The Right Chemo For Your DNA

What causes two patients with the same type of cancer to respond so differently to the same chemotherapy drug? For starters, no two people share the exact same cancer, genetically speaking. Cancer researchers today are using genomic sequencing to help explain why some drugs work for some patients and not for others, what DNA has to do with it. With genomic sequencing, we can better fight cancer by creating drugs that match portions of DNA sequences that make up individual patients’ cancer cells.

Bryan P. Schneider is a classically trained MD with a depth of knowledge in biomarkers of disease and drug toxicity. A little over a decade ago, Bryan began his journey as a faculty member at the Indiana University School of Medicine, where he started off seeing breast cancer patients and running a translational laboratory focused on precision medicine. He now works with IU’s Precision Health Initiative, where the goal is to find new therapies for cancers that have limited therapeutic options today. These include multiple myeloma, pediatric sarcoma and triple negative breast cancer.

During his training, Bryan worked with George W. Sledge, now Chair of Oncology at Stanford University, learning about clinical trial designs. He was also advised by the late David Flockhart, a leader in the field of personalized medicine and pharmacogenomics, the study of the role of the genome in drug response. Flockhart helped identify the metabolism of the drug tamoxifen. He helped discover that the effects of this cancer drug are almost completely inhibited in breast cancer patients who are also taking serotonin reuptake inhibitors (SSRIs).

SSRIs are often prescribed for women with breast cancer to treat depression or anxiety and to reduce hot flushes, a common side effect of taking the cancer drug tamoxifen. However, thanks to work by Flockhart we now know that the “risk of breast cancer related death is higher in women taking tamoxifen plus the selective serotonin reuptake inhibitor (SSRI) paroxetine” (BMJ 2010;340:c783).

Learn more: Pharmacogenomics of Tamoxifen in a Nutshell

In his 60’s, Flockhart tragically developed glioblastoma, a rare brain cancer. He was one of the first patients to be sequenced as part of the Bryan and Milan Radovich‘s Precision Genomics Program at IU, an experience Bryan recalls as “surreal”. Bryan has carried Flockhart’s legacy forward by studying the interaction of chemotherapy drugs with specific genes and genetic mutations.

Bryan went on to investigate drug toxicity and drug-drug interactions in the context of chemotherapies. His lab group at IU has now identified several markers that predict who will suffer more side effects from particular chemotherapy drugs.

“We’ve spent the last 10-15 years thinking about how genomics can make our therapies better,” Bryan said. “Some of the earlier work we did explored genetic changes that we inherit and how those might impact side effects of drugs.”

“Our research group has spent a substantial amount of time and effort looking at how minor inherited genetic differences can impact some of the devastating side effects of chemotherapy,” Bryan said.

The C-Word: Making Chemotherapy Less Scary

Many of us who have had friends or family members with cancer might agree that “chemotherapy” is often a scarier word than cancer itself. This is because of some of the devastating and potentially life-threatening side effects of chemotherapy drugs – side effects like congestive heart failure and irreversible peripheral neuropathy conditions, or nerve damage that can cause pain, weakness and numbness.

Over time, Bryan and colleagues at IU including Milan Radovich, Vice President for Oncology Genomics at Indiana University Health, have realized that genomics might be key to predicting drug side effects before they happen. If clinicians could learn which genetic variants are associated with side effects to particular chemotherapy drugs, they could help prevent these effects in the first place by prescribing alternative drugs to at-risk patients. This idea led to the IU Health Precision Genomics Program.

Bryan helped launch this genomics clinic at IU for patients with all cancers, including breast cancer. The clinic has been providing patients with real-world genomic sequencing in an effort to find the best medications for individual patients based on their unique genetic factors. Through data about patients’ genomes and other data about their symptoms, lifestyles, etc., Bryan has been working to identify genomic drug targets within cancer and markers of drug success. The program has sequenced over 2000 patients to date.

“We use comprehensive whole-genome sequencing. This is a really neat way to look at an abundance of drug targets,” Bryan said. “I think the most exciting thing about using whole-genomic sequencing with cancer patients is that it allows you to learn so much beyond the question of which drug works best. But we haven’t been able to fully harness this technology until now.”

This is where modern genomic sequencing technologies and LifeOmic’s Precision Health Cloud platform come in.

“These technologies will finally provide us the opportunity to look at an unprecedented depth of genomic information along with real-life outcomes in terms of how cancer patients do with particular therapies,” Bryan said. “We can now learn from both cases of devastating side effects as well as cases in which patients do exceptionally well.”

Using the Cloud to Find New Cancer Drug Targets

Many healthcare programs across the world now are starting to use genetic markers to help patients get the drugs that will work for them based on individual variation. Many clinics are using limited panel testing to look for genetic markers known to affect cancer outcomes or side effects associated with particular therapeutics. However, more robust whole genome sequencing approaches allow researchers in Bryan’s group to explore a wide variety of genetic markers that might be helping particular patients respond extremely well to particular drugs.

Being able to comb through whole genome sequences to look for genetic markers of drug success requires a talented team of interdisciplinary researchers, including oncologists and bioinformaticians. It also requires huge data storage and analysis capabilities. Machine learning and artificial intelligence technologies, such as those built into LifeOmic’s Precision Health Cloud platform, are also integral to finding trends, outliers and other associations between genetic markers and patient outcomes that clinicians might not know to look for.

With machine learning, “smart” systems can learn from patient data and uncover hidden patterns between unique genome sequences and patient outcomes. Clinicians will increasingly rely on machine learning to make better drug treatment decisions for individual patients as we uncover increasingly complex relationships between genetic markers and chemotherapy drugs effects.

