Curious Dr. George
Cancer Commons Editor in Chief George Lundberg, MD, is the face and curator of this invitation-only column

Genitourinary Oncologist and Researcher, Beth Israel Deaconess Medical Center; Instructor in Medicine, Harvard Medical School

Cancer patients often ask their doctors, “What would you do if you were me?” Here, our Curious Dr. George asks Harvard oncologist Jon Arnason, MD, how he would handle his own diagnosis of chronic lymphocytic leukemia (CLL).

Curious Dr. George:Imagine a hypothetical scenario in which, as an active clinical and research oncologist in general good health, you have lately felt less energetic than usual, have been feeling fatigued at night, and have lost some weight without dieting. You decide to visit your primary care physician for a check-up. On physical examination, she notes that you are a little pale, discovers moderately enlarged lymph nodes in the axilla and groin, and a slightly enlarged liver and spleen. She orders several lab tests. Your hematocrit is 38, hemoglobin 12, WBCs 90 000 with mostly mature lymphocytes, a proper number of granulocytes, and normal platelets. A provisional diagnosis of chronic lymphocytic leukemia (CLL) is made. How would you proceed?

Jon Arnason, MD: CLL is the most common kind of leukemia. It is a diagnosis that is often made in situations similar to the hypothetical one you present above. In addition, many patients present incidentally when laboratory testing is performed and demonstrates an incidentally elevated white blood cell count. Upon suspicion of CLL it is important to confirm the diagnosis with peripheral blood flow cytometry.

A new diagnosis of CLL can be quite distressing for patients. However, it is important to determine whether the patient is actually suffering pathology from their disease. A large percentage of patients who present with CLL will not have any active symptoms and may go for an extended period time, sometimes many years, before developing any symptoms that are attributable to their disease. In addition, a subset of CLL patients may never develop harm related to their cancer. Therefore, we do not recommend initiation of treatment if patients are asymptomatic from their CLL. Signs and symptoms that would suggest that patients require treatment include significant anemia, low platelet count, enlarged lymph nodes that are causing or about to cause symptoms, an enlarged spleen, or other systemic symptoms such as fever and weight loss that may be attributed to their disease. In addition, if the patient’s CLL-related white blood cell count doubles in less than 6 months, this suggests that the patient will imminently have symptoms and would benefit from treatment.

In the absence of the above signs or symptoms, we recommend a strategy of active surveillance. The goal of active surveillance is to identify when the patient will be about to develop symptoms and discuss treatment options at that time. At diagnosis there are number of tests that we can perform to help us predict whether a patient will have a relatively rapid progression or a more indolent course. Perhaps the most important testing to perform at diagnosis is a cytogenetic analysis. There are a number of recurrent cytogenetic abnormalities that we see with CLL. Deletion 13q is associated with a relatively good prognosis. Trisomy 12 is associated with an intermediate prognosis. Deletion 11q abnormalities are associated with large lymph nodes and a relatively more rapid progression. Finally, 17p abnormalities are associated with a poor prognosis and lack of response to traditional chemotherapeutic strategies. In addition to cytogenetic abnormalities, there is a test called IGVH mutational status. Mutations of IGVH are associated with a better prognosis than the absence of mutation.

Traditionally, CLL was treated with a combination of chemotherapeutic agents and antibody therapy. However, recent clinical studies have demonstrated that targeted therapy with oral agents and antibody treatments are more effective with less toxicity than traditional chemotherapeutic strategies. Current guidelines suggest the use of a Bruton’s tyrosine kinase inhibitor, such as acalabrutinib or zanubrutinib for continuous therapy. Alternatives include the use of the Bcl-2 inhibitor venetoclax with CD20 antibodies, potentially in addition to the use of Bruton’s tyrosine kinase inhibitors. The majority of patients will have a response to these treatments and improvement in any underlying CLL-related symptoms.

If I was diagnosed with the above findings and labs, I would not want to receive therapy. As an aging, busy clinician I am always fatigued at night, have less energy than I used to, and struggle to maintain my weight as I’m running around between clinic patients and the inpatient unit. The current data suggests that I’m not going to live any longer if I treat my CLL at diagnosis compared to at the time that I have clear symptoms. Furthermore, treatment options have significantly improved during my career, and I am hopeful that there is the potential for long term remission with some of our emerging therapies, such as CAR-T cells.

