TAMPA, Fla. August 13: Researchers at Moffitt Cancer Center have developed three mathematical biomarkers that may help physicians personalize adaptive therapy for prostate cancer by predicting treatment outcomes early in treatment. The biomarkers are specialized mathematical measures calculated from routine prostate-specific antigen (PSA) blood tests.
The study, which appears in JAMA Oncology, was co-led by joint senior authors Alexander R. Anderson, Ph.D., chair of Moffitt’s Integrated Mathematical Oncology Department and Philip Maini, Ph.D., director of the Wolfson Centre for Mathematical Biology at Oxford University.
Standard cancer treatments focus on killing cancer cells by administering continuous high doses of anti-cancer drugs. However, this aggressive approach can result in treatment resistance by allowing resistant cell populations to expand unchecked.
To lower the chances of resistance, under an approach called adaptive therapy, treatment is paused when the cancer is under control and restarted when signs of growth return. This strategy is designed to slow the rise of drug-resistant cells that can eventually make therapies ineffective. Previous Moffitt studies have shown that some patients can benefit from this line of attack, but others do not. Until now, doctors have had few ways of predicting who is most likely to respond.
The Moffitt team set out to change that. They created three mathematical biomarkers that use PSA measurements collected during the first treatment cycle. Those early PSA changes can provide clues about how a patient’s tumor is behaving and how it may respond to treatment in the future.
The biomarkers are:
- The AT Score (Adaptive Therapy Score): AT estimates how much a patient is likely to benefit from an adaptive treatment schedule compared with standard continuous therapy. Higher scores suggest that a patient’s cancer may be well suited to treatment breaks.
- The Expected Time to Progression (eTTP): This measure projects how long the cancer is likely to stay quiet before starting to grow again. It gives doctors, patients, and their families an early indication of what the treatment course might look like.
- The Expected Mean Daily Dose (eMDD): This measure estimates the average amount of medication a patient is expected to receive over time while following an adaptive therapy schedule. Because adaptive therapy includes planned treatment breaks, some patients may require less overall drug exposure than they would with continuous treatment, potentially reducing the burden of treatment.
Instead of requiring constant, complex updates or invasive tissue biopsies, the new Moffitt-led framework uses data from just the very first treatment cycle to probe and understand how a patient’s unique tumor evolves over time.
The team evaluated their approach using data from 53 prostate cancer patients enrolled in two independent clinical studies. Group one comprised 40 patients whose cancer was still highly responsive to standard hormone therapies. At this stage, reducing testosterone levels can often keep the disease under control.
The second group included 13 patients whose cancer had already spread to other parts of the body, even after standard hormone treatments had successfully lowered testosterone levels.
For both groups of patients, the AT Score was strongly associated with how long patients remained free from disease progression. When looking at the long-term survival of the advanced, drug-resistant group, patients with a high AT Score or a longer expected timeline (eTTP) lived longer.
In contrast, standard tracking methods, such as looking at the lowest point a patient’s PSA blood test has dropped to, how fast it fell, or how quickly it doubled, showed far less ability to predict patient outcomes.
Although the findings will need to be validated in larger prospective studies, the researchers believe their method could eventually serve as a decision-support tool to help doctors match patients to the treatment strategy most likely to benefit them.
Q&A with Kit Gallagher, Ph.D., first author and Alexander R. Anderson, Ph.D., Joint Senior Author, Chair, Integrated Mathematical Oncology Department, Moffitt Cancer Center
Your study suggests that information from the first treatment cycle can predict outcomes months or even years later. What does that tell us about how quickly a tumor reveals its evolutionary behavior?
We believe that these clinical outcomes are often predetermined near the start of treatment. If we can identify the behavior of cancer cells that exist at the start of treatment, we can forecast into the future to predict how they will ultimately respond.
Cancer centers already collect PSA data routinely. What would be required to integrate your biomarkers into everyday clinical practice?
The biomarkers that were present here could be computed automatically by decision support software, enabling them to be generated and used by clinicians without requiring specific mathematical expertise.
Current genetic tests are effective at showing a tumor’s mutations, but they can be invasive and expensive to repeat. How can your non-invasive measures work alongside existing genetic tests to give doctors a better, real-time picture of a patient’s cancer?
Genetic tests can give a picture of the tumor at the time of the test, which is typically conducted before treatment, but they are less able to predict future trends in the cancer. These measures supplement genetic tests by providing ongoing and updated predictions of tumor behavior throughout treatment.
The study mentions that some patients are highly sensitive to unexpected disruptions in their treatment, such as illness or drug shortages. Can the biomarkers identify these vulnerable patients early enough for doctors to adjust monitoring or treatment schedules?
This is an area of ongoing research. We hope to be able to identify the patients who are most at risk here in advance, to allow doctors to adapt their care.
Do you think similar measures could eventually be developed for other cancers?
For cancers that have accessible and non-invasive tumor-monitoring metrics equivalent to PSA in prostate cancer,such as CA125 in ovarian cancer, we are working on adapting these models to generate relevant predictive tools to improve patient care.