Trovagene (San Diego) sees an opportunity to transform cancer care through noninvasive genomic monitoring using cell-free DNA. Unlike other technologies that rely upon blood samples, Trovagene’s precision cancer monitoring technology can utilize urine samples to determine mutational status and to quantify response to treatment based upon dynamic shifts in mutational load. The platform allows for single-molecule analytical sensitivity with DNA inputs of up to 100,000 genome equivalents.
recently spoke with Trovagene CEO Antonius Schuh, Ph.D., to discuss the future of genomic monitoring of cancer patients.
When looking at cell-free DNA, why is urine preferable to blood?
Our focus is really cell-free DNA and we will extract and detect cell-free DNA in any specimen we get, including plasma or blood. However, we believe there are features associated with urine that make certain applications clinically much more feasible—most importantly, monitoring.
There are monitoring applications where you need to acquire samples often. Our early clinical data indicates that you can observe changes within one week and, depending on the type of treatment, you can see indications informative of response within one day. When you are looking at higher-frequency sampling, blood becomes increasingly less feasible. Also, for all practical considerations, urine samples are not volume constrained. As a result, there is a significantly larger input of DNA, which means a higher chance of detecting low-abundant mutations. We just published a paper inCancer Discovery
showing with histiocytic patients (a malignancy where it is hard to obtain a usable biopsy), not only are we able to determine mutational status reliably from urine, we actually outperformed biopsies by a significant margin when it comes to successfully typing a patient’s mutational status.
Trovagene has aggressive goals for publishing clinical trials in the next year. How does this contribute to the company’s overall strategy?
In order to achieve reimbursement, at least two peer-reviewed publications demonstrating the utility of your approach are required by most insurance carriers to initiate technical and clinical assessment of a novel test. We are just now meeting the criteria for insurance carriers to look at our technology.
All our studies have the same objectives. There are three stages to demonstrate clinical utility and health economic impact. The first stage is simply diagnostic. Can I determine mutational status of a malignancy from urinary cell-free DNA, and how well does it correlate to results obtained from a tissue sample? This may be easy in treatment-naive late-stage cancer patients, because they typically have large amounts of circulating tumor DNA (ctDNA). But if patients are delivering plasma or urine samples six weeks after start of treatment, and if they respond well, ctDNA levels can be drastically reduced and an input-constrained sample like blood may simply not have enough ctDNA in it to reliably detect the mutation of interest.
So, again, this is the first clinical question: Can we determine mutational status reliably from cell-free urinary DNA, meaning can we save a biopsy? This would clearly be of clinical utility, and there is a strong health economic argument here because taking a urine sample is simply cheaper than taking a biopsy.
The second level of clinical utility we investigate is can we monitor for treatment response? Can we monitor quantitative changes in the mutational signal indicative of treatment response, or lack thereof? Can we observe response to treatment as fast as conventional tools, such as imaging modalities, or hopefully, significantly faster? And are we able to detect the onset of progression as fast as with standard of care, or are we able to observe progression even earlier?
A third aspect of our clinical program focuses on the emergence of resistance mutations that are relevant for a given treatment. The clinical utility is obvious. It is important to determine tumor dynamics under treatment. For us the question is, is the qualitative and quantitative mutational signal in ctDNA providing valuable information incremental to imaging? If the cell-free DNA signal is highly informative, can we then significantly reduce the number of imaging studies and repeat biopsy procedures, which would save significant expenses and radiation exposure?
Noninvasive prenatal testing (NIPT), which also uses cell-free DNA, has been hailed as a commercial success for molecular diagnostics. How does cell-free DNA monitoring for oncology monitoring compare to NIPT?
I was the CEO of Sequenom from 2000 to 2005, when we started working with Dennis Lo on cell-free DNA in pregnancies. At the time, next-generation sequencing (NGS) was commercially out of reach. Now that has dramatically changed. I see a strong parallel between both clinical applications.
