New Model Transforms Prostate Cancer Screening Decisions

Prostate cancer remains a formidable health challenge for men in the United States, ranking as the second-leading cause of cancer death among American men. According to University of Michigan Health Lab, about one in eight men will face a prostate cancer diagnosis in their lifetime, with risk factors varying considerably by age and race. Despite the prevalence of the disease, the tools available for interpreting the results of the most common screening test—the prostate-specific antigen, or PSA—have long been limited in their ability to help patients and doctors make informed decisions about next steps.

Each year, an estimated 10 million PSA tests are performed across the country. Yet, as Dr. Kristian Stensland, Assistant Professor of Urology at the University of Michigan, pointed out, "Current tools don’t take into account how long someone may live or the benefit a patient may receive from treatment." The challenge has been not just in detecting prostate cancer, but in determining which patients will truly benefit from further screening and, potentially, aggressive intervention. Overdiagnosis and unnecessary biopsies have been persistent concerns.

Now, a new model—developed by researchers at the University of Michigan and detailed in the Annals of Internal Medicine—aims to change the landscape of prostate cancer screening. This prostate cancer-specific mortality (PCSM) model harnesses PSA results and a suite of patient-specific factors to predict the risk of dying from prostate cancer, rather than simply identifying the presence of cancer itself. The model’s creators hope it will provide both patients and clinicians with a clearer understanding of what a PSA result means for long-term health and life expectancy.

The development of the model relied on data from the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial, which recruited more than 33,000 men aged 55 to 74 years between 1993 and 2001. This cohort, with a median age of 62, provided a rich dataset for the researchers, who incorporated not only PSA levels but also family history of prostate cancer, race, age, body mass index, smoking status, and medical history (including hypertension, diabetes, or stroke) into their calculations.

To ensure the robustness of the model, the team externally validated it using PSA data from over 174,000 patients in the Veterans Affairs Healthcare System—again, men aged 55 to 74—collected between 2002 and 2006. This validation cohort closely mirrored the derivation group, with a similar median age and body mass index, and two-thirds of participants having a history of smoking. The median PSA at first screening was 1.13 ng/dL in the PLCO cohort and 0.97 ng/dL in the VA group.

The results were striking. The model was able to predict prostate cancer-specific mortality risk with greater accuracy than previous tools, such as the Prostate Biopsy Collaborative Group (PBCG) model. Specifically, the PLCO model demonstrated better discrimination at 29.5 years in the derivation cohort (AUC 0.6664) and at 20 years in the validation cohort (AUC 0.776), outperforming the PBCG model in both cases. As reported by Patrick Lewicki, MD, MS, and colleagues in the Annals of Internal Medicine, "PSA screening for [prostate cancer] will reach its greatest efficacy when conducted in a risk-adjusted manner. To date, few tools exist to implement risk-adjusted screening. We designed and externally validated a prediction model for risk stratification of patients at the point of PSA screening."

In practical terms, the model identified a clinically relevant threshold of 0.5% prostate cancer-specific mortality for deciding whether to discontinue PSA screening. This threshold captured 17% of patients who had a 4% likelihood of dying from prostate cancer before the age of 85. The net benefit of this approach was calculated at 0.013, suggesting a meaningful improvement in screening decisions.

It’s important to note, as the researchers themselves did, that the model was built and tested on data from two decades ago. Prostate cancer treatments and diagnostic approaches have evolved since then. "Even though prostate cancer treatment is different now, our model improves on previous tools and can be used to decide how we do PSA screens," Dr. Stensland told University of Michigan Health Lab. The research team is now working to implement the model in clinical settings, with the goal of reducing unnecessary biopsies and ensuring that only those who stand to benefit most from further intervention receive it.

Another key finding from the validation process was that risk prediction for prostate cancer-specific mortality before age 85 was better with the PLCO model (AUC 0.709) than with the PBCG model (AUC 0.642). Among patients classified as the highest risk decile by the older PBCG model, the new PLCO model significantly outperformed it in both the derivation and validation cohorts, confirming its utility for identifying those at greatest risk.

The issue of overdiagnosis—identifying cancers that would not cause harm during a patient’s lifetime—remains a core concern in prostate cancer screening. Some men with a PSA level below the traditional threshold of 3 ng/mL may still harbor aggressive cancers. To address this, research from the STHLM3 trial (as referenced in European Urology) suggests that combining PSA with more advanced tests, such as the Stockholm3 blood test, can improve detection of high-risk cancers while reducing the identification of indolent, non-threatening tumors. The Stockholm3 test integrates multiple proteins, a germline polygenic risk score, and clinical variables, offering a more nuanced risk assessment. In the STHLM3 trial of over 59,000 men, using Stockholm3 in addition to PSA increased the detection of high-risk biochemical recurrence by nearly ninefold compared to PSA alone.

Despite these promising advances, the researchers acknowledge limitations. Patients in the original PLCO cohort were treated more aggressively than might be typical today, potentially affecting the model’s generalizability. Additionally, while the model incorporates a broad range of risk factors, ongoing changes in prostate cancer management—such as the adoption of active surveillance for low-risk cases—mean that continual refinement and validation will be necessary as new data emerge.

The development of this new PCSM model marks an important step toward more personalized prostate cancer screening. By integrating life expectancy, comorbidities, and individual risk factors, clinicians and patients can move beyond a one-size-fits-all approach. As Dr. Stensland and his colleagues continue to refine and implement their model, the hope is that fewer men will undergo unnecessary procedures, and those at highest risk will receive timely, effective care.

For now, the message is clear: interpreting a PSA test result is more complex than a simple number. With the advent of risk-adjusted models and next-generation screening tools, the future of prostate cancer detection and management looks increasingly precise—and, perhaps, more hopeful for men facing this common disease.