In June 2020, The University of Texas MD Anderson Cancer Center and Royal Philips announced a collaboration to provide oncologists with biomarker-driven, evidence-based therapy and clinical trial guidance through Philips’ oncology informatics solutions and MD Anderson’s Precision Oncology Decision Support (PODS) system. Through this collaboration, physicians around the world will be able to personalize cancer care based on a patient’s genomic profile and better match patients to clinical trials.
Journal of Clinical Pathways spoke with Kenna Shaw, PhD, cancer genomics laboratory, MD Anderson, and Louis Culot, general manager, genomics and oncology informatics, Philips, to better understand the PODS system and how this collaboration will improve patient care worldwide.
Can you tell me about MD Anderson’s PODS system? How do your physicians interact with this system, and what does it offer them that they did not have before?
Mrs Shaw: Our clinical molecular testing allows physicians to order tests depending on a patient's tumor type. Physicians can order mutation or fusion testing based on a list of approved tumor types and genes. Those orders go to our clinical laboratory and we receive a report back on which mutations have been found. If a mutation is found, a red circle goes around the gene that has a mutation or alteration.
The specific alterations are then listed on a separate page. If it's a BRAF V600E mutation, BRAF is circled on the first page, and the specific amino acid alteration is listed on a second or a third page.
We looked at patterns of ordering sequence data followed by patterns of utilization of that sequencing data to put patients on clinical trials. Before PODS, we found there was very low utilization of next-generation sequencing (NGS) data in terms of people who ordered the test but did not use the data to change care decisions or enroll patients in clinical trials. After investigating why this was the case, we found that physicians often did not know if there were clinical trials available for indicated mutations. In other words, they did not know if the mutation was actionable.
PODS was created and designed to address this issue – to provide our clinicians with information in real time on mutations, whether they are actionable, and available clinical trials for those that are actionable.
Why did Philips enter into this collaboration with MD Anderson? What is Philips bringing to the table?
Mr Culot: Philips is a health technology company. In terms of what we bring in scope are strong capabilities on a technical side, and then the ability to deliver solutions with global scale. We are a big company with a presence in basically every market in the world that provides health care.
MD Anderson is one of the leading cancer centers in the world. They have a broad reach themselves, but they are limited in terms of being able to carry out the mission on a global basis. So our partnership brings together a commercial and technical entity (Philips) with a clinical, academic, and research entity (MD Anderson).
In most of the US and the world-over, doctors and patients do not have access to the decision‑making resources and information that an institution like MD Anderson provides. Philips’ oncology informatics solutions allow for further virtualization of the MD Anderson PODS system; we take the hard work and collective wisdom on the part of Dr Shaw and her colleagues and make it more broadly accessible.
How are Philips’ oncology informatics solutions designed for analyzing genomic data and scaling large quantities of data into a report that is easy to read and actionable?
Mr Culot: From our view, the role of precision diagnosis is the integration and appropriate use of all the available information we have around a patient, including the genomic data. The genomic information provides another lens as to what is happening with a patient. In particular, it sheds light on what is happening with tumor biology, but there are other molecular results that matter; there are cytogenetic features and information from imaging analyses and tissue pathology that are important.
When you bring all of this information together, you get a much more comprehensive view of the patient that can feed into treatment decisions.
Historically, there has been a bit of a disconnect between genomics and the rest of the diagnostic modalities. When you look at what we do in imaging informatics, the scale is mind‑numbing in the terms of the amount of information that we are able to capture, orchestrate, and deliver to radiologists, oncologist, and pathologists in their care.
As a technical challenge, we believe we are well‑suited for it. But things become tricky when deciding from a fundamental standpoint whether the information is used in a complementary or deterministic fashion. You might have a lung cancer patient with a driver mutation, and also a high PD-L1 score, and so multiple options are available. We think this is where you need to bring the information together, surface to potential for discordance, and then rely on a doctor to adjudicate that.
We can break the decision-making process into two core challenges: arriving at a correct initial diagnosis and, somewhat separately, arriving at the best next-step or options for a patient. The system itself cannot adjudicate a discordant result between different diagnostics, but we can help surface them to the clinicians. Following the clinical judgement, which includes but certainly extends beyond biomarker status, this partnership allows doctors to explore clinical trial options for all patients, and options for patients who have exhausted standard-of-care and could access new therapies through philanthropy or compassionate use.
How does MD Anderson hope to leverage the broader exposure and use of the PODS system to aid in continued and overall better refinement of treatment and clinical trial suggestions?
Mrs Shaw: As more groups utilize the PODS knowledge base, one of the requirements will be a more complete and detailed annotation. Not only do we currently annotate specific molecular alterations, we also annotate every clinical trial that is relevant to a patient population. To date, that patient population has been primarily at MD Anderson, which means our focus has been on MD Anderson clinical trials.
On clinicaltrials.gov, the trials are not annotated for their molecular relevance. Some trials require a specific biomarker (eg, BRAF‑V600E positive and activating EGFR mutation). There is also a different class of clinical trials we call “biomarker relevant.” For example, a trial could be evaluating the use of vemurafenib, but without a specific biomarker needed for a patient to be eligible. The trial does not necessarily require the presence of a BRAF mutation. In fact, the trial may not even mention BRAF. We annotate specific clinical trials for those connections based on the drugs that are available and being used in the trial.
This level of detailed annotation is a bit unique at this point for clinical trials and not available for public domain. As more groups utilize the knowledge base, we plan to annotate more clinical trials.
We often see that precision medicine is considered second priority to drug matching on clinical trials at the moment. In other words, trials are more often designed to evaluate the effectiveness of a given drug and less often designed to evaluate the outcomes of patients with a given mutation.
Historically as a field, we have not necessarily been consistently rigorous with optimal clinical trial design. For example, we might put a given patient on a matched drug, but the patient may have a variant that is of unknown significance. We do not know for certain that the variant is responsive to a specific targeted agent.
Thus, the time is now to utilize decision-support tools to focus on these smaller populations of patients and match them with the best clinical trials available.