Dr. Arnason can be reached at jarnason@bidmc.harvard.edu.

***

Copyright: This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Curious Dr. George
Cancer Commons Editor in Chief George Lundberg, MD, is the face and curator of this invitation-only column

Michael Kramer
Director of Technology, Cancer Commons

Artificial intelligence (AI) holds promise to transform cancer research and treatment. Here, Curious Dr. George asks our very own Director of Technology, Michael Kramer, about the powerful role Cancer Commons can play in shaping the future of AI in oncology. Having lost his son, Theo, to a rare form of brain cancer—Diffuse Intrinsic Pontine Glioma (DIPG)—Kramer is deeply committed to helping others facing similar struggles.

Michael Kramer: This year’s PMWC in Silicon Valley was a tremendous experience. It was my first time attending the conference—and I couldn’t have picked a better year. The energy was palpable. Jensen Huang, founder of NVIDIA and arguably the “Steve Jobs of the AI computing era,” gave the keynote address. His presence was symbolic: a torch has been lit. AI is no longer on the horizon. It’s here.

But if this moment sparked hope, it also exposed the deep fissures between promise and practice. I came away inspired—but also sobered. Healthcare, especially oncology, stands at a crossroads. And while AI might help us move faster, more accurately, and more personally, it will only thrive if we raise it with care—giving it the values, structure, and support systems it needs to grow into something truly beneficial.

As someone who came to this work not as a scientist or physician, but as a grieving father and a systems thinker, I’ve seen this story before—in another domain. I spent nearly two decades in the arts as a film and television composer, often building new technology to help me tell stories through music. I once believed that creativity—the essence of our humanity—would be the last frontier for AI. Surely art was safe.

I was wrong.

AI came for art first. Not because it understood the soul of the artist, but because it could mimic the product. Because in the commercial world, humanity is often less valued than efficiency. And now, healthcare is staring down the same paradox: how do we use a dehumanizing technology to become more human?

Four Key Lessons from PMWC 2025
Over the course of three days, one truth echoed across panels and disciplines: most of medicine is not ready for AI. But that’s not a reason to be cynical—it’s an invitation to get it right. And Cancer Commons is uniquely positioned to do just that.

Here are four critical insights from the conference:

1. Patient-Centric, Multimodal, Real-Time = the Future
From real-time clinical trial matching tools to AI systems that synthesize genomics, imaging, and electronic health record (EHR) data on the fly, the trend is clear: the patient must be at the center. Not data. Not the institution. Not even the technology.

Cancer Commons has always operated from this premise. We don’t provide direct care, but we educate and empower patients, families, and even the clinicians working tirelessly on their behalf. We listen, adapt, and act in real time, curating information and matching people with possibilities. In this sense, AI isn’t a disruption to our model—it’s a powerful ally. We’re modeling how to build the kind of responsive, patient-informed approach that will define the next era of medicine.

2. A Tale of Two Cultures: Big Data vs N=1
Precision medicine lives in a paradox. On one side, multi-million-dollar AI models trained on millions of data points. On the other, a child with DIPG brain cancer—a disease so rare it is invisible to those very datasets.

Most AI systems don’t know what to do with outliers. But Cancer Commons was built for outliers. Our strength lies in combining AI with “HI”—human intelligence: expert curation, personalized insight, and support for patients when the standard of care has failed them. Our small size is not a weakness; it’s a strategic advantage in a world where N=1 matters more than ever.

3. Hope Through Shared Intelligence
One of the most exciting threads at PMWC was the emergence of federated learning, synthetic data, and collaborative AI training. Again and again, we heard: no one has enough data and resources alone. Collaboration isn’t just nice—it’s necessary.

And yet, large institutions often find themselves paralyzed by competition, legal silos, and outdated business models. Cancer Commons can move nimbly where others can’t. We’re small. Agnostic. Mission-driven. And we’re building partnerships—like our collaboration with xCures—that can help overcome one of the field’s greatest obstacles: the inability to share.

