When we think about kidney transplants, we imagine the dramatic moment of surgery, a list of immunosuppressive drugs, and the first blood tests after the procedure. We rarely think about what happens many months later, when everything seems to be fine, yet a damaging rejection process begins in the organ.

Under the microscope, the pathologist sees damage in various parts of the transplanted kidney and, on this basis, assesses the type of rejection we are dealing with. One type of rejection is microvascular inflammation. Until a few years ago, it was underestimated and overlooked. However, today we know that this pathological process, as part of the antibody-mediated rejection group, may be one of the causes of damage to the transplanted kidney. However, it is not easy to diagnose it non-invasively, i.e., without performing an organ biopsy.

Given the latest criteria for diagnosing antibody-mediated rejection and microvascular inflammation, there is also a large gap between the number of patients in whom this type of rejection can be diagnosed by biopsy and the positive results of classic, well-established immunological tests, such as antibodies against Human Leukocyte Antigens (HLA). Simply put, without performing an invasive biopsy, even if the blood test results are expected, it is currently impossible to rule out with certainty that a damaging process is taking place in the transplanted kidney.

This problem was examined by a team of researchers from Wroclaw, led by Prof. Mirosław Banasik, a specialist in nephrology, transplantology, and internal medicine at Wroclaw Medical University. The work by MD. Jakub Mizera, a doctoral student at the WMU Doctoral School, published in the journal Frontiers in Immunology, combines classical transplantology, pathomorphology, biochemistry, and AI data analytics.

Their question was: Are there risk factors we do not routinely check in patients that may contribute to transplant damage? Can artificial intelligence help to analyze them better?

At the University Clinical Hospital in Wroclaw, the transplant clinic cares for over 1,000 patients with transplanted kidneys. For some of them, the transplant is not an episode, but almost their entire adult life.

"Our goal is to prolong the survival of the transplanted kidney. Currently, at the transplant clinic, where we care for about 1,200 patients after kidney transplantation, there are over 120 people who have been functioning with a transplanted organ for over 20 years. We also have patients who have been living with a transplanted organ for over 30 years", emphasizes Prof. Mirosław Banasik.

Behind these numbers lies an ambitious plan not only to maintain good organ survival statistics but also to improve them. And, equally important, to understand why some transplants, despite treatment, are rejected and lost prematurely.

Diagnostic criteria are changing

Under a microscope, a transplanted kidney is very complex in structure. To someone unfamiliar with the histopathology of this organ, it may look like a city seen from a bird's-eye view. There are wide arteries—large vessels; smaller streets—capillaries; dense buildings—glomeruli; tubules. Various problems can arise in each of these areas.

If the process diagnosed by a pathologist in a kidney biopsy affects the glomeruli or capillaries, it is called microvascular inflammation (MVI). As a result of inflammation, the sensitive elements responsible for the proper functioning of the organ and blood filtration are damaged. Interestingly, until a few years ago, many cases that we now recognize as microvascular inflammation were not formally classified as rejection. Today, following the 2022 update of the Banff International Classification, which is the gold standard for histopathological diagnosis of transplanted kidneys, we know this is a severe warning sign, suggesting that something is wrong with the transplanted organ

At the same time, another, more difficult problem is emerging.

"At the Department and Clinic of Nephrology, Transplant Medicine, and Internal Medicine at Wroclaw Medical University, we are conducting extensive research into the mechanisms of transplant rejection. It turns out that antibody-mediated rejection, known as humoral rejection, is responsible for over 50% of transplanted kidney loss, which consequently leads to the need for dialysis or another transplant. We are working on understanding and limiting this adverse immune response," says Prof. Banasik.

In simple terms, more than half of transplant losses can be attributed to the fact that the immune system still considers the donated kidney to be an intruder, despite medication, supervision, and the experience of clinicians.

Classic laboratory tests are no longer sufficient

For years, antibodies directed against human leukocyte antigens (HLA) – individual and unique molecules characteristic of each patient – were considered the main factor responsible for antibody-mediated rejection. Simply put, the recipient's immune system recognizes foreign antigens present on the donor organ and tries to fight them. It was these antibodies that were at the forefront of diagnosing so-called humoral rejection. But in practice, a significant discrepancy increasingly emerged. In the biopsy, the pathologist described features typical of antibody-mediated rejection; kidney function was deteriorating, and standard tests assessing the recipient's immune system after transplantation did not show evidence of anti-HLA antibodies. In other words, something was damaging the transplant, but the tools at our disposal did not signal a problem.

