New approaches to clinical trials can lead to more successful results in the challenging near future. Parallel, comparative studies and a better selection of patients are just some positive developments in the generation of useful evidence for prostate cancer treatment, according to Dr. Laurence Collette (Brussels, BE).
Dr. Collette spoke as part of the Friday morning session on prostate cancer at the 10th European Multidisciplinary Congress on Urological Cancers (EMUC18) in Amsterdam. She offered a biostatistician’s perspective in the two-hour session that also included pathologists, urologists, radiation and medical oncologists and research scientists. Dr. Collette is the senior manager in charge of statistics and independent data monitoring at the European Organisation for Research and Treatment of Cancer (EORTC).
Good evidence in challenging times
Tasked with offering a view on evidence gathering in 2025, Dr. Collette spoke of trials taking place in a “fast-changing, high-tech and economically challenging environment.” She pointed out five major themes: which drugs work (best), which sequence, the impact of modern imaging, new technologies and the emergence of big data.
On drug trials: “Typically, trials end up adding to existing treatments. Their controlling arm is usually placebo, and the trial is focused on rapid authorization before the long-term information is available.” Collette cited the PROSPER trial, praising its results. Nevertheless, she considers it striking that information about the patient’s quality of life is only collected until the time of progression, without any information available for a complete assessment of the treatment.
“Trials should be designed not just to achieve clearance for the drug or treatment in question, but ultimately to inform healthcare systems on which treatment is effective for the patient and worth the investment. We need to move on from drug-centered to patient-centered trials.”
In order to determine which drugs work best, parallel and comparative studies are needed. Collette quoted the STAMPEDE trial as a good example of a flexible, adaptive trial that compares different treatment options in different arms that could be added or stopped early. “This leads to practice-changing results in a much shorter time than sequential trials. This requires pharmaceutical companies to work together, and here academia can play a role.”
With regards to new technologies: “There is a danger of a catch-22 situation: people need to buy the technology to test it, but are compelled to use it once the investment has been made.”
The application of big data in cancer research has its potential in the near future, as well as some challenges, according to Prof. Collette: “Big data in oncology is obviously not on the scale of Google’s use of data, or social media. It can also not be monitored on a day-by-day basis, unlike conditions that can be monitored through wearable devices like blood pressure or heart rate.”
“Furthermore, there is the challenge of new laws. We are currently trying to negotiate an exception to the GDPR rules that impact our trials by preventing us from sharing data or in some cases even invalidating existing data.”
In summary, for generating good evidence in the coming decade: “Patient-centric trials are the future, as are multi-stakeholder collaborations with more than one company. Effective digitalization of our results will lead to bigger data for researchers.”
Predictive statistics
The morning continued with a session on controversies and contradictions in the staging of prostate cancer. This mainly took the form of an hour-long multidisciplinary case discussion concerning a 69 year-old man with hypertension and a PSA of 14.7, a family history of PCa but no suspicious lesions after a DRE was performed. Sixteen cores were taken in a random TRUS biopsy. A panel of six experts debated the course of action.
Preceding the case discussion was a presentation by Dr. Daniel Sjoberg (New York City, USA), biostatistician of Memorial Sloan Kettering Cancer Center who in the past also worked to develop the 4Kscore.
“Recently, there has been a huge increase in modern markers for prostate cancer,” Dr. Sjoberg began. “We are starting to move beyond the ROC curve, which is insufficient for making a useful clinical prediction. We need make a more pragmatic assessment of these markers.”
Dr. Sjoberg lined out three qualities that make a good biomarker: discrimination, calibration and ultimately clinical utility.