4. Living on the Edge: Redefining Risk in a System That Plays It Safe
Many of the people who come to Cancer Commons are already living on the edge—medically, emotionally, and existentially. They’re facing advanced cancers, rare diagnoses, or systemic barriers that have pushed them beyond the reach of standard care. For them, risk isn’t hypothetical—it’s part of daily life.

And yet, many medical institutions are built to avoid risk, not engage with it. I saw this firsthand during my son Theo’s treatment. One day in the hospital, I found myself in a tense exchange with an anesthesiologist who hesitated to proceed because Theo had a do-not-resuscitate (DNR) order. For them, it meant “don’t engage.” For us, it meant “don’t prolong suffering”—but we were still fighting, still seeking options that balanced quality and time. Hope and realism were never in conflict. They lived side by side.

This willingness to act within uncertainty is something our clients understand deeply. It also shapes how we think about innovation.

At PMWC, Andrew Rister from Verily offered a striking metaphor: when cars first appeared, they had to share roads with horses. It was chaotic and risky, but over time, new rules emerged. The system adapted. That early turbulence wasn’t failure—it was transformation.

The same is true now. Where others see emerging technologies like AI as too risky, our unique clientele not only affords us that risk—it compels us to take it.Not carelessly, but with urgency, integrity, and care.

AI can help us act wisely in the face of uncertainty, if we’re brave enough to guide it there.

Where We Fit In
If AI is still in its infancy, Cancer Commons is uniquely positioned to raise it with care, intention, and purpose. Consider this:

• Most large institutions are not nimble enough to adopt AI. We’re small and adaptive.
• Competition prevents data sharing. We can be a neutral, open-source bridge.
• Most AI models are trained on flawed or inconsistent data. Our hand-curated patient data is small but clean.
• Protected Health Information (PHI) regulations prevent data flow. Our collaboration with partners like xCures may become a bedrock solution.
• And most importantly: our patients don’t have time to wait. My son Theo didn’t either.

For me, this isn’t theoretical. It’s personal.

When Theo was diagnosed with DIPG, we didn’t have five years to wait for perfect systems or large-scale trials. We needed real help, real insight, real options—now. That urgency is still with me. It’s why I believe Cancer Commons is not just a nonprofit—it’s a blueprint for how we all might one day learn enough to truly outsmart cancer.

What Kind of Parents Will We Be?
AI has arrived, kicking and screaming. Like any powerful new force, it’s immature, chaotic, and full of promise. The question now isn’t just how we use it—but how we raise it.

We can neglect it—feed it biased data, isolate it in silos, train it on systems we already know are broken.

Or we can nurture it—guide it with ethics, context, humility, and care. Show it the best of humanity, not just the most efficient. Teach it, like a child, how to grow into wisdom—not just intelligence.

Because at the end of the day, AI will reflect us. And if we’re brave, patient, and generous enough, it might even help us become something better than we were.

Mr. Kramer can be reached at michael.kramer@cancercommons.org.

***

Copyright: This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Curious Dr. George
Cancer Commons Editor in Chief George Lundberg, MD, is the face and curator of this invitation-only column

Jerald P. Radich, MD, MA, MS
Professor, Translational Science & Therapeutics Division, Kurt Enslein Endowed Chair, Fred Hutchinson Cancer Center

When facing a frightening new cancer diagnosis, some people ask their doctors, “What would you do if you were me?” Here, our Curious Dr. George asks expert oncologist Jerald P. Radich MD, MA, MS, how he would handle his own hypothetical case of chronic myelogenous leukemia (CML). Dr. Radich is the Kurt Enslein Endowed Chair in the Translational Science & Therapeutics Division of the Fred Hutchinson Cancer Center in Seattle, Washington.

Curious Dr. George: Please consider this hypothetical scenario: you are an active clinical and research oncologist in general good health, but lately you have lacked energy, have been feeling more tired than usual, have some dizziness on exertion, recent easy bruisability, have lost some weight without dieting, and are experiencing sweating at night. You decide to visit your primary care physician for a checkup. On physical examination, she notes that you are a little pale, finds arm bruises, and discovers an enlarged liver and spleen. She orders several lab tests. Your hematocrit is 38, hemoglobin 12, WBCs 120 000 with mostly mature granulocytes, a few blasts, and a normal lymphocyte count. Platelets are normal. A provisional diagnosis of chronic myelogenous leukemia is made. When you learn of this hypothetical situation, how do you proceed?

