s first US registrational study, which was performed in 2014 and 2015 across four clinical centres. The purpose of the trial was to identify the correct parameters, such as NGAL thresholds for the clinical identification of AKI using both plasma and urine. The study compared blood preserved with two anticoagulants (EDTA and heparin) so see if that had an impact as well. It enrolled 245 patients from ICUs and AKI was determined using the KDIGO guidelines by a panel of physician adjudicators. The data reported by the investigators segregated the patients into two populations: those with stage 2 or 3 AKI, and those with stage 1 AKI or no AKI. Samples were taken from the patients daily.
Exhibit 6: The NGAL Test statistics for stage 2/3 AKI
Fluid |
Sensitivity |
Specificity |
AUC |
EDTA plasma |
78.8% |
73.0% |
0.76 |
Heparin plasma |
72.7% |
73.8% |
0.77 |
Urine |
69.7% |
76.8% |
0.79 |
Source: Tecson et al. 2017
The statistics from the study were positive. AUC measurements were 0.76 and above, and sensitivity and specificity measurements were approximately 70% or higher. These results are largely in line or better than those previously presented in the literature on NGAL, although it should be noted they represent the identification of just stages 2 and 3 of AKI. When the investigators then focused on patients with persistent (two days or more) stage 2/3 AKI, the results were further improved to a maximum AUC of 0.85 with the use of EDTA plasma.
The FDA rejected the company’s application package including these data. There are limited details on the reason behind the FDA’s decision, although BioPorto has stated the rejection was ‘primarily because the dataset for mild cases of AKI did not support approval’. When explaining the decision to exclude patients with stage 1 AKI from the primary end point, the investigators cited the observation from prior studies that many patients classified in this category are subject to transient sCr elevations without AKI associated with tissue damage. In other words, the stated reason for this exclusion was aforementioned variability in the sCr gold standard.
BioPorto currently has two development programmes for the NGAL Test: adult AKI using blood NGAL and paediatric AKI using urine NGAL. The product for adults represents the biggest market opportunity for the product, but BioPorto believes the paediatric programme may be easier from a regulatory perspective and provide insight for the adult application process.
The company has had additional clinical and regulatory setbacks on its pathway towards clearance of the NGAL Test. After the initial rejection by the FDA above, in 2018 BioPorto conducted an additional study and submitted a second 510(k) application to the FDA for clearance of the NGAL Test in adults based on this data from over 500 ICU patients across 17 clinical sites in the US where it only enrolled hypotensive patients within the first 24 hours of their admission to the ICU. Four blood samples were collected from each patient and AKI was subsequently graded based on KDIGO guidelines by an independent adjudication panel. In October 2018, the FDA requested additional information to continue the application process citing further data was needed to support the claim that the test could rule-out AKI. The company initially estimated it can complete the dataset with an additional 150–200 patients.
Concurrent with this, however, the company’s application for the paediatric NGAL Test was rejected in late 2019. The data in the paediatric 510(k) application were based on samples collected in a third-party study of AKI in children (the AWARE study) and tested urine samples for NGAL retrospective to the study. The AWARE dataset had 4653 paediatric patients, of which 1261 had AKI (543 severe AKI). A subset of these patients were tested, finding a sensitivity of 65.0% and a specificity of 81.8%. We believe these performance metrics are sufficient to support market authorisation. However, the FDA was concerned that the enrolment criteria for the AWARE study may have introduced a bias in the data, as only patients expected to be in the ICU for 48 hours or more were tested, which was evaluated by just the general impression of the investigators. The company determined it would need to collect new data to support an application given the FDA feedback. Based on our communication with the company, management believes that 300 paediatric ICU patients should suffice.
The rejections of the company’s marketing applications have caused unfortunate delays, but do not lessen our opinion of the ultimate approvability of the product because the FDA decisions to date have mostly centred on the statistical strength of the application package as opposed to providing doubt on the test’s utility. For this the upcoming submission for the paediatric NGAL Test, BioPorto is planning on preparing a De Novo application (as opposed to 510(k)), which has more stringent safety and efficacy guidelines, but we expect the completion of the clinical studies to answer any questions of safety or efficacy. BioPorto is in the final stages of gathering the clinical data and the current timeline for submission of the application is in Q121. A response from the agency is expected after the 150 day statutory review period.
