Sameek Roychowdhury, MD, PhD, is an assistant professor of medical oncology at The Ohio State University Comprehensive Cancer Center (OSUCCC), where he also leads their precision cancer medicine program. Dr Roychowdhury focuses on the translation of basic research to patient care; i.e., a treatment goal. His research aims to develop novel molecular diagnostic tools to help guide the treatment of individual cancer patients. Specifically, he intends to match patients who have advanced cancer of any histology or tissue of origin with new, targeted therapies in clinical trials. This is part of their program’s initiative for precision cancer medicine—matching patients to therapies based on molecular alterations in the genome instead of tissue of origin.
We spoke to Dr Roychowdhury about the research being conducted in his laboratory.
Is the ongoing precision cancer medicine clinical trial one of the main projects for the cancer center?
Yes. The study welcomes patients with any type of cancer that is advanced or unresponsive (refractory) to standard therapies. We examine a patient’s current disease, rather than archival disease from the original diagnosis. Our study allows patients to benefit from novel testing of their tumors, including both DNA and RNA sequencing of material from current metastatic and normal cells. Sequencing DNA from normal cells identifies the patient’s germline, helps us understand their tumor, and can provide information about future cancer risk.
In our CLIA-certified cancer genomics lab (see definition, Clinical Laboratory Improvement Amendments), we develop diagnostic tests that meet analytical validity criteria for CLIA standards. The tests must be accurate and precise, so they can be offered in the study and used to return information to patients and their referring oncologists for subsequent treatment decisions.
How does a patient get selected for your trial, and what is the process?
Patients are referred to our precision cancer medicine clinic either by our oncologists at The James Cancer Hospital and Solove Research Institute or from community oncology clinics throughout Ohio and beyond. The clinics refer patients who have advanced metastatic disease that is either resistant to standard therapies or, especially for diagnosis of a rare cancer, has no effective standard treatment.
After the patients have given their consent, we offer them a tumor biopsy through interventional radiology, in collaboration with the radiology department. This provides a fresh sample of the tumor that can be immediately frozen. We also take a blood sample or buccal swab for germline testing. The pathology group evaluates this sample for tumor content and quality. Our CLIA-certified lab then isolates DNA and RNA from the sample to prepare appropriate libraries for target capture and sequencing.
We design custom xGen® Lockdown® Probes (IDT) to capture genomic regions of interest for sequencing on benchtop MiSeq® instruments (Illumina). Our computational team processes the sequencing data to generate a report that we can share with the patient and the referring physician. Typically, the referring physician is a medical oncologist or hematologist in need of a new systemic therapy option via clinical trial.
How many patients are you working with on this study, and what is your throughput?
This is a rolling study. We envision this approach as the future of oncology—the ability to characterize a patient’s cancer at the molecular level in real time. For the tumor sequencing, we have enrolled approximately 400 patients and are currently enrolling 4–6 patients per week for testing. We expect to expand the study to include more patients.
What is the duration of the testing processes; i.e., how quickly after seeing a patient can you return results?
It can take up to 3 weeks, but we generally return results within 10–14 days from the time of biopsy or receipt of the patient’s tissue. This fast turnaround is possible because we can focus our attention on panels of genes, as opposed to the whole exome or the whole genome.
This is the difference between discovery and the clinical approach. During discovery, we look at the whole genomic landscape and are comprehensive. In the clinical approach, we are constrained to clinically actionable genes, those for which therapies exist, the time frame within which results can be delivered, and the cost of the assay. These factors are important in determining how to build a molecular diagnostics assay.
The data we generate will be shared through a common mechanism such as dbGaP (the NCBI database of genotypes and phenotypes). Other cancer centers have adopted similar strategies to systematically profile their patients. With their permission, and with proper security, we can make this information widely useful for all of us to mine and learn from, especially if it is connected with measureable clinical outcomes.
Tell us more about the panels. What panels are you using, and are they predesigned or custom?