Jerald P. Radich MA, MS, MD: How ironic that I have been diagnosed with CML. On the other hand, if one is going to get the disease that they have devoted decades of studying, I guess I chose wisely.

How so, you might ask (after all, you are curious)? Because in CML, we now have an embarrassment of riches, with five approved tyrosine kinase inhibitors (TKIs) for newly diagnosed, chronic phase CML cases (which I am taking editorial license to place myself in that category, rather than advanced-phase disease, thank you very much). Taking these medications daily places you on the same natural life expectancy curve as the non-CML population. It is truly a mind-blowing example of the “bench-to-beside” paradigm so often utilized in grant writing, but so rarely attained in real life.

The current list of TKIs for newly diagnosed, chronic phase CML includes the original, imatinib, and the “second generation” TKIs, bosutinib, dasatinib, and nilotinib. Ponatinib is approved by the U.S. Food and Drug Administration (FDA) for resistant chronic phase and advanced phase disease. All these inhibitors work to block the ATP-binding domain in the tyrosine kinase of the chimeric BCR-ABL protein. Recently, a new allosteric inhibitor known as asciminib was approved for newly diagnosed CML (should we dub this a third-generation compound?). It acts outside the ATP domain by causing a structural change in the protein that abrogates kinase activity.

These TKI drugs share various degrees of common toxicity (low counts, lipase increase, potential cardiovascular events) and all have drug-specific toxicities (e.g., pleural effusions in dasatinib, diarrhea with bosutinib). In general, the second-generation TKIs are more potent, leading to faster reductions of disease, and less evolution to advanced phase disease compared to imatinib. But at the end of the day, large phase-three studies have shown that all these drugs yield near identical survival.

How then would one choose? Here are the factors I would consider, in my own order of prioritization:

1. What are my goals of therapy? What do I want? First, to control the disease and put me on a normal lifespan trajectory. This requires a treatment strategy that allows for maximum compliance (the easier to take drug, the better), minimal toxicity (both preventing off-target injury and minimizing time off the TKI) and the prevention of progression to blast crisis (as noted above, the second-generation-and-beyond TKIs seem a little better than imatinib for blocking progression). If I was considerably younger, and especially if I wanted to have kids, then there would be the consideration of getting to a deeper molecular response faster, since there is ample data that roughly 25% of CML cases can successfully discontinue the drug after several years of a deep molecular response and stay in a treatment-free remission. Evidence suggests that patients taking the stronger, second-generation TKIs may have an advantage in getting a deeper molecular response, faster.
2. What’s my risk of progression? The major therapeutic mission is to block progression, since once that happens, the only real path to cure is through allogeneic transplantation, and that’s not easy for a guy my age. There are various risk scores that map well to risk of progression (Sokal, Eutos, Hasford). I didn’t get my blast count, but I am worried about the large spleen, weight loss, and night sweats—all suggest a disease on the pathway towards progression. Thus, I would be hedging my bets and leaning towards the second-generation TKIs, including asciminib.
3. Comorbidities and toxicity. TKIs are truly remarkable, since they are exceedingly effective, yet have relatively little toxicity compared to standard chemotherapeutic agents. A pervasive complication across all these drugs—except imatinib—is the shadow of cardiovascular events, including venous and arterial thrombosis. Hypertension is also a common problem. If I had some cardiovascular risks, I’d be leaning towards options that seem be on the lower side of the risk spectrum, such as imatinib, lower dose dasatinib, or asciminib.
4. Financial toxicity. These drugs are very expensive. If I had to pay for these out of pocket, they run more than $100,000 per year, and since I’d likely have to take them the rest of my life, the overall cost would be insane. Fortunately, I have good insurance. But if not, I could go generic. For example, Mark Cuban’s drug store lists prices for generic imatinib at less than 1% of branded imatinib! A generic of dasatinib has been launched, and some new formulations of dasatinib designed for better bioavailability have been approved, though it is unclear where they are going to weigh in on cost.