The plan is for BioPorto to seek clearance for the test initially with the paediatric NGAL Test and to subsequently follow this with an application for the adult test. This will allow the company to tailor the application based on the feedback provided by the FDA during the paediatric review process. Additionally, BioPorto will be able to use the paediatric test as the predicate device for the 510(k) application. In Europe the product has received a CE mark, making it commercially available, but adoption in that region is largely limited to research use as in the US currently. Adoption of diagnostics in Europe is driven more by guidelines and other decisions by organising bodies than simply marketing approval. More widespread adoption and more routine diagnostic use will require the company to publish the upcoming results from its registrational studies.
Sales, licensing and intellectual property
The NGAL Test is already commercially available for research purposes and sales were DKK11.6m in 2019 and DKK9.8m for the first nine months of 2020, an improvement of 58% over the same period in 2019. BioPorto has global distribution agreements with Roche and Siemens, as well as a small direct sales channel that provides the test. We expect these distributors to be the primary sales channels following FDA market clearance, although we expect BioPorto to hire a small sales team dedicated to the promotion of the NGAL Test. The primary commercial hurdle will be altering the longstanding clinical practice on the use of sCr for AKI, which will take physician outreach and education, which we expect the internal sales team to perform. Other aspects of the launch should be smooth considering the product seamlessly integrates into the existing workflow and is billable under existing DRG codes.
BioPorto entered into an arrangement with Abbott in 2014 to cross license their respective intellectual property after Abbott’s NGAL diagnostic test was rejected. BioPorto has also in-licensed additional patents from Columbia University regarding the NGAL technology. The company’s proprietary and in-licensed patent families cover a range of aspects of the NGAL technology and its applications including the use of urine and plasma, the use of serial sampling and the diagnostic threshold, among others. BioPorto’s NGAL Test is not without some degree of intellectual property risk: patents begin to expire in 2025, although we expect the portfolio as it stands to provide a decent commercial runway to approximately 2028. To our knowledge, there are no other NGAL-based tests under active development as a commercial use AKI diagnostic.
Other testing methodologies
In addition to NGAL, a number of other biomarkers have been investigated as alternatives to sCr (Exhibit 7). The most prominent research (other than into NGAL) has been into kidney injury molecule 1 (KIM-1) and IL-18. KIM-1 is a protein that is upregulated in the kidney following reperfusion injury in renal tubules and is a direct measure of injury (although not exclusively kidney injury). IL-18 in contrast is a proinflammatory cytokine secreted by macrophages that are released in response to various inflammatory conditions, including AKI. The performance of these biomarkers has generally underperformed NGAL in the clinic (AUC of 0.68 and 0.63 respectively compared to 0.74 for NGAL in one study). However, the performance of these markers tends to improve with more severe AKI and one study demonstrated superior results for the use in combination: AUC of 0.93 for KIM-1 + IL-18 compared to 0.89 for NGAL + sCr for the ability to predict stage 3 AKI or death.
Exhibit 7: Selection of alternative AKI biomarkers
Marker |
Name |
Notes |
NGAL |
Neutrophil gelatinase-associated lipocalin |
Component of the innate immune system, secreted by neutrophils and the kidney and other tissues following injury. |
L-FABP |
Liver-type fatty acid-binding protein |
Long chain fatty acid transporter, elevated in response to tissue damage of multiple types. |
IL-18 |
Interleukin 18 |
Proinflammatory cytokine produce from macrophages, associated with ischemic injury. |
KIM-1 |
Kidney injury molecule 1 |
Protein specific to the kidney, upregulated following ischemia. |
Cys C |
Cystatin C |
Protease inhibitor, ubiquitously expressed, clearance rate associated with GFR like sCr. |
TIMP-2 + IGFBP-7 |
NephroCheck |
Only branded proprietary AKI test available for the identification if imminent stage 2/3 AKI. |
The only commercially available alternative to sCr is NephroCheck, 510(k) cleared and marketed by Astute Medical. Astute Medical was a private healthcare company that was acquired by BioMérieux in April 2018 for approximately $90m. NephroCheck and associated systems are its sole products, but there is limited visibility on sales.
