Welcome to Molecular Ideas. We’re back today with a Startup Showcase for an organization on the cutting edge of molecular diagnostics for ovarian cancer and so much more.
Ovarian cancer affects 1 in 78 women in the US, and accounts for more deaths than any other cancer of the female reproductive system. While not as prevalent as breast or uterine cancer, ovarian cancer often evades detection until the late stages where treatments are largely ineffective. This translates into simply stunning statistics – tracking studies by the National Cancer Institute (NCI) show the 5-year relative survival rate for ovarian cancer is 49.1% - a full 32% lower likelihood of survival than uterine cancer.
Put another way, a women’s average chance to survive five years after being diagnosed with ovarian cancer is about the same as a coin toss. Those aren’t favorable odds.
5-Year relative survival stats for women's breast, uterine, and ovarian cancer. Up-to-date stats can be found at https://seer.cancer.gov/statfacts/index.html.
Ovarian cancer is known colloquially as ‘the silent killer’ for two reasons. The first reason is that the most common symptoms are often masked or mistaken for period symptoms or other normal biological processes. Vaginal bleeding, pain, and bloating are all commonly associated with periods or common medical conditions. The second reason is that there has been limited success in developing screening protocols that provide confidence in a diagnosis. Both the transvaginal ultrasound and CA-125 protein test can show reliable results when the cancer has significantly grown. However, the latter is often elevated for benign reasons, which leads to more false alarms than true early detection. What’s worse is that invasive tests are needed to confirm the veracity of these results – which cost patients time and money.
Building a Better Ovarian Cancer Diagnostic with HYPERmelt
Last week, I was honored to sit down with the founder of HYPERmelt, Dr. Chrissy O’Keefe, Ph.D., to discuss her novel and low-cost approach to a non-invasive ovarian cancer diagnostic. The company’s name stands for ‘High-Density Profiling and Enumeration by melt’. That translates into rapidly identifying small pieces of unique genetic material in a sample and evaluating whether they reflect a potential cancer.
HYPERmelt’s molecular diagnostic uses pap-smear samples, and can be supplemented with blood samples. Imagine – instead being forced to go to the clinic because something feels wrong, you could be diagnosed as part of an annual OB/GYN checkup. It radically alters the cost paradigm since earlier detection lowers treatment costs. More importantly, it can save countless lives.
Building a molecular diagnostic requires three pieces of knowledge:
We must know what target we’re looking for, and that the target reflects a specific disease state.
We need to know that we will have enough of it in the sample when testing to statistically confirm a diagnosis.
We need to make sure our test is sensitive enough to ‘perceive’ the target substance while not mistaking it for something else.
We also must filter out all the other ‘noise’ in our body – the various platelets, hormones, sugars, fats, and other compounds – to find what we’re looking for and avoid false positives or negatives. This noise itself can be deafening for diagnostics leveraging traditional technologies. In a sea of noise, silence is deadly.
‘Now wait a minute –’ I hear you saying, ‘this sounds familiar.’ Unlike Theranos (the fallen angel of Silicon Valley), HYPERmelt’s technology is grounded in over five years of translational research at one of Johns Hopkins’ most prestigious molecular diagnostics research labs led by Dr. Jeff Wang. It was here that Dr. O’Keefe made an important observation that led to a whole new direction for her diagnostic.
Traditionally, cancer diagnostics look for unusual metabolites (byproducts of reactions in the cell or large proteins, like CA-125). However, these metabolites are rarely a smoking gun, and can easily be missed in the early stages of disease. As such, newer diagnostics are looking for specific sequences of circulating free DNA sequences that are ‘shed’ by cancerous cells and signal that a cell has become cancerous. Dr. O’Keefe is taking that one step further – instead of looking for mutations, they are looking for changes in methylation patterns of this circulating free DNA.
Methylation – The 'Dimmer Switch' of Gene Regulation
Methylation is a key signaling tool your cells use to regulate when genes are expressed. While certain tumor suppressor proteins like p53 work as binary on-off switches for gene expression, methylation patterns work more like dimmer switches. Unlike mutations or other abnormalities in your DNA itself, changes in the methylation patterns - called 'epigenetic changes' - can change how your body reads a DNA sequence.
“[Examining methylation patterns] makes so much more sense to study for an early-stage diagnostic.” She told me. “When someone starts to develop cancer, these switches are probably what start to break down first, before key mutations occur. Mutations are so much rarer and strictly regulated.”
