If you know someone with a medical crystal ball who can predict a prostate cancer patient’s outcome with 100% certainty, I’d like to hire that person to help my PCa patients gain the optimum treatment plan. Until that day, I continually seek top technology advances I can use in the service of my patients.

Increasingly, I find them in Artificial Intelligence (AI) platforms designed for clinical practice. Are you aware of how we rely on AI every day? Do you use a rideshare app like Uber or Lyft? Do your phone texts “autocorrect” your spelling? Do pop-up ads reflect what you buy? There it is, AI at work!

What if, when you’re diagnosed with PCa, you could immediately harness a vast, lightning-quick tool to calculate your prognosis, meaning the most likely outcome, tailored to your individual case?

Before a PCa patient and doctor decide on a treatment, they need to know if there’s a hidden pitfall called adverse pathology (AP). You might think, with all the tests, probing and scans leading to the diagnosis, nothing would have been missed. However, roughly 25% of patients who have a radical prostatectomy (RP) experience return of their cancer within 5-15 years; with hindsight, this recurrence is often linked with unidentified pre-surgery AP. Thankfully, most patients with no AP are deemed “cured” if they have zero PSA, called biochemical recurrence free survival (bRFS), for the rest of their lives.

Merging statistics, imaging, pathology and Artificial Intelligence

Since we don’t have a crystal ball, I’m excited to share a pioneering new probability tool based on imaging and clinical factors. I will be using the following terms to describe it:

• Nomogram – As I wrote in an earlier blog, a nomogram is a set of scales used to calculate an unknown value. When adapted for medicine, it’s a statistical modeling tool. If you ever used an online prostate calculator, you experienced a nomogram used to gauge probability. You enter your key clinical factors (age, clinical stage, Gleason grade and PSA) and it gives the odds that your cancer has spread outside the gland at the time of treatment. Technically, “…within the medical field, a nomogram is a graphic calculating device, a two-dimensional diagram designed to allow the approximate graphic computation of the likelihood of a clinical event.”^[i]
• Radiomics – A relatively new field that marries medical images such as MRI or CT scans with Artificial Intelligence (AI). AI uses computational algorithms applied to a “training” data set, so it “learns” to recognize patterns. It can then make predictions about new incoming information, from which it also learns more. When used in medical imaging, radiomics is a quantitative way to harvest imaging features that the naked eye can’t perceive, and mathematically extract image texture features like pixels or intensities. In turn, they are analyzed according to the algorithms. Radiomics can thus shape key clinical decision-making by utilizing nomogramic probabilities.
• Clinicopathologic – This refers to the characteristics (clinical factors) of actual tissue samples (pathology) following a prostate biopsy or RP. The physical PCa samples are sent to a specialist called a pathologist for analysis as well as a type of molecular analysis called genomics if needed. Just as AI can be trained on image features, it can also be trained on clinicopathology.

A new prognostic imaging-based nomogram called RadClip

A personal colleague from Case Western Reserve University, Anant Madabhushi, and an international research/development program developed an integrated radiomic-clinicopathologic nomogram called RadClip to predict the probability of adverse pathology and post-prostatectomy recurrence in PCa patients.^[ii] Their work builds on existing research that “radiomic derived features have shown to capture of sub-visual texture patterns for quantitative characterization of tumors phenotypes, and help in PCa risk stratification.”^[iii] For their study, they retrospectively analyzed 198 prostatectomy cases in which a pre-operative MRI had been done and which had average 35-month follow-up.

The team used image features from each of the 198 cases as the data for the AI training set. To bolster risk stratification, the team also correlated into AI recognition the respective pre-RP clinical factors for each case with its image features. Since all case records included RP outcomes (AP, bRFS), the team was able to create a radiomic risk score (RadS) and the RadClip nomogram.

Using the 198 cases, they tested the probability power of RadClip against two well-established PCa risk predictor nomograms, Decipher and the Prostate Cancer Risk Assessment (CAPRA). As the authors write, “Following an independent, multisite validation, we found that pre-operative RadClip was more prognostic of BCR [biochemical recurrence] and AP compared to Decipher and CAPRA for PCa patients who had undergone radical prostatectomy.”^[iv]

AI at the Sperling Prostate Center

Although I was not personally involved in the above endeavor, Dr. Madabhushi and I have been research collaborators over the past 10 years. I am a co-author with him and others in four published studies on multiple AI and radiomics platforms, and am proud of our pioneering work.

As I stay abreast of medical/clinical advances in AI, I integrate them into our practice here at the Sperling Prostate Center. Our MRI guidance system has the benefit of AI-assisted recognition, and RadClip is just one of the radiomics resources for predicting AP and the probability of BCR based on in-bore MRI-guided targeted biopsy results. Finally, by ruling out AP and probable recurrence, we are able to identify and support PCa patients who are candidates for Active Surveillance. We are confident that by incorporating RadClip and other technology/AI breakthroughs, we offer our patients superior diagnostic and treatment resources in our one-of-a-kind center of PCa excellence.

NOTE: This content is solely for purposes of information and does not substitute for diagnostic or medical advice. Talk to your doctor if you are experiencing pelvic pain, or have any other health concerns or questions of a personal medical nature.

References

[i] Bianco, FJ. Nomograms and medicine. Eur J Urol. 2006;50:884-6.
[ii] Li L, Shiradkar R, Leo P, Algohary A et al. A novel imaging based Nomogram for predicting post-surgical biochemical recurrence and adverse pathology of prostate cancer from pre-operative bi-parametric MRI. EBioMedicine. 2021 Jan;63:103163.
[iii] Ibid.
[iv] Ibid.