Natural history model of prostate cancer using PSA testing data

In the landscape of modern medicine, few tools are as ubiquitous yet as misunderstood as the prostate-specific antigen, or PSA, test. This simple blood test measures a protein naturally produced by the prostate gland. In a healthy body, only a tiny amount of this protein leaks into the bloodstream. However, when the prostate is affected by cancer, the cell walls become disrupted, allowing more of the protein to escape, causing levels in the blood to rise. Because the test is inexpensive and easy to perform, it has become a common way to look for early signs of the disease. Yet, the test is far from perfect. PSA levels can also rise due to benign conditions like inflammation or an enlarged prostate, leading to false alarms. This creates a difficult dilemma for doctors and patients: when a level is high, the only way to confirm cancer is to perform a biopsy, a procedure that carries its own risks and discomfort. In many places, including Sweden, there is no national program that invites every man to be tested at regular intervals. Instead, testing happens opportunistically, driven by individual choices and doctor recommendations, creating a vast but messy collection of data that is hard to interpret.

For decades, researchers have tried to understand the "natural history" of prostate cancer—the invisible timeline of how the disease begins, grows silently, and eventually becomes noticeable. Knowing this timeline is crucial. If the disease moves slowly, screening might catch harmless growths that would never have caused harm, leading to unnecessary treatment. If it moves quickly, missing a test could be fatal. The challenge has been that we cannot watch this process unfold in real time for a single person; we only see snapshots taken at irregular intervals. A new study by researchers at the Karolinska Institutet in Sweden has developed a sophisticated way to piece together these snapshots. By using a massive database of real-world testing records from Stockholm, they built a mathematical model that reconstructs the hidden life of the disease, revealing just how slowly it often progresses and how the timing of tests influences what we find.

The researchers started with a unique resource: the Stockholm Prostate Cancer Diagnostics Register. This database links together millions of records, tracking when men living in Stockholm had their PSA tests, when they underwent biopsies, when they were diagnosed with cancer, and when they died or moved away. The study focused on nearly two million men who were alive and cancer-free at the start of 2006. Because the testing in this region was not organized into a strict schedule, the data looked chaotic. Men tested at different ages, with different frequencies, and some were tested only once while others were tested dozens of times. The team's goal was to turn this chaotic stream of numbers into a clear picture of the disease's behavior. They created a model that treats a man's life as a journey through three states: being healthy, having a preclinical form of cancer that is detectable by a test but causes no symptoms, and finally having clinical cancer that causes symptoms or is found through a positive test.

To make sense of the irregular testing patterns, the researchers had to account for the fact that the decision to perform a biopsy depends heavily on the PSA result. If a man's PSA level is low, he is unlikely to get a biopsy. If it is high, the chance of a biopsy increases. The model also had to handle the fact that a biopsy is not perfect; it can sometimes miss cancer even when it is there. By feeding all the available data into their system, the researchers could estimate the hidden parameters of the disease. They simulated millions of possible scenarios to ensure their method worked correctly before applying it to the real Stockholm data. This process allowed them to separate the noise of opportunistic testing from the true signal of the disease's progression.

The results offer a strikingly clear view of the disease's pace. The study found that for the average man in this population, the time between the start of the preclinical cancer and the point where it would cause symptoms is incredibly long. The median duration of this silent phase was estimated to be 33.2 years. This means that for many men, the disease begins decades before it would ever become a problem. Furthermore, the researchers observed how the PSA levels changed over time. Before the cancer actually starts, the PSA level rises very slowly, doubling only once every 26.8 years. Once the cancer begins, the rate of increase speeds up significantly, with the level doubling every 4.9 years. This slow initial growth helps explain why the disease is so prevalent in older men; it accumulates over a lifetime. By age 79.2, the model suggests that half of all men will have developed a form of prostate cancer that is detectable by a biopsy.

The study also highlighted the limitations of the current testing approach. Because the disease progresses so slowly, a significant portion of the cancers found through screening are likely to be those that would never have caused symptoms or death during a man's natural lifespan. This supports the ongoing debate about overdiagnosis, where men are treated for conditions that do not require treatment. The researchers noted that their model works best when there is a steady stream of data, but the opportunistic nature of Swedish testing meant that for many men, there were very few data points. This made it difficult to estimate certain complex variations in how the disease progresses, leading the team to rely on the most robust findings: the slow speed of the disease and the long window of opportunity before symptoms appear.

Ultimately, this work provides a powerful tool for understanding the balance between the benefits and harms of screening. By reconstructing the natural history of prostate cancer from routine data, the researchers have shown that the disease is often a slow-moving companion rather than a rapid aggressor. This insight is vital for policymakers and doctors who must decide how to guide men through the confusing landscape of screening. The study confirms that while PSA testing can detect cancer early, the sheer slowness of the disease means that finding it early does not always mean saving a life, and it often means treating a condition that would have remained harmless. The model serves as a foundation for future studies that could help design better screening strategies, ensuring that medical resources are used where they can truly make a difference.