Home pregnancy tests are easy to use and they give accurate and fast results. If the use of a simple biomarker test in urine could be applied to chronic kidney disease, the test would have the potential to improve the lives of millions of people. 10% of the population suffers from chronic kidney disease, or CKD, and currently the diagnostic procedures are not uniformly applicable to all patients. There is no cure either.

“We hope that, someday, it will be possible to use alterations in the patterns of metabolites in patient urine samples to predict the onset and progression of CKD,” says Robert Fenton, who was heading the KidDO project. Together with an interdisciplinary team of researchers from Aarhus University and AstraZeneca, the KidDO project focused on exploring the changes in metabolites that occur during CKD. “Metabolites are the products - or intermediate products – of all the chemical reactions that take place inside a cell. In the KidDO project, we integrated knowledge from metabolomics and proteomics across rodent animal models and human CKD biopsies. We aimed to identify specific metabolites that may be useful as biomarkers for CKD,” he elaborates.

CKD is a complex disease, where the kidney tissue is scarred, a process that is aggravated by other conditions, primarily hypertension and type 2 diabetes. As these are both on the rise, the number of CKD patients is increasing: “CKD causes more deaths than breast cancer or prostate cancer, and more than 1 in 7 adults suffer from it. The disease progresses slowly over many months or years, and most patients simply don’t know that they have CKD before it’s too late,” Robert Fenton explains and continues: “When significant symptoms appear, the kidney function is irreversibly impaired. Therefore, it’s important that we identify biomarkers that can help clinicians diagnose patients at the early stages of CKD before the disease has progressed too far.”

Even when patients are diagnosed with CKD, there is no effective cure. Most often clinicians must settle for treatment of the underlying diseases, but there is nothing they can do to reduce the scarring that has already occurred. At the late stages of the disease, when the kidney function is completely lost, the patients must be on dialysis, and some undergo a kidney transplantation.

“Biomarkers should not only help diagnose patients. It is equally important that they can predict progression of the disease,” says Robert Fenton. He knows how important evaluation is in clinical trials: To choose the best drug candidate for future treatment of the patient, it must be possible for researchers to accurately and objectively assess how potential drugs affect the disease. “When the level or presence of specific biomarkers can tell researchers what the kidney tissue looks like, it becomes much easier to evaluate how a given drug candidate can help patients at specific stages of the disease,” Robert Fenton says.

As mentioned above, the KidDO team focused on metabolites as biomarkers for CKD. But the metabolite biomarkers themselves might also become relevant when it comes to CKD therapies. “Simply put: If there is too much of a metabolite in CKD kidneys compared to healthy kidneys, we might be able to come up with strategies to lower the level of that specific metabolite. If there is too little of a metabolite, the patients might be able to increase the levels though the diet,” says Robert Fenton. He and the rest of the KidDO team employed different strategies to identify relevant metabolites involved in the onset and progression of CKD.

Metabolites are intermediate or end products of the chemical reactions that take place within the cell. But the levels of these metabolites, especially intermediate products and the speed at which they are produced are not necessarily the same between humans and mouse. In many cases, this can pose a problem in a research setting. “We need complex models for CKD when we look for biomarkers and new drug targets. But whole-body models are always animals. And if the animal doesn’t completely mimic the human CKD condition, we might not be able to use the results after all. Therefore, well-characterized animal models are essential,” explains Robert Fenton. Acknowledging this, the KidDO team compared five different animal models with biopsies of human CKD kidneys in different ways.

Since the KidDO project started in 2020, the team characterized and compared the five different animal models. For instance, they linked the clinical pathology – how the CKD kidneys look – to the clinical chemistries of urine and plasma. They found that the scarring of kidneys was not always similar between models. “Some models developed a striped pattern of fibrosis, like spokes in a wheel, whereas other models displayed a radial pattern more like a small circle spreading from the center of the kidney,” says Robert Fenton.

Not all members of the KidDO team were involved in the characterization of the animal models. The team had broad spanning competencies, which made it obvious to divide their work into different work packages. In one work package, the team aimed to identify which, and how many, metabolites were present in CKD in both human and animal kidney samples. As part of this work, the team developed a method for isolating the healthy areas of the kidney from the scarred areas. “In this way, we were able to look at the two types of tissue separately – but it also provided an opportunity to compare the tissues and see which proteins and potential metabolites were changed due to the disease,” says Robert Fenton.

The purpose of another work package was to compare the profile of metabolic enzymes between animal models and human samples. “Initial results showed that more than 400 proteins were changed in the same direction across 4 of the 5 animal models,” states Robert Fenton and adds that the team has started to develop new approaches for correlating data from single cells sequencing with quantification of proteins in the cells: “When we look at the collective proteome for all cells, we can now model the data back to the single cells to find out where the changes in the overall proteome are taking place.”

Other competencies in the KidDO team included computer science and machine learning. Originally, the aim was to combine the proteomic and metabolic profiles using computational tools, but the task proved more difficult than expected due to a small number of statistically significant identified metabolites. Data analysis for further machine learning processes in the KidDO project is still ongoing, and Robert Fenton explains that the team has already obtained valuable initial results: “Although we are still working on this particular part of the project, the initial analysis identified a number of metabolites and genes that were universally altered across the different animal models.” Similar to the ongoing machine learning processes, the team is still testing approaches for efficient delivery of metabolites to kidney cells.

“A lot of work is still ongoing,” says Robert Fenton and continues: “We hope that the open approach will be valuable for researchers around the world as they can use our data for both model selection and they can also ‘cherry pick’ metabolites and proteins to examine in their own targeted research.”

Read more about the KidDO project here.