We initially tried IDT xGen Pan-Cancer and AML Panels to determine how well these probes performed. Since then, we have created and are using several custom xGen Lockdown Panel designs. One design targets approximately 280 genes to detect point mutations, copy number alterations, and tumor mutation burden in DNA. A second panel targets 93 full-length RNA transcripts to detect gene fusions by RNA sequencing, and can also be used for correlation with our DNA sequencing for aberrations, such as expressed point mutations, or for expressed transcripts for genes that may be amplified on the DNA panel. A third design is aimed at evaluating microsatellites across the genome and measuring microsatellite instability.
Combining these sequencing strategies should allow us to integrate data at the DNA and RNA levels. This analysis will provide critical information about what is occurring in the transcriptome.
Do you expect to observe large differences in the results you get from RNA vs. DNA sequencing?
This is something we expect to learn from our assays. We might find in one patient, depending on their type of cancer, few or many point mutations. Interpreting those changes for clinical use can pose challenges. Sometimes the involvement of a specific driver mutation is obvious; at other times, it is not. Additionally, patients may have more than one prototypical driver point mutation or gene amplification. So interpretation can be challenging, and we are often limited by the prevailing clinical trials or available drugs. We hope that combining results from RNA sequencing and targeted DNA sequencing will help us generate additional information for interpretation.
We will evaluate DNA alterations and use the RNA sequencing to determine whether those alterations are expressed or not. Many CLIA-certified laboratories are currently performing testing at the DNA level, looking at 10–500 genes. We are looking at the exons of 280 genes and planning to supplement that with targeted transcriptome sequencing.
Certainly, looking at DNA-level alterations alone has not been enough. It is a good starting point, but not everybody with the same genotype responds to the same therapies. Obviously, there is more to cancer than the genomic landscape—the transcriptome is important, the proteome is important, and the epigenome is important. I believe examining the transcriptome, and being able to correlate it to DNA-level changes, is a great starting point for novel assay development. Our next goal is to evaluate the clinically actionable epigenome.
For years, cancer was typed at the organ level—liver, lung, head and neck, etc. That appears to be changing, with cancer increasingly recognized as a disease of the genome. Now Drug A, which was originally intended for, say, liver cancer, can be recommended for someone with cancer of a different organ but with the same mutational profile. Is this becoming a wide-ranging practice or do you think there is much more to be uncovered regarding genomic and transcriptomic profiling of cancer?
That question comes up a lot: whether we are going to change from a tissue-of-origin to, for example, a pathway-based classification of cancer. It turns out that pathways in one cancer show up in another cancer type, and features of some cancer types show up regardless of the pathway.
We are discovering that cancers in 2 different organs may, in fact, have a lot in common. Certainly, cancers will have features and behaviors that are unique to their tissue of origin. Having expertise for lung cancer versus expertise for prostate cancer or leukemia is important clinically, because the various diseases display certain patterns of behavior. I think that NGS and, perhaps, the next “omics” technologies that come along will serve as new tools to further classify that molecular taxonomy on top of the clinical picture that we have. It is an exciting time for cancer research, moving from “tissue of origin” to “tissue of origin plus molecular taxonomy”. We are currently performing the clinical research studies to figure out how best to utilize new high-throughput profiling approaches.
You have spoken about combining different channels of data (genomics, proteomics, etc.). Are you also creating a database to try to model some of the more typical clinical information that you obtain from interviewing a patient, such as family history?
Absolutely. That is definitely part of the consultation process—to enquire about the patient’s prior history, response to previous treatment, etc. That information is certainly required for personalized health care in general. Another initiative with much national interest is the “learning healthcare system”. I consider that another great opportunity to leverage what we do here versus what happens at other cancer centers—mine the pooled data, come up with hypotheses that we can test, and develop new strategies for patient management. Capturing the clinical data that allows us to do basic research is another field, research informatics.