The test combines readouts of two biomarkers, TIMP-2 and IGFBP-7, in the company’s proprietary linear flow devices to be used in a dedicated testing platform (the Astute140 device). These proteins were discovered in a longitudinal study of 300 biomarkers in 2013 and thus have a shorter history of study compared to other biomarkers. The test is intended for use in patients following major cardiac or pulmonary events for AKI monitoring. NephroCheck was evaluated in two clinical studies. The first clinical study enrolled 408 patients and found a sensitivity of 90–93% and a specificity of 45–49% (with values varying based on the laboratory used). However, in the second clinical study of 126 patients, sensitivity dropped significantly to 76% with a specificity of 51%. This implies a negative predictive value of only 88% in the second study, meaning 12% of patients that were ruled as AKI free in the study really had kidney injury. However, despite these limitations, given the low bar set by sCr the test has been shown to improve outcomes following cardiac surgery,, and the test (or more accurately TIMP-2 and IGFBP-7 testing) was recently included in the consensus guidelines from the Enhanced Recovery after Surgery (ERAS) Cardiac Surgery group presented to the American Association for Thoracic Surgery (moderate level of recommendation).
Aside from BioPorto’s biologic products, it has developed a generalisable lateral flow device called the gRAD (Exhibit 8). The device consists of a paper test strip similar to that used for at-home diagnostics such as urine glucose strips, ovulation tests, etc. However, unlike these products, the gRAD does not come preloaded with detection antibodies. Instead, it can be arbitrarily used in any detection system consisting of a biotinylated antibody and a gold conjugated antibody. The strip contains a biotin binding region (presumably with some type of avidin, a protein that binds biotin tightly and used extensively in this context) that captures the biotinylated antibody and immobilises the analyte, which is subsequently detected by the gold conjugated antibody. The product can therefore be used to assay biomarkers that otherwise lack a point-of-care test, without the need for high-cost capital equipment such as clinical chemistry instruments. The product has potential in segments of healthcare where capital equipment is unavailable, such as in field work, at the bedside and in the office of general practitioners. The gRAD was launched in 2015, but only constitutes a small portion of the company’s revenues to date.
gRAD development programmes: From COVID-19 to snake bites
BioPorto has several early-stage programmes using the gRAD as a platform with its proprietary antibodies. The first and most obvious programme is to use this platform with the company’s NGAL antibodies to provide a dipstick for assessment of NGAL. In December 2020, the company received a CE mark for the so-called NGALds (ds for dipstick). This product could provide a rapid assessment of kidney health, for instance in the triage environment, where no such testing capacity exists. Other potential uses include monitoring post-surgery, allowing a patient to leave the hospital and test for kidney issues over the following days at home.
Another programme in development intending to use the gRAD as a platform is BioPorto’s COVID-19 programme. The company partnered with the University of Southern Denmark to develop antibodies to be used on the platform to detect the SARS-CoV-2 virus. In December 2020, BioPorto announced it had completed development and was ready to start testing patients. The FDA previously guided the company that it could move forward with the test with 150 patient samples, of which at least 30 would need to be positive. The company noted that non-invasive samples will be taken, presumably nasal swabs. Patients are being enrolled at the University of California Davis. BioPorto believes it can complete this study in early 2021 and it expects to file a EUA and CE mark application shortly thereafter if the results are positive. This product would fill a niche in the COVID-19 testing landscape. Tremendous improvements have been made in the throughput of COVID-19 testing, but the majority of testing in the US remains nucleotide based (eg, PCR). This still requires significant infrastructure and resources to perform. A dipstick-based antigen test, like the one in development at BioPorto, could be used to prescreen patients and provide an immediate feedback on COVID-19 status to be followed up with more robust nucleotide based testing (if positive), which would dramatically improve resource utilisation for COVID-19 testing. Moreover, there are many circumstances in which individuals undergo routine screening, such as healthcare workers, and a robust antigen test could improve screening capacity in these circumstances by providing immediate results without the need for infrastructure. That said, the utility of the test in any circumstance will be entirely dependent on its performance (eg sensitivity and specificity) and will be demonstrated in the clinical study. There are already a number of antigen-based tests that have EUAs and are commercially available, many with sensitivities over 95% and specificities at 100%. Therefore, we believe the bar is high for the performance of the BioPorto test.
Finally, BioPorto has two other announced development programmes using the gRAD, about which we know relatively little. First, the company has announced it is developing a gRAD assay for sepsis, using thrombomodulin as the biomarker. Thrombomodulin is a protein that is released as a result of damage to the lining of blood vessels, which is common with sepsis. Finally, BioPorto has partnered with VenomAid Diagnostics to develop a series of test to quickly identify the venom from snake bites for immediate treatment. A dipstick test is attractive in this case because snakebites are a more prevalent problem in regions that are underserved by medical care. Relatively few details has been released about either the sepsis programme or the snake bite programme.