No matter what drug I start with, I would be diligent about receiving routine peripheral blood testing for BCR-ABL to monitor response. If I didn’t meet the milestones (as set out in standard guidelines from the National Comprehensive Cancer Network), there are plenty of options for switching to another TKI. In general, if I need to switch because of side effects, options are open to move across generations (i.e., from a second-generation to the first-generation imatinib). However, if I need to change TKI for resistance, changes should be made within a generation (from dasatinib to nilotinib), or escalating to the next generation (dasatinib to ponatinib or asciminib).

Dr. Radich can be reached at jradich@fredhutch.org.

Curious Dr. George
Cancer Commons Editor in Chief George Lundberg, MD, is the face and curator of this invitation-only column

Michael Schmainda, MBA
President and CEO, Imaging Biometrics, LLC

For people with glioblastoma brain tumors, it can be vital yet challenging to determine the precise size, location, and extent of the tumor. A company called Imaging Biometrics, LLC, has developed novel imaging strategies to address this issue. Here, our Curious Dr. George asks its president and CEO Michael Schmainda about these innovations.

Curious Dr. George: Determining the extent of an intracranial neoplasm can be as important as securing the diagnosis. This also applies to postoperative follow-up: Was the tumor fully or only partly extirpated? Is the tumor recurring or is the evolving positive imaging only locating organizing hematoma or scar tissue?

Michael, how is your company striving to help solve these vital clinical puzzles?

Michael Schmainda: You’ve highlighted a critical challenge in neuro-oncology: distinguishing between tumor progression and treatment effects like pseudo-progression or post-treatment radiation effect (PTRE) using imaging techniques. This distinction is vital for accurate diagnosis, treatment planning, and monitoring.

Contrast enhanced magnetic resonance imaging (CE-MRI) is the standard imaging method but has limitations in differentiating between tumor and PTRE. In fact, PTRE often mimics tumor on CE-MRI, further conflating the issue. This often necessitates invasive biopsies for confirmation, which can be problematic, especially in heterogeneous tumors like glioblastoma. Targeting a biopsy site that is not consistent with aggressive tumor can lead to suboptimal treatment plans.

Response assessment techniques for brain tumors, such as response assessment in neuro-oncology (RANO), rely on CE-MRI to measure tumor size, which can be misleading if PTRE is present, potentially resulting in overly aggressive treatments with severe side effects.

At Imaging Biometrics (IB), we have developed and validated an automated processing platform that overcomes these limitations. This advanced imaging platform uses two quantitative technologies. The first, called a Delta T1 map, helps delineate true regions of contrast enhancement. Delta T1 maps incorporate an exclusive image-intensity calibration step, removing variability and artifacts to objectively highlight the true contrast-enhancing region.

Within the enhancing region identified by the Delta T1 map, additional information is obtained using our IB Neuro product. This uses dynamic susceptibility contrast (DSC) MR perfusion to measure several parameters including relative cerebral blood volume (rCBV). As tumors form and grow, they need increased oxygen and nutrients, which are supplied via the blood. Before a tumor can be detected on standard imaging, IB Neuro can accurately measure this increase in rCBV. IB Neuro also incorporates a calibration step that standardizes the rCBV output (sRCBV) independent of how it was collected. This enables a direct comparison of sRCBV measurements across time, MR scanners and field strengths, and patients. Furthermore, the sRCBV values have been independently validated by multiple academic centers using spatially matched tissue samples to establish cutoff thresholds that differentiate tumor from PTRE.

Combining IB Delta T1 and IB Neuro enable the creation of fractional tumor burden (FTB) maps, providing a quantitative assessment of tumor progression and treatment response across time. The maps are also imported into surgical navigation systems to target biopsies, plan surgical resection and radiation and, more recently, to help plan laser interstitial thermal therapy (LITT).

Several recent and ongoing studies have demonstrated the clinical benefit of using FTB for treatment surveillance for both primary and metastatic brain tumors and for biopsy guidance such that the most aggressive part of the tumor is sampled. They also have been shown to result in the greatest inter-reader agreement and confidence for glioblastoma response assessment over other techniques to which they were compared.

I encourage interested readers to learn more by watching Imaging Biometrics’ video on getting more out of your MRI exam, as well as our webinars on Delta T1 and the clinical applications of perfusion MRI in neuro-oncology.

Michael Schmainda can be reached at mike@imagingbiometrics.com.

***

Copyright: This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.