Concentric Hypertrophy as a Prognostic Factor and Mediator of Diabetes-Related Risk in HFpEF: Development and Validation of Prediction Models Based on Cox Regression and Machine Learning Approaches Across Two Randomized Controlled Trials

Heart failure is a condition where the heart struggles to pump enough blood to meet the body's needs, but it is not a single disease with a single cause. One specific and increasingly common type, known as heart failure with preserved ejection fraction, occurs when the heart muscle is actually strong enough to squeeze blood out normally, yet the organ itself has become stiff and unable to fill up properly. This stiffness often develops alongside other health issues, particularly diabetes, which affects nearly half of the patients with this condition. For decades, doctors have known that diabetes makes the outlook for these patients worse, but the exact biological pathway connecting high blood sugar to a failing heart has remained somewhat of a mystery. Understanding this link is crucial because without knowing how the damage happens, it is difficult to predict who is most at risk or how to treat them effectively.

Researchers at Shengjing Hospital in China set out to solve this puzzle by looking closely at the shape and structure of the heart muscle itself. They focused on a specific pattern called concentric hypertrophy, a condition where the heart walls thicken and the chamber inside becomes smaller, much like a balloon that has been squeezed from all sides until it is tight and rigid. This thickening is measured by two specific numbers: the total weight of the heart muscle relative to the person's body size, and the thickness of the walls relative to the size of the chamber. The team wanted to know if this specific shape change was merely a side effect of diabetes or if it was actually the mechanism driving the poor outcomes. To find the answer, they analyzed data from two major international clinical trials involving thousands of patients, carefully filtering the records to focus only on those with complete measurements of their heart structure.

The study began by sorting patients into groups based on the shape of their hearts: some had normal walls, some had thickened walls but normal size, and others had both thickened walls and a smaller chamber. When the researchers tracked who experienced serious events like death from heart causes or hospitalization for heart failure, a clear pattern emerged. Patients with the specific "squeezed" shape, known as concentric hypertrophy, were significantly more likely to suffer these bad outcomes than those with other heart shapes. The data showed a direct, linear relationship: as the heart walls got thicker and the chamber got smaller, the risk of trouble increased steadily. This finding was robust, holding true even after the researchers adjusted for age, gender, blood pressure, and other common health factors.

Perhaps the most significant discovery was how this heart shape related to diabetes. The researchers found that having diabetes did increase the risk of bad outcomes, but when they accounted for the presence of concentric hypertrophy, the direct link between diabetes and the bad outcome became much weaker. This suggests that diabetes does not damage the heart in a vague, general way; rather, it appears to cause the heart to thicken and stiffen, and it is this specific structural change that then leads to the failure. In statistical terms, the heart shape acted as a mediator, carrying the harmful effects of diabetes to the final outcome. This means that the thickening of the heart wall is not just a symptom, but a critical step in the chain of events that leads to heart failure in diabetic patients.

Armed with this new understanding, the team built two different types of prediction tools to help doctors identify high-risk patients. The first tool used a traditional statistical method, while the second used a more complex computer learning approach that could detect subtle, non-linear patterns in the data. Both tools incorporated the presence of concentric hypertrophy along with other factors like age, blood pressure, and kidney function. When tested, both models proved highly accurate at predicting who would face serious heart events over the next few years, with their accuracy improving over longer time periods. The researchers even created online calculators so that doctors could input a patient's specific numbers and get a personalized risk estimate.

To ensure these tools were reliable, the team tested them on a completely separate group of patients from a different trial. The models performed just as well in this new group as they did in the original one, confirming that the findings were not just a fluke of the first group of patients. The researchers also discovered that using both models together provided an even better way to sort patients into risk categories. By combining the results, they could identify a group of high-risk individuals with greater precision than either model could on its own. This combined approach revealed that patients flagged as high risk by both tools faced a much higher chance of events than those flagged by only one, allowing for a more nuanced view of who needs the most urgent care.

The study also clarified why some other heart shapes did not show the same level of risk. Patients with a different type of thickening, where the heart chamber gets larger instead of smaller, were actually quite rare in this group of patients. This makes sense because that specific shape is usually caused by volume overload and is often linked to a weak pumping ability, which these patients did not have. The specific "squeezed" shape, driven by pressure and stiffness, was the one uniquely tied to the poor outcomes in heart failure with preserved ejection fraction. The research also highlighted that certain common measurements, like the level of hemoglobin in the blood and the classification of heart failure symptoms, were among the strongest predictors of risk, alongside the heart shape itself.

While the study provided powerful new tools and insights, the authors noted some limitations. The number of patients with complete heart measurements was smaller than the total number of participants in the original trials, which meant the final group was a specific subset. Additionally, some important blood markers that could have refined the predictions were missing for many patients, preventing the models from being even more precise. Despite these constraints, the findings offer a clear path forward. By identifying the specific heart shape that signals danger, doctors can now better understand how diabetes harms the heart and use new prediction tools to spot the patients who need the most aggressive treatment. The work suggests that targeting this specific thickening of the heart muscle could be a key strategy for improving the lives of those living with this difficult form of heart failure.