A team from Wroclaw Medical University decided to look at antibodies outside the main list —non-HLA antibodies. These are antibodies directed not against tissue compatibility antigens but against other molecular elements present in the transplanted kidney. The focus was on those targeting the angiotensin II type 1 receptor (AT1R), a protein that regulates blood vessel tone.

During each transplant, the transplanted organ spends some time outside both the donor's and the recipient's body. During this time, even under appropriate transport conditions, such as reduced temperature, processes begin in the organ, damaging its structure due to the lack of blood supply. We noticed that, at this critical moment, due to damage, the transplanted kidney may be exposed to the very elements against which non-HLA antibodies are directed.

Transplant archive and digital magnifying glasses

The study, published in the journal Frontiers in Immunology, included 167 kidney transplant patients. What they all had in common was that they had undergone a biopsy in the past due to deterioration of transplant function. Pathologists evaluated these specimens according to the latest Banff 2022 criteria, looking for signs of microvascular inflammation. At the same time, blood samples were analyzed for the presence of AT1R antibodies.

Two groups were formed: 79 patients with MVI and 88 patients without MVI.

Even a first glance at the data was intriguing. Nearly half of the patients with MVI had positive AT1R antibodies, while in the group without MVI, less than a quarter did. Something clearly linked high levels of these antibodies to microvascular inflammation.

But the Wroclaw team went a step further. Instead of stopping at a simple positive/negative, they treated the AT1R level as a scale and began looking for an answer to the question: at what point does this noise become really dangerous?

Podcast: How to better protect a transplanted kidney?

Click here to listen!

Artificial intelligence enters the game

This is where artificial intelligence comes into play. Not in its spectacular form, generating images or texts, but in its analytical form. Algorithms help calculate relationships that the human eye cannot detect.

"The challenge was to use artificial intelligence tools. I began my adventure in this area relatively shortly before starting my doctorate, while I have been involved in medical data analysis for several years. Even during my studies, I conducted various research projects as part of the Nephrology Student Scientific Association's activities. Currently, I am systematically expanding my knowledge of AI methods, learning new data analysis techniques and their practical applications,“ says Jakub Mizera, MD. ”When I started my research work, artificial intelligence was not yet as common in everyday life as it is today. The choice of my doctoral thesis topic was therefore a natural consequence of the growing importance and potential of these technologies and the increasingly easy access to advanced analytical tools. My assistant supervisor, Prof. Maciej Pondel from Wroclaw University of Economics and Business, provides me with invaluable substantive support in the field of AI," he adds.

The study used, among other things, logistic regression, ROC curve analysis, and association rule mining, a technique that searches for patterns in data such as “if X occurs, what is the probability that Y will occur.”

The comparison showed that the higher the level of anti-AT1R antibodies, the greater the risk of microvascular inflammation. The correlation was statistically significant, although not very strong overall. An increase in AT1R levels also affected the results. Each additional unit increased the risk by approximately 6%. However, the most interesting finding was the division of patients into 5 groups based on antibody levels. Only in the highest quintile, above approximately 12 U/ml, did the risk of microinflammation clearly “jump” above the average. In this group, microvascular inflammation occurred in almost two-thirds of patients, and the risk was about 37% higher than in the entire study group. In other words, not every positive result is equally dangerous. Only high AT1R titers appear particularly significant.

MD. Mizera sees this as a starting point for the next step:

"We are currently conducting research that includes not only anti-AT1R antibodies, but also their subclasses. In addition, we are analyzing other non-HLA antibodies, including anti-ETAR, anti-ETBR, anti-PAR1, and anti-PAR2. In the future, we plan to use artificial intelligence tools to develop a comprehensive panel of non-HLA antibodies. This could enable not only earlier detection of organ rejection, but also the identification of patients at increased risk of organ loss. Such a panel would be a valuable addition to the tests currently in use, including anti-HLA DSA antibody testing and C4d assessment in biopsies", he points out. "Antibody testing in blood is much less invasive than a biopsy and, in the future, could significantly facilitate post-kidney-transplantation patient monitoring and improve transplantologists' clinical decision-making."

