Gram Research analysis of a 2026 study suggests two proteins, GSTA1 and ALDH1A1, may help flag early kidney damage in people with diabetes. In 810 participants, a computer model using seven candidate proteins reached a test score above 0.8, where 1.0 is perfect. This is an early-stage finding that still needs confirmation.
Kidney damage is a serious complication of diabetes, and it can be hard to catch early. A new study from a diabetes clinic looked at proteins and small molecules in the body to search for warning signs. Researchers compared people with diabetes alone, people with a low level of protein leaking into their urine, and people with higher levels. Two proteins, GSTA1 and ALDH1A1, stood out as the most important in a computer model that sorted people into groups. The model reached a test score above 0.8 on a scale where 1.0 is perfect, which is encouraging. The study also found that a form of kidney disease without any protein leakage was rare, appearing in about 0.7% of people with diabetes. According to research reviewed by Gram, these findings are early and still need to be confirmed in larger studies before they can guide everyday care.
Key Statistics
A 2026 cross-sectional study of 810 people with diabetes found that GSTA1 and ALDH1A1 ranked highest in importance among candidate protein markers for low-level albuminuric diabetic kidney disease.
In a 2026 study of 810 people, a computer model built from seven candidate proteins reached a five-fold cross-validation mean AUC above 0.8, where 1.0 represents perfect classification.
A 2026 study from a single diabetes clinic found that non-albuminuric diabetic kidney disease accounted for 0.7% of all people with diabetes and 1.6% of people with diabetic kidney disease.
The Quick Take
- What they studied: Researchers asked whether certain proteins and small molecules in the body can act as signals for early kidney damage in people with diabetes, especially a stage where small amounts of protein begin leaking into the urine.
- Who participated: People with diabetes treated at one hospital endocrinology department. The team first reviewed records of all diabetes patients from 2023 and 2024. Newly diagnosed patients were then enrolled from January to March 2025. The final comparison included 810 people: 270 with diabetes alone, 270 with low-level kidney leakage, and 270 with higher levels.
- Key finding: Two proteins, GSTA1 and ALDH1A1, ranked highest in importance in a computer model that separated people with early kidney damage from those without it. A group of seven candidate proteins together reached a score above 0.8 on a scale where 1.0 is perfect. The version of kidney disease without any protein leakage was rare, found in about 0.7% of people with diabetes.
- What it means for you: These results are early and are not a test you can use today. If you have diabetes, keep up your regular kidney checks, including urine protein tests and blood tests for kidney function. Work with your doctor to manage blood sugar and blood pressure, since these steps are the proven ways to protect your kidneys.
The Research Details
This study combined two approaches. First, the researchers looked back at medical records of all diabetes patients seen in 2023 and 2024. They sorted patients by how much protein was leaking into their urine, which is measured by a urine albumin test. This helped them estimate how common each type of kidney disease was.
Second, the team enrolled newly diagnosed patients from early 2025 and collected blood or body samples. They used a method called integrated proteomic-metabolomic analysis, which measures large numbers of proteins and small chemical products of metabolism at once. This allowed them to look for differences between groups in a wide range of body chemicals.
Finally, the researchers randomly selected 810 people to test whether the most promising proteins could tell the groups apart. They used a computer learning tool called XGBoost to build a prediction model. They also used a method called SHAP to rank how much each protein contributed to the model’s predictions, and they checked the model’s accuracy by repeatedly testing it on different slices of the data.
Using both record review and fresh sample collection lets researchers see how common each kidney disease stage is, and then look closely at the body chemistry of the people who have it. Measuring many proteins at once can reveal patterns that a single test would miss. The random selection step is also useful because it reduces the chance that the groups were chosen in a biased way. Together, these steps make the findings a reasonable starting point for future work, even though they are not yet proof.
The study has several strengths. It used a large verification group of 810 people, selected at random, and it tested its model on data it had not been built from. The team also used standard, widely recognized statistical methods. However, the study was run at a single hospital, so its results may not apply to people in other regions or health systems. The sample size for the detailed protein analysis was not clearly reported in the summary, which makes it harder to judge that part of the work. Readers should treat the findings as promising but unconfirmed.
What the Results Show
The researchers found that people with low-level protein leakage in their urine differed from people with diabetes alone and from people with higher levels of leakage. These differences showed up across 33 metabolic pathways, which are chains of chemical reactions in the body. The differences involved 29 proteins that were measured in the detailed analysis.
From these proteins, the team selected seven candidates that worked well together in a prediction model. The model’s accuracy, measured by a score called AUC, stayed above 0.8 in a five-fold cross-validation test. This test repeatedly checks the model on different parts of the data, so the score reflects how well it works on new information. A score above 0.8 is generally considered strong, though it is not perfect.
Within the model, GSTA1 and ALDH1A1 ranked highest in importance. The researchers reported their contribution scores in both the training data and the testing data, and the two proteins held up in both sets. This consistency is a reason the team highlighted them, though the numbers still come from one study.
