Cancer: A Multidimensional Challenge
“The finished house never completely resembles the initial blueprint,” is how Dr. Timothy Triche begins his explanation of working with DNA to diagnose childhood cancers. What we call cancer is much more complex than many realize, and having genetic information informs only so much about this “constantly evolving and changing tableau of sequential and coordinated events.”
Dr. Triche is co-director for the Center of Personalized Medicine at Children’s Hospital Los Angeles. He believes that our understanding of cancer is still largely one-dimensional. Oncologists can now identify specific genetic mutations, but for many cancers that may tell only part of the story.
Seeing cancer in multiple dimensions
A diagnosis of cancer is frustratingly nonspecific. What we observe as “lung cancer,” for example, is more precisely the result of highly complex DNA, RNA and protein interactions. And there’s more: Dr. Triche is uniquely focused on non-coding RNAs, the part of the genome that does not produce proteins yet contains powerful biomarkers that may predict what can happen to patients and may suggest therapeutic targets. “Understanding how DNA, coding genes, non-coding RNA and proteins really work in the real world in three-dimensional space over time is the key,” says Dr. Triche.
“Looking at genomic data alone is not three-dimensional,” cautions Dr. Triche. In his view, precision medicine involves going beyond what we can observe with the patient and basic lab results, adding new dimensions of data to our current understanding. “What’s most conspicuous today is genomic data. But the advocates for proteomics will tell you there’s another missing dimension. So too will those who specialize in imaging,” Dr. Triche adds. “Ultimately, cancer functions at the single cell level. Grinding up tumor tissue, extracting the DNA, and looking for mutations provides important information but may not provide all the information we need.”
This all leads to a fundamental difference in how we approach patients: Each person and their cancer is unique, and now we have evidence that supports treating them differently, sometimes with a completely personalized approach.
And this takes us back to the lung cancer example. Until recently, this was an anatomic diagnosis, meaning you have a malignancy involving your lungs. But what do we really know from that information? Almost nothing, says Dr. Triche.
Today, in the age of precision medicine, we don’t simply schedule surgery and remove the lung tumor; instead, Dr. Triche says, we attempt to understand which of the many subtypes of lung cancer you have. That answer could, for example, determine that surgery will be ineffective and that certain drugs might work better. This is the first step in delivering more precise, personalized therapy.
Tackling pediatric cancer
With news of new therapies such as Keytruda and Kymriah, the cancer community is abuzz. But pediatric cancer is different warns Dr. Triche. “We don’t see the same types of tumors and mutations and, therefore, most of our patients aren’t eligible for these new and notable drugs.”
The understanding that pediatric cancer is unique has led physicians such as Dr. Triche to look for different approaches. “The well-publicized NCI-MATCH panel initially targets adult cancers only. What we need, however, is a panel that covers all the common types of childhood cancer, which, believe me, are very different from adult cancers,” says Dr. Triche. “Developing targeted gene panels based on next-generation sequencing technology, which look at the genomic alterations specific to a child’s tumor, are one way we can get there.”
Dr. Triche and other pediatric oncologists believe that targeted panels can eventually be used broadly to predict a better or worse prognosis for a young patient, leading to more (or less) aggressive therapies, or a completely different therapy if necessary. For example, when faced with a diagnosis of aggressive childhood neuroblastoma, a tumor with, for example, MYCN gene amplification that carries a dire prognosis, an oncologist may instead recommend a non-standard therapy for lack of better options. The known uniform failure to respond to conventional therapy may make the patient eligible for a phase two trial for an experimental therapy. Dr. Triche points to the benefits of choosing this pathway earlier in the process instead of using standard risk therapy, which is ineffective in such patients.
Is every disease an orphan disease?
Unlike adult cancers, there are far fewer cases of pediatric cancer, especially when each type is considered. Also, the incidence of “targetable mutations” with matched therapy is far lower in childhood cancer. Consequently, it’s difficult to enroll enough patients on a clinical trial intended to match a diagnostic to a companion therapy. Dr. Triche points out that this is a challenge for “orphan diseases” overall. In fact, the more we understand about the unique genetic character of each patient’s disease, the more difficult it is to treat by diagnosis, as opposed to targetable genetic defect. We are likely entering a phase in cancer management where treatment will be based on gene defect as opposed to a simple diagnosis like “leukemia” or “brain tumor.”
Dr. Triche argues that precision medicine is moving toward a point where diagnoses are defined in terms so narrow that a single therapy may ultimately apply to far fewer patients. “With what we know today, we may infer that a single treatment is appropriate for a large cohort, but in the future – with more data – we may move to a completely different therapeutic approach for subsets of patients in that cohort with different genetic defects in their tumor.” What this means for orphan diseases, which affect fewer than 200,000 people per year, is intriguing: presumably many more diseases will one day qualify as orphan. The diagnosis and treatment principles developed for childhood cancer may then prove useful for treatment of other orphan disease, particularly given that the FDA is now publicly more open to fast-tracking orphan drug approvals.