For the layman. The idea is to create an “immunological profile” of the patient that can be read from the blood, and for the algorithm to tell the doctor: “this patient is in a higher risk group, it is worth reacting faster here and, for example, intensifying immunosuppressive treatment.”

AI as the new stethoscope of transplantology

Is medicine ready to use AI not only for data analysis but also in other areas? The Wroclaw team responds with a cautious but decisive “yes.”

“The results of patients at the transplant clinic of the University Clinical Hospital in many cases significantly exceed the average survival time of a transplanted organ,” says Prof. Banasik. "Our goal is to identify new factors and the relationships between them that can lead to premature loss of transplanted kidneys. Suppose we can recognize them and introduce appropriate treatment to limit their negative impact. In that case, we can not only extend the transplant's functional lifespan beyond average but also avoid the need for retransplantation in the same patients. As a result, transplants will be able to go to other patients, increasing organ availability and improving patients' quality of life. We believe that artificial intelligence tools — from analyzing the relationships between test results to autonomous analysis of biopsy images — have enormous potential to accelerate and improve the diagnosis of transplant rejection. Medicine, including transplantology, should be open to these technologies, as their presence in clinical practice is becoming inevitable. Failure to take advantage of the opportunities offered by AI would mean giving up valuable support that can realistically improve the effectiveness and efficiency of patient care."

In this sense, artificial intelligence does not replace the doctor. It is more like a new type of stethoscope —a tool that lets you hear what was previously lost in the noise of data.

Young science

Finally, there is one aspect that is easy to overlook in the world of large numbers and complex algorithms: this research is primarily the work of young scientists.

MD. Mizera is still pursuing his doctoral degree. He works in a team that brings together nephrologists, transplantologists, pathomorphologists, biochemists, and data analysis specialists. When asked what is most valuable for a doctoral student in such a project, he replies:

"For me, the most valuable aspect of working on this publication was the opportunity to collaborate with recognized authorities in the field of transplantology and nephropathology. I am incredibly grateful to all my co-authors for sharing their knowledge and experience and for their support at every stage of the research. The Doctoral School of Wroclaw Medical University creates excellent conditions for the development of young scientists — not only by improving their research skills, but also by providing real opportunities to collaborate with experts from various fields. Support in obtaining the funding needed to conduct research is also extremely valuable, allowing doctoral students to carry out ambitious, interdisciplinary projects at a high level", says MD. Mizera.

This is a good place to end the story. Although the title of the thesis contains difficult words incomprehensible to people not directly involved in medicine, such as “microvascular inflammation” and “non-HLA antibodies,” it is actually about a straightforward topic: how to keep a donated organ functioning for as long as possible.

For this to happen, we need to identify factors that, although unknown to date, contribute to premature organ loss. And, as the example from Wroclaw Medical University shows, this requires experienced clinicians, young researchers, and algorithms that patiently review thousands of rows of tables, looking for patterns hidden from the human eye.

D. Sikora

MD. Jakub Mizera, Department of Nephrology, Transplant Medicine and Internal Medicine

FAQ: Angiotensin II Type 1 Receptor Antibodies and Microvascular Inflammation (MVI) in Kidney Transplant Recipients

What is the main focus of this research article?

The study focuses on evaluating the potential role of Angiotensin II type 1 receptor antibodies (AT1R abs) as a contributor to microvascular inflammation (MVI) in kidney transplant recipients. The researchers aimed to provide insights using both traditional statistical methods and artificial intelligence (AI) based approaches.

Why is Microvascular Inflammation (MVI) a significant concern in kidney transplantation?

MVI is a subcategory of antibody-mediated rejection (ABMR). The classification and diagnosis of ABMR, particularly MVI, were recently reappraised in the Banff 2022 classification. Following this update, ABMR has emerged as one of the most frequently diagnosed causes of immunological graft damage. Traditionally, ABMR was primarily associated with donor-specific anti-HLA antibodies (DSAs), but this study explores non-HLA factors as a potential contributor, since traditional markers fail to accurately diagnose this type of rejection.