The proteins identified in the study were mostly located in parts of cells such as the cell’s outer membrane, its interior fluid, and small packets that cells use to send messages. In a pathway analysis, the proteins were linked to processes such as metabolism, amino acid production, protein building in cells, folate use, and carbon metabolism. These connections help explain why the researchers looked at these specific proteins.
The study also found that the version of kidney disease without any protein leakage was uncommon. It made up about 0.7% of all people with diabetes and about 1.6% of people with diabetic kidney disease. Patients with this form tended to be older, which suggests it may develop through a different path than the low-level leakage stage.
The authors proposed a possible sequence of disease: diabetes, then low-level protein leakage, then more advanced protein leakage. They suggested that damage to the small tubes in the kidney that filter and reabsorb substances may be an early part of this path. This idea is plausible but has not been directly confirmed by this study.
Kidney damage in diabetes is well studied, and doctors already use urine albumin tests and blood tests for kidney function to find it. Earlier research has linked oxidative stress, which is a form of cell wear caused by harmful molecules, and damage to kidney tubules with diabetic kidney disease. The proteins highlighted in this study fit with that broader picture. Still, this is one study from one center, so it adds a new candidate rather than overturning existing knowledge.
The study is a snapshot in time rather than a long-term follow-up. That means it can show that certain proteins differ between groups, but it cannot prove that people move from one stage to the next in the order the authors suggest. The work was done at one hospital, so the results may not match other populations. The detailed protein analysis was not clearly sized in the summary, and the verification step is preliminary. The model has not yet been tested in an outside group of patients. Finally, a high accuracy score in a research setting does not guarantee the same performance in routine clinics.
The Bottom Line
Confidence that GSTA1 and ALDH1A1 will become useful clinical tests is low at this stage, because they have only been studied in one early-stage study. Confidence in standard kidney care is high. Regular urine albumin testing, kidney function blood tests, blood sugar control, and blood pressure control are backed by long-standing clinical guidelines and remain the best current approach for protecting kidneys in people with diabetes.
People living with diabetes, their families, and healthcare professionals who manage diabetes may find this research interesting. Researchers who study kidney disease may use it as a starting point for larger studies. People who do not have diabetes, or who already follow their doctor’s kidney monitoring plan, do not need to change anything based on this study alone. Anyone worried about their kidneys should talk with a doctor rather than rely on research summaries.
Turning early research like this into a reliable test usually takes years. The next steps would include studies in larger and more diverse groups, follow-up over time to see how disease progresses, and testing in regular clinical settings. Because of this, any new blood test based on these proteins is unlikely to be routinely available for several years, and it may take longer. Meanwhile, the standard steps for protecting kidneys can begin now and show benefits over months and years.
Frequently Asked Questions
Can a blood test detect early diabetic kidney disease using GSTA1 and ALDH1A1?
Not yet. A 2026 study found GSTA1 and ALDH1A1 ranked highest in a computer model of protein markers, but this preliminary work involved 810 people from one hospital. Doctors currently rely on urine albumin tests and kidney function blood tests.
What is low-level albuminuric diabetic kidney disease?
It is an early stage of kidney damage in people with diabetes, where small amounts of a protein called albumin leak into the urine. The levels are above normal but below those seen in more advanced disease. Catching this stage early gives more time to slow progression.
How rare is non-albuminuric diabetic kidney disease?
In a 2026 study, the form without any albumin leakage was uncommon, accounting for 0.7% of all people with diabetes and 1.6% of those with diabetic kidney disease. Patients with this form tended to be older than those with leakage.
How can I protect my kidneys if I have diabetes?
Keep blood sugar in your target range, control blood pressure, get yearly urine albumin tests, and check kidney function with blood tests. Your doctor can set personal targets, and these steps are supported by long-standing clinical guidelines.
Want to Apply This Research?
- Record your urine albumin-to-creatinine ratio, HbA1c, blood pressure, and estimated kidney function (eGFR) each time you get them checked, along with the date. Track these numbers together so you can see trends over time rather than focusing on a single result.
- Set a reminder in the app for your yearly urine albumin test and your routine kidney function blood test. Log your daily blood pressure readings if your doctor has asked you to monitor them, and bring the log to appointments.
- Review your logged values every three months and note any changes in urine protein, kidney function, blood sugar, or blood pressure. Share the trend summary with your healthcare team at least once a year, and ask them to explain any results that are moving in the wrong direction. Keep an eye on new research, but use it to start conversations with your doctor rather than to change treatment on your own.
This article summarizes a single early-stage study and is for general information only. It is not medical advice and should not be used to diagnose, treat, or change any medication. People with diabetes should talk with their doctor or kidney specialist before making decisions about testing or treatment based on this research.
This research translation is published by Gram Research, the science division of Gram, an AI-powered nutrition tracking app.