The role of data interpretation
The complex interactions that comprise an individual’s molecular profile generate massive amounts of data. This points to the challenge of interpretation. Dr. Triche believes this speaks to an expanded role for “molecular” pathologists, or those trained in both pathology and genetics. He also speculates that this will change oncologists’ roles too. “Few clinicians involved in patient care are trained or have the time to interpret raw genetic data,” he says. “Others with training in both medicine and molecular genetics will likely be tasked with interpreting the raw data and presenting it in a clinically useful form to clinicians, who are primarily concerned with how best to treat their patient.”
Dr. Triche believes the profession is still learning and changes are certain. In fact, we could ultimately see impacts on medical licensure and training. This, Dr. Triche believes, will lead to fundamental changes in the practice of medicine, starting in medical school and eventually leading to focused areas of expertise and board certification. In the interim, however, molecular pathology and genetics experts are working alongside clinicians to manage the process from generation of patient genetic data to potential clinical use.
“We’re still in the early days of understanding what a gene defect means in terms of disease onset, progression, prognosis, outcome and treatment response, and we’re all still learning. To pretend otherwise is naïve,” says Dr. Triche.
Part of navigating the new precision medicine landscape is managing patient expectations. What, for example, should a physician do when a patient argues for treatment based on a genetic test, such as pre-disposition to heart disease or cancer, especially when we know these diseases are multi-factorial and no single assay can predict whether the disease will occur? Is there enough data to pursue a costly treatment for a disease that is yet to present? And who is to make that call? For many, the choice is binary: Is this an actionable or non-actionable mutation? For the former, a course of therapy is pursued, but for the latter the mutation can be documented so the patient could be matched at some point in the future with a therapy should the disease in question one day occur.
Where are the gaps?
One gap Dr. Triche sees as we increasingly rely on panels for diagnosing cancer patients is seeing the full molecular picture, which should, he believes, consider RNA too. Fusion genes, which are common in pediatric cancer but less so in adult cancer, produce abnormal proteins that drive tumor formation and are a diagnostic and prognostic hallmark of childhood cancer, but they are difficult to detect at the DNA level. They are, however, easily detected at the RNA level
A similar situation exists with normal genes, which can be expressed in many variant forms, due to variation in the way the pieces of the gene are spliced together to make the messenger RNA that specifies the resulting protein. In cancer cells, many abnormal variants are made, resulting in many abnormal proteins. Dr. Triche believes that future protein-based assays must enable detection of these thousands of abnormal proteins. He’s confident that we’ll ultimately be able to fill all the gaps between raw genetic data and protein expression to understand how these fusion genes and other cancer genes work. “Right now, however, we’re largely ignorant of many things in between,” he says.
Dr. Triche believes that the “really interesting” interplay is between non-coding RNAs and proteins. “It has become so obvious that RNA interacting with protein is part of the master switch in your genome. Nothing happens in your DNA without both RNA and proteins. Those two, often working together, determine whether genes get expressed, whether they get turned off and whether you can even get to them,” says Dr. Triche. “Remember, life began as non-coding RNA; protein encoding genes and DNA came later. Not surprisingly, many of the most basic functions in a cell, like the ribosome that turns message into protein, is at heart an RNA machine.”
These “master switches” are incredibly important. And Dr. Triche believes that one of the most important emerging areas in genomics is the integration of knowledge about proteins and the regulatory RNA with which they work to control the genome. In other words, looking at DNA sequence is like looking at the blueprint of a house and predicting what the house will look like. “And, remember, all contractors make changes to the blueprints while the house is being built,” Dr. Triche asserts.
Toward functional genomics
Dr. Triche says we’re still overly reliant on DNA sequence-based, “linear” genomics, but that is changing. The next destination is functional genomics, where we understand how DNA, RNA and proteins interact and function with one another in three dimensions within the cell over time. And we need more advanced analytical tools to support this. “Soon, we’ll be able to understand whether key proteins and their regulatory DNA and RNAs are normal as well as who they are working with and where,” says Dr. Triche. “Then you begin to understand key players and pathways in a holistic context that cannot be discerned from DNA sequences alone.”
Functional genomics puts us on a pathway toward understanding why certain parts of the genome would be “open for business” and genes there would be expressed or not. Today this is still largely a mystery, though there are clues: childhood cancer is often described as “embryonal” in appearance, and as it does appear it often recapitulates embryonal, developmental pathways. Learning how those normal developmental pathways are hijacked and misused by cancer cells may provide insight into future targeted therapy. But, for now, we’re at least closer to understanding how to turn expression on and off in an organized, logical way that leads to a healthy cell instead of a malignant one. “And then we’ll be closer to understanding the difference between effective and ineffective therapy based on actual mechanistic evidence,” says Dr. Triche.
Rather than inferring that a therapeutic disruption within a gene pathway for, say, cell growth, will work, in the future we may be able to control the on/off switches to target key steps that enable the disease to progress, or even emerge in the first place. We’re obviously in the earliest stages of this, but work by many researchers (see Targeting the Causes of Infant Brain Tumors above) is already showing promising results. This is one of many reasons that Dr. Triche believes we’re closer than ever to meeting the multidimensional challenge of cancer.