How was Microvascular Inflammation (MVI) defined and diagnosed?

MVI was diagnosed histologically based on biopsy findings. According to the Banff Classification of Renal Allograft Pathology, MVI was defined by the presence of a glomerulitis score (g) > 0 and/or peritubular capillaritis score (ptc) > 0. The evaluations were based on the 2022 Banff Classification.

What was the key finding regarding AT1R antibodies and MVI prevalence?

The study found that patients with MVI demonstrated a statistically significantly higher prevalence of AT1R antibodies compared to those without MVI.

In the group with MVI, 37 out of 79 patients (46.8%) tested positive for AT1R antibodies.

In the group without MVI, 21 out of 88 patients (23.9%) tested positive.

The difference was statistically significant (p=0.0018).

Do all levels of AT1R antibodies correlate with an increased risk of MVI?

No. The AI analysis, specifically using association rule mining (ARM), demonstrated that high AT1R abs titers (>12 U/ml) were particularly associated with MVI. Lower levels did not show a meaningful association.

For the highest quantile of AT1R abs (12.388–40.0 U/mL), the likelihood (confidence) of MVI was 64.7%.

The lift value for this highest quantile was 1.37, indicating that the probability of MVI was 37% higher in this group compared to the general population.

What other factors were associated with Microvascular Inflammation in this cohort?

The study identified several other significant associations:

C4d Positivity: A highly significant association was observed between MVI and C4d positivity (p=0.000029). C4d positivity (grades 1, 2, and 3) was more frequently observed in patients exhibiting MVI.

Histopathological Alterations: Patients with MVI showed significantly elevated mean histopathological scores for the interstitium (i score), tubules (t score), and arterial vessels (v score).

Warm Ischemia Time (WIT): Kidneys exposed to longer warm ischemia time were statistically more often affected by MVI (p=0.01).

Were other non-HLA antibodies investigated, and what were the results?

The study screened for anti-endothelial cell (AECA) antibodies and anti-endothelin type A receptor (ETAR) antibodies.

Anti-ETAR antibodies: Analysis of anti-ETAR antibodies did not reveal a statistically significant association with MVI in this study.

Anti-AECA antibodies: An independent analysis was not conducted for anti-AECA due to a high number of missing values (59 unknown records).

However, when analyzing the combined presence of AT1R, AECA, and ETAR antibodies, seropositivity for at least one was significantly more frequent in MVI patients (63.2%) compared to non-MVI patients (42%) (p=0.024).

How did the researchers use Artificial Intelligence (AI) techniques?

Given the complexity of the issue, the methodology extended beyond conventional statistical approaches by incorporating AI techniques to maximize insights. These advanced tools included:

Correlation Analysis: To assess the strength of the association between continuous AT1R abs levels and MVI.

Logistic Regression Modelling: To quantify how AT1R abs levels affect the probability of MVI. This model showed that with each unit increase in AT1R abs, the odds of inflammation increase by approximately 6.1% (Odds Ratio 1.0614).

Receiver Operating Characteristic (ROC) Curve Analysis: To evaluate the diagnostic performance of AT1R abs for predicting MVI (AUC was 0.600).

Association Rule Mining (ARM): To uncover patterns in categorical groupings of AT1R abs values (quantiles) and their association with MVI, which strongly supported the significance of the highest AT1R quantile.

What are the main conclusions and implications of these findings?

The study concludes that AT1R antibodies may contribute to MVI and subsequent graft injury in kidney transplantation. These findings highlight the need for expanded immunologic monitoring beyond traditional anti-HLA antibody screening.

The presence of AT1R abs may be associated with MVI development, potentially contributing to allograft injury in a subset of cases.

The findings suggest that anti-HLA antibodies, while traditional, may lack sufficient sensitivity to capture all instances of antibody-mediated injury, particularly those driven by non-HLA antibodies like AT1R abs.

The researchers recommend that future studies should focus on validating integrated diagnostic models that combine these parameters (non-HLA antibodies and histopathology) to lead to more targeted therapeutic interventions.

Powerd by NotebookLM