There are three billion base pairs in a single patient’s genome. There exist in this genome innumerable genetic variants that may or may not matter to how a given patient responds to a cancer therapy. Determining which drugs work for which cancer patients is not as simple as playing “Go Fish” and finding a single biomarker best matched to a single drug. It involves investigating constellations of genetic biomarkers that might be associated with drug side effects or drug success.

“To date, it has been an incredibly difficult task for us to take a huge genome file, match it up to a single patient and then look for associated phenotypes, such as positive patient outcomes, survival or toxicity,” Bryan said. He and Milan are now using LifeOmic’s Precision Health Cloud (PHC) platform to power their search for biomarkers of drug success in cancer therapy. The PHC stores, integrates and analyzes any and all types of patient data. This includes clinical data, genomic data and data patients generate when they log their symptoms and daily health behaviors in a mobile app, for example.

“Having all of this data in one place, where it is searchable and where we can apply machine learning to it, takes what was once an overwhelming task and makes it manageable,” Bryan said. “We will have a database set up where we can query drug responses and associated genomes immediately, which will provide us real power and speed in this search. The PHC is almost like a wine cellar – the older the wine and the data get, as we collect patient genomic and phenotypic data over time, the more valuable it’s going to be.”

Precision Medicine and the Apple IIe Computer

Over time, as your cells grow and divide, typographical errors are introduced into the three billion letter blueprint inside each of your cells – your genome. If the spellchecker machinery in your cells doesn’t fix these typos, or mutations, your cells can start to misbehave and even become cancerous.

There are other types of changes, apart from mutations, that can also influence the growth of a cancer. For example, you can inherit genetic variants from your mother or father, such as BRCA1 and BRCA2 variants, that put you at higher risk for cancer. Fortunately, these variants can often be drug targets, points of weakness that our doctors can harness during cancer therapy. You may also develop too many or too few copies of a gene that influences the behavior of cancer cells in your body. If researchers can identify such changes in your genome, they can target them with existing or newly designed drugs, giving your body a better chance of killing off cancerous cells. This is the premise of applying genomics to precision medicine for cancer therapy.

Precision medicine hasn’t lived up to its initial promise, yet. This is partly because identifying rare genetic markers person-by-person is a hugely complex and time-consuming task.

“I look at precision medicine, especially as it relates to genomics, to be like the old Apple IIe computer,” Bryan said. “I had the Apple IIe growing up, and it didn’t do much other than maybe some simple video games and basic calculations. But you can see now how computers have changed our lives completely. It took people seeing past what the Apple IIe wasn’t at the time, and I think we are in the same place with precision medicine and genomics today. We are going to look back on what we are doing today and wonder at how simplistic we were being, trying to look for single markers to match to a really limited cadre of chemotherapy drugs. I think in the future we are going to get really fast at finding patterns of genomic markers and use this information to match to an expanding pipeline of medications.”

Today, drugs designed to target rare genetic markers like NTRK fusions and RET fusions, proteins made by aberrant gene fusions, are resulting in patient response rates of over 80 percent. These drugs may only work for a small number of cancer patients, but for those patients they work really well. The more of these rare genetic markers that we can identify, the closer we will get to effective precision cancer treatments for a great number of people.

Matching Drugs to Your Own Cancer DNA

As Bryan spends his time across both the laboratory and the clinic, he sees breakthroughs in terms of discovering new genetic markers and cancer drug targets reach real-life patients.

“Every day we meet a patient in the clinic who has a really interesting or rare genetic marker for which we try a medication that works really well,” Bryan said. “This is what gets me up in the morning and excited to go to work. My hope is that over the next decade, the fraction of those patients in our clinic will grow.”

Bryan recently had a patient enter his clinic with a very aggressive anaplastic thyroid tumor. The median survival for patients with this incurable type of cancer is only 3-6 months. It’s also a fairly chemo-resistant disease.

“There’s not much that works for this cancer,” Bryan said. “This young gentleman came into our clinic with anaplastic thyroid cancer having tried two different traditional chemotherapies without success. While trying to keep our expectations realistic, we gave whole genome sequencing a go with this patient.”

Bryan’s group discovered two different genetic markers that they could target with drugs for this patient. They found a BRAF mutation, for which there are drugs that have been used in melanoma patients, and an over-expression of programmed death-ligand 1 (PD-L1), which indicated that an immune-based therapy might also be beneficial for this patient.

The patient responded extremely well to a treatment combining a BRAF inhibitor with a PD-L1 inhibitor.

“He had complete clearance of disease for over a two-year period of time, and is still doing well,” Bryan said. “Without a precision medicine-based approach, this patient would not have experienced the dramatic results that he had.”

Moving Forward

“We have found that the earlier you do whole genome sequencing for a cancer patient, the more valuable it seems to be,” Bryan said. “The problem now is that we use most of our more advanced tools very deep into a patient’s journey with cancer. Patients at that point don’t have a lot of options and are wanting to try new things. The downside is that the cancer genome has become more complicated by then because of additional mutations. The ideal situation is to perform whole genome sequencing earlier in a patient’s journey, when the genome is less complicated and when you have more therapeutic options.”

With precision medicine, we have the ability to make the patient’s journey better, Bryan says. But he is even more excited that precision medicine and associated technologies can also help the entire field forward, which in turn helps future generation of patients have even better outcomes. Cloud platforms like LifeOmic’s Precision Health Cloud can store genomic sequencing data for long periods of time at low cost, providing valuable information and context around a patient’s cancer journey over time.

In the future, LifeOmic’s PHC will help Bryan and other researchers combine complex patient datasets, including protein levels, gene expression data, and many others. This will provide a more comprehensive view of each patient and allow an even more personalized approach to therapy.