A visualization of how methylation patterns make specific genes available for transcription, i.e., turning them 'on' or 'off' (visual from Wikipedia)
Exploring methylation patterns of circulating free DNA is a fairly new approach in diagnostic development, but critical in ovarian cancer as a ‘silent killer’. Different cancers leave different metabolic signatures in the blood. Identifying the appropriate methylation patterns of circulating free DNA unique to ovarian cancer adds a whole new set of letters to our arsenal that can be used to decode that signature.
While breaking new ground in science is never easy, Dr. O’Keefe and her team has had great success. “Our first test had just eight samples – four cancerous and four healthy tissues – and we had almost perfect results in three out of four samples each. So, we were like, maybe, maybe this could work.” Since then, her team has gone on to iterate and is currently validating their diagnostic against 500 clinical samples through a recently awarded R01 for early ovarian cancer screening in collaboration with Dr. Ie-Ming Shih and Dr. Tian-Li Wang at Johns Hopkins Hospital. This robust analysis may seem excessive, but validating their methodology is only one part of what makes HYPERmelt’s approach so innovative – and so challenging.
Putting a Lab on a Chip
The methylation patterns are what we are looking for; now, we turn to how to build a diagnostic. What is just as impressive as the method used to identify these methylation patterns in circulating free DNA is how Dr. O’Keefe and her team have miniaturized the reaction using microfluidics technology.
After removing the metabolic ‘noise’ in the blood sample through traditional prep procedures, the extracted material is inserted into a small cartridge, or ‘chip’. This chip attempts to capture relevant pieces of circulating free DNA with the methylation patterns that specifically correspond to ovarian cancer. Depending on the strength of the reaction and number of similar reactions on the chip, a skilled human reader or algorithm can identify whether a patient has cancer or precancerous cells before a tumor would show up on a scan.
Example 'lab on a chip' from Castillo-Leon et. Al (2015). Note how the sample is filtered ('split') before working with reagents to react in the appropriate test site.
To ensure accuracy, the chip is preloaded with dozens of reaction chambers to isolate each piece of circulating free DNA. “The DNA gets split across 10,000 different chambers, or wells. If we load a thousand pieces, the probability that any two DNA are in the same chamber is less than 0.1%. So to analyze each individual piece of DNA, we just have to make sure we have enough chambers for the math to work out.” says Dr. O’Keefe.
Other companies – whether established or startup – use this technology for enhancing the full gamut of life science research, from basic research to drug discovery & development, to diagnostics. This ‘lab-in-a-chip’ approach is incredibly cost effective in three critical ways:
It saves considerable time that clinical lab technicians would spend running multiple reactions to achieve the same level of diagnostic accuracy. This also reduces per-patient costs by allowing them to be informed of their results earlier – possibly even in the same visit.
It lowers material and labor costs for executing multiple reactions.
It promotes accuracy by testing for multiple points of data per test, rather than just one or two – reducing the rate of false positives and hidden negatives.
There is no doubt that ovarian cancer represents one of the most critical unmet medical needs in the oncology category. That said, it represents a significant hurdle, made all the higher by the need to prove new science.
When I asked Dr. O’Keefe why she chose this therapeutic category, she replied, “We hope this platform will have a broad impact for many cancers lacking suitable screening methods. As a start, after speaking with multiple organizations and taking stock of the samples we had available to us, we thought we could have the greatest impact in ovarian cancer. Starting from zero and going to anything above zero sounded like a good place for us to start, rather than trying to tackle one of the larger cancers with more established diagnostic pathways. Plus, our lab does a really great job connecting science to potential large impact and mixing biology and instrumentation.”
Dr. O’Keefe has been working on further validation studies and building her business in collaboration with other Johns Hopkins resources, including the legendary accelerator, FastForward U, and the world-renowned technology-transfer office, Johns Hopkins Tech Ventures to prepare for commercialization.
“It’s funny,” she told me, “After five years of research, what I think of as ‘the hard part’ keeps changing. The hard part turns into, oh, I know how to tackle this problem. And now I'm like, 'how do I write a licensing agreement?' 'Or build a website?' That's really hard now. But that probably takes a lot fewer years to learn.”
Plus, this is just the beginning. Dr. O’Keefe and her team have also been experimenting with expanding this diagnostic platform to other types of cancer, as well as to detect the presence of sepsis-causing bacteria. This ‘lab-in-a-chip’ would lower diagnostic costs and time to diagnosis drastically, enabling higher quality-of-care.
Regardless, it’s clear that the best is yet to come.
If you are interested in supporting a funding round or learning more about HYPERmelt, please reach out via the Molecular Ideas ‘Contact’ page to be put in touch.
That’s all for today! Please share this article, and sign up to leave your thoughts, ideas, and opinions in the comments. Your feedback is always welcome and helps Molecular Ideas grow!