Skip to main content
Premium Trial:

Request an Annual Quote

Early Chronic Kidney Disease Detection, Progression Targeted by RenalytixAI With Machine Learning Assay


NEW YORK (360Dx) – Startup RenalytixAI has set its sights on the kidney disease diagnostic space with its blood-based assay, which it believes will help improve clinical management of patients with type 2 diabetes and individuals of African-American descent with fast-progressing kidney disease.

The London-based firm recently began a clinical validation study for its chronic kidney disease (CKD) diagnostic, KidneyIntelX, with intentions of seeking de novo 510(k) approval from the US Food and Drug Administration later this year.

In order to diagnose patients with CKD and potential renal failure, clinicians have traditionally monitored estimated glomerular filtration rates or the amount of albumin in a patient's urinary tract. However, in earlier stages of CKD, patients with type 2 diabetes don't always manifest fluctuating levels of albumin or other biomarkers, leading clinicians to often misdiagnose and miss the condition in patients.

"That's why there's been a need to bring in other diagnostics and risk assessment tools to help identify these at-risk patients," RenalytixAI CEO James McCullough explained. "The thesis [behind KidneyIntelX] was to bring together machine learning and diagnostics and to integrate disparate datasets — which include key circulating biomarkers, along with EMR data — to create predictive and prognostic equations around kidney disease in large populations."

Founded in January 2018 as a subsidiary of EKF Diagnostic Holdings, RenalytixAI started listing under the London Stock Exchange this past November. The firm was established as a partnership between EKF Diagnostics and Mount Sinai Health System in order to validate and commercialize artificial intelligence-enabled clinical diagnostic tools for the early detection of kidney disease and transplant management. RenalytixAI currently has a CLIA laboratory in New York and an additional lab in Georgia.

McCullough explained that KidneyIntelX uses a patient's blood sample to identify three blood-based biomarkers that are well-known to be predictive of kidney disease; sTNFR1, sTNFR2, and KIM1. After a clinician extracts 50 microliters of a patient's blood, they send the sample to RenalytixAI's CLIA lab, where the firm's researchers run the samples on a multiplex electrochemiluminescence assay to identify the three biomarkers. The test then combines the biomarkers with data derived from a patient's EMR — including demographics, diagnostic and procedural codes, laboratory values, and medications — to generate a risk score of the patient's progressive kidney decline.

Earlier this month, RenalytixAI began a clinical validation study for the KidneyIntelX assay in collaboration with investigators from Johns Hopkins Medicine, Emory University, Mount Sinai, Northwestern University, Harvard University, and Brigham and Women's Hospital.

According to McCullough, the researchers will collect about 5,000 blood samples and EMRs from the multicenter collaboration over the next six to seven months. RenalytixAI Cofounder Steve Coca noted that the study's sample size will include patients with type 2 diabetes or are of African descent, whom he highlighted have a higher risk for CDK progression.    

"In the past, [researchers] discovered that this gene called APL1, which is a mutation that occurs largely in people from the West African region, protects them against Human African trypanosomiasis, or sleeping sickness," Coca explained. "[They've] found over the years that it also acts as a strong genetic risk for CKD in this population."              

McCullough explained that RenalytixAI will use the data generated from the clinical validation study as part of its goal to seek regulatory approval from the US Food and Drug Administration for the approach. He noted that the firm will also establish "the finer points" of the predictive value to distinguish patients with different levels of kidney deterioration.

Coca also emphasized that the study will help RenalytixAI "lock in" the machine learning's algorithm and allow the firm to develop a better version of the assay for clinical utility studies.

"If you're able to stratify those patients in the fast progression [group] now, you can apply clinical utility and impact against those patients to make a difference," McCullough said. Clinicians can then "slow the progression of the disease and ultimately start to change the outcome."

Prior to possibly getting FDA approval, RenalytixAI expects to commercially launch the assay as a laboratory-developed test out of its CLIA-approved lab after it concludes the clinical validation study in Q2 2019. The estimated price will be below $1,000 per run. 

According to Coca, doctors and care managers will be able to review the private data through a customized portal on the company's website. Additionally, patients "will be able to track this score on their own, and see how it changes over time, and what they can do to modify the risk score," Coca explained.

According to McCullough, the overall process for the KidneyIntelX to develop a CKD risk score requires between five and seven days. Because the firm envisions the test to be used in the outpatient setting, Coca said that it aims to offer a turnaround time that is "well before a repeat doctor's visit."  While the firm declined to comment on the assay's clinical sensitivity and specificity, McCullough believes that the assay's accuracy will be "an improvement over the standard of care."

In addition, McCullough said RenalytixAI is currently filing a patent in the USfor the machine learning platform and related material.

As it pushes to commercialize the KidneyIntelX assay, RenalytixAI will encounter multiple other groups within the CKD diagnostic space developing their own methods to detect and track kidney failure in patients.

Siemens Healthineers announced in October that it is partnering with Israeli startup to market a smartphone tool for CKD monitoring in a patient's home. Siemens will integrate its urinalysis reagents into the Israeli's smartphone-based urinalysis system to allow increased albumin-to-creatinine ratio testing and response.

In addition, Australian proteomics firm Proteomics International offers its PromarkerD immunoassay test, which uses mass spectrometry to measure blood-based proteins to predict diabetes patients' risk of developing diabetic kidney disease.  The firm is collaborating with Janssen Research & Development for predicting decline in kidney function and drug response.

A group of Columbia University Medical Center researchers are also developing an exosome sequencing method to detect a large subset of CKD cases. In a study published in December in the New England Journal of Medicine, the team found that its genetic-based method identified kidney disease-related pathogenic variants in exomes for about 9 percent of individuals with CKD.

McCullough argued that RenalytixAI's focus on a wet-lab component and integrating the data with EMRs will provide a stronger and earlier predictive value than diagnostic assays developed by other groups. In addition, Coca highlighted that the platform will monitor specific proteins that respond differently over time in response to therapy rather than static genetic mutations. He believes that patients will therefore need to be retested as they start the appropriate medications or engage in other activities that impact their health.

"Unlike a genetic test, which is just fixed in one's  germline, the biomarkers that are input into the [KidneyIntelX assays] are providing a readout of all these clinical and subclinical dysfunctions that we may not have been aware about," Coca said. "Moreover, they can change over time and in response to therapy."

McCullough emphasized that the user can pull a patient's EMR on a repeated basis to update their health status, which can alter the CKD risk profile.

Coca explained that the longitudinal component will help inform physicians and motivate patients to improve their lifestyles, since they can potentially change their risk score over time. He noted that patients could also be relieved that they have some component of a lower eGFR, but that they're at low risk based upon other health factors. 

"We can uncover these people at high risk, but there is also an opportunity at the other end of the spectrum to de-risk patients that might get inappropriately anxious for mild forms of kidney disease that aren't likely to progress over time," Coca noted.

According to McCullough, RenalytixAI has raised about $30 million in financing since it was founded last year, with most of the funding stemming from its listing on the London Stock Exchange.                  

McCullough highlighted that RenalytixAI also plans to partner with drug developers to use its machine-learning diagnostics and growing data pool to efficiently enroll fast-progressing kidney disease patients in clinical trials. The firm also plans to increase the number of biomarkers on its platform as it expands into precision medicine development and therapy in the kidney space.

"As we expand into our validation and clinical utility studies, we may potentially add other orthogonal or additive biomarkers, whether they're in the blood, circulating DNA, or in the urine, but we need time to validate those approaches empirically," Coca said. "We intend to increase the level of information as the patient population grows and as new biomarkers are validated."