There has been a convergence of three factors over the past two decades. We have learned that all cancer is caused by DNA changes, or more simplistically damage to cellular genomes. As a result, we have kicked off a massive effort to develop targeted cancer treatments that are educated by these causative genomic changes. And lastly, as stated before, NGS has become so affordable that it can be used routinely in clinical practice. I don’t have hard numbers, but if you would ask how many cancer patients had their tumors sequenced in 2010 at Moores Cancer Center here in San Diego, the answer would probably be close to none. In 2014 it is likely many hundreds. Memorial Sloan-Kettering announced they plan to sequence tumor tissue from 10,000 patients this year, with estimated numbers increasing significantly over time. We predict that genomic assessment and genomic monitoring of cancer patients will become clinical mainstream fairly quickly.
How do you predict adoption of cell-free DNA oncology monitoring will unfold?
Even in small study populations, cohorts as small as 15 or 20 patients, you can achieve high statistical significance when identifying responders and nonresponders. That is possible because you are looking at discrete yes or no answers. In theory that is what you would expect to see, much like in infectious disease, where, for example, reduced viral count indicates response to treatment. The hypothesis in oncology is if you deplete the body of tumor cells, you should see less ctDNA. It makes logical sense, but we know a lot of clean theories don’t really translate in a clean manner to clinical settings. This one does. That is why we expect adoption will be fast.
It is all about the performance of your assay in a highly degraded sample. That is what cell-free DNA by definition is. If you look at currently available tests to detect KRAS mutations, the specified analytical sensitivity is not that strong and typically reaches the low single-digit percentage range. We have demonstrated that we can achieve single-molecule sensitivity for DNA inputs of up to 100,000 genome equivalents. That’s not 2 percent to 4 percent sensitivity. That is 0.002 percent or 0.001 percent, and we can do this when the target sequence that carries the mutation is as tiny as a 30-mer. A traditional assay would simply miss this fragment entirely. It would be a tiny sardine swimming through your net and you would say, “no fish in this pond.”
Our early clinical data show that our superb analytical sensitivity and also quantitative performance are clinically meaningful. You would otherwise be at risk to classify patient samples as mutation negative. Mutational signals can be so faint that you simply need this level of analytical performance to see it. We believe this is clinically relevant, and our thinking as we generate data is that really urine is not just more convenient and easier to sample, but the combination of feasibility and sample size simply translates into better clinical performance.
You are approaching commercialization at a time that the Food and Drug Administration (FDA) may begin regulating laboratory-developed tests. How is Trovagene preparing for that possibility?
The regulatory framework we are operating in is always on our mind. When we started building Trovagene’s diagnostic laboratory operations, we considered that there will be more regulation in this space. Our CLIA lab, for example, is set up in compliance with ISO standards. That is just one component of thinking forward. As far as the proposed draft guidance for laboratory-developed tests is concerned, is there some incremental regulation necessary in the molecular diagnostics space? I’d say probably yes. I think, however, the FDA is acutely aware of the fact that we don’t even know yet how this should all be regulated. The clinical science is very complex, conventional clinical validation models are not feasible, and we are moving close the sensitive area of regulating clinical practice and a physician’s access to information. I believe this is the reason why the proposed transition time periods are very long, such that the clinical community and the diagnostic innovators can work through this.
Of course every diagnostic company is a little worried that they could face overreaching regulation. But assuming regulation is going to be pragmatic and focused on ensuring that clinical diagnostic data reported to a clinician is reliable, that it’s good data from validated platforms, and that the clinical meaning you read into these observations is based on solid science, then we actually appreciate more regulation. Frankly, when you have excellent analytical performance and a regulatory framework that allows you to objectively differentiate yourself from competitors, that’s an advantage.
In the next few years how will cancer monitoring evolve?
The new paradigm will be a mix of less imaging and more genomic monitoring. Genomic monitoring is an absolute necessity because oncology drugs are more and more directly targeting the genomic changes driving a patient’s cancer. Every larger oncology hospital and large integrated health care delivery networks are thinking intensively of how they are going to integrate genomic diagnostics and monitoring into cancer care and what tools they will need to bring genomic patient management into practical care. I think this will become standard very quickly. We aren’t going to wait another five years. It can be done now. Early data is very strong.
- Number of Employees: 25
- Number of Collaborations: 12
- Number of Clinical Trials Under Way: 30
- Number of Mutations Under Research: 10