podcast Peter Attia 2022-07-11 topics

#213 ‒ Liquid biopsies and cancer detection | Max Diehn, M.D. Ph.D.

Audio

Show notes

Max Diehn is a Professor of Radiation Oncology at Stanford and a clinical radiation oncologist specializing in lung cancer. Max’s research focuses on developing novel methods for detecting circulating tumor DNA in the blood of cancer patients and on elucidating the molecular pathways and genes associated with cancer. His interests also include uncovering biomarkers that can predict patient survival, responses to therapy, and disease recurrence. In this packed episode, Max discusses the history of blood-based cancer screening and the importance of understanding the predictive value of tests—sensitivity, specificity, negative predictive value, positive predictive value – and how these metrics play into cancer screening. Max then goes in depth on the topic of liquid biopsies, including the history, current landscape, and possible future of liquid biopsies as a cancer detection tool. He discusses how these non-invasive blood tests can detect DNA/RNA from tumor cells released into the blood as well as the different methods one can use to predict if a cancer is present. He gets granular on the topic of cell-free DNA/RNA signature, methylation patterns, and the importance of knowing mutation information, and he ends with a discussion on the exciting future of liquid biopsies and how we can possibly get to the panacea of cancer screening.

Subscribe on: APPLE PODCASTS | RSS | GOOGLE | OVERCAST | STITCHER

We discuss:

Max’s training that planted the seeds for development of liquid biopsies [4:30];
Max’s decision to specialize in radiation oncology [11:45];
A culture at Stanford that values research and physician scientists [17:00];
The motivation to develop liquid biopsies [19:15];
History of blood-based cancer screening and understanding the predictive value of tests [25:30];
Current state of lung cancer and the need for better screening [32:45];
Low-dose CT scans: an important tool for managing lung cancer but with limitations [42:00];
Using liquid biopsies to identify circulating tumor cells [47:00];
Liquid biopsy research moves from circulating tumor cells to cell-free DNA [1:03:00];
Zeroing-in on circulating tumor DNA in cell-free DNA [1:10:48];
Cell-free RNA and Max’s vision for cancer detection from a blood sample [1:22:00];
Methylation patterns and other informative signatures found in DNA [1:24:30];
Mutation-based methods of liquid biopsies [1:26:30];
Understanding the sensitivity and specificity of a diagnostic test [1:30:30];
Existing clinical liquid biopsy tests and their limitations [1:37:30];
The future of liquid biopsies [1:44:00];
How we get to the panacea of cancer screening [1:52:00];
More.

Show Notes

*Notes from intro:

Max Diehn is a Professor of Radiation Oncology, Vice Chair of Research, and Division Chief of Radiation and Cancer Biology at Stanford University
Max is a co-founder of Foresight Diagnostics , a precision medicine company, developing novel liquid biopsy tests for measurement of minimal residual disease
He’s also the co-founder of CiberMed , a company that applies data science for biomarker discovery
Max consults and advises a number of companies in similar spaces
Max’s current research involves the development of novel methods for detecting circulating tumor DNA in the blood of cancer patients He also works to understand cancer cells by identifying molecular pathways and genes associated with disease He’s interested in uncovering biomarkers that can predict response to therapy or predict patient survival and return of disease as early as possible, which is something we’ll get into in the discussion so you can understand why it’s so important to predict recurrence as soon as it happens
Clinically, Max is a radiation oncologist and he specializes in lung cancer
He manages a broad clinical research portfolio, and he focuses on improving these personalized therapies for patients with lung cancer
Max and Peter were classmates in medical school
In this episode, they talk about Max’s background and how he became interested in liquid biopsies
We go into great detail here on sensitivity, specificity, negative predictive value, positive predictive value These are things that everybody needs to understand if they want to be smart on diagnostics, and if they want to understand cancer screening We talk about how they play into cancer screening, especially when it comes to understanding prevalence and pretest probability
We spend some time talking about lung cancer, which is the number one killer for both men and women It’s not just a smoker’s disease, 15% of people who die of lung cancer have never smoked a cigarette in their life
We dive really deep into liquid biopsies; the landscape, the history, the possible future of liquid biopsies In this episode, we get a lot more granular around the nuances of the different ways in which not just we can look at circulating tumor cells versus cell-free DNA
When looking at cell-free DNA, what are the different methods that can be used to predict if a cancer is present? How can we look at the actual genes from the actual cancer that we know we’re searching for versus in a screening situation when we don’t know the gene that we have to look for other clues.
We talk about these cell-free DNA, RNA signatures
We talk about methylation patterns
We talk about the importance of knowing mutation information
We talk about the difference in some of the screenings being approved by the FDA versus those that are being permitted to use for patients without FDA approval formally
We talk about the path to using blood alone as a screening for early detection and the use case for liquid biopsies
We also talk about cancer screening beyond when you know the mutations Peter thinks this question is the hardest one to get at because when you start to think about pan screening for early detection of cancer, it seems unlikely that you’re going to know the mutations
There’s a lot packed into this episode, but it is truly one of the most important subjects given the difficulty in treating cancer when it becomes advanced
He also works to understand cancer cells by identifying molecular pathways and genes associated with disease
He’s interested in uncovering biomarkers that can predict response to therapy or predict patient survival and return of disease as early as possible, which is something we’ll get into in the discussion so you can understand why it’s so important to predict recurrence as soon as it happens
These are things that everybody needs to understand if they want to be smart on diagnostics, and if they want to understand cancer screening
We talk about how they play into cancer screening, especially when it comes to understanding prevalence and pretest probability
It’s not just a smoker’s disease, 15% of people who die of lung cancer have never smoked a cigarette in their life
In this episode, we get a lot more granular around the nuances of the different ways in which not just we can look at circulating tumor cells versus cell-free DNA
How can we look at the actual genes from the actual cancer that we know we’re searching for versus in a screening situation when we don’t know the gene that we have to look for other clues.
Peter thinks this question is the hardest one to get at because when you start to think about pan screening for early detection of cancer, it seems unlikely that you’re going to know the mutations

Max’s training that planted the seeds for development of liquid biopsies [4:30]

Max’s training at Stanford

Peter and Max were in medical school together at Stanford
After the 1st 2 years of med school, Peter went off into the clinical stuff while Max went off into the lab
Most MD PhD programs split medical school in half, focusing for the 1st 2+ years on medical school classes, then doing PhD work in the laboratory, and finally going back for clinical work
There’s an important transition point there when you’re finishing the classroom part of the medical school and deciding what lab to work in
Max chose to do his dissertation with Pat Brown , who was a professor in biochemistry He was the founder and CEO of Impossible Foods until recently
What attracted Max to Pat Brown’s lab‒ Pat had invented technology for measuring the expression of basically all the genes in the genome with a technique called DNA microarrays At the time, this was revolutionary Before, only one or a handful of genes were measured at a time This technology was opening up whole new fields that would allow us to learn so much
Max’s dissertation was a little unusual; he worked on many different projects
It was a unique time in the lab where a new technology is transformative It opened up so many doors simultaneously A new lens through which to view biology He immediately thought of thousands of questions that would be interesting to ask
He was the founder and CEO of Impossible Foods until recently
At the time, this was revolutionary
Before, only one or a handful of genes were measured at a time
This technology was opening up whole new fields that would allow us to learn so much
It opened up so many doors simultaneously
A new lens through which to view biology
He immediately thought of thousands of questions that would be interesting to ask

“ You have this new tool that no one’s ever had before ”‒ Max Diehn

His dissertation focused on 2 main areas‒ 1) immunology and 2) cancer biology These were his 2 interests coming into medical school One project focused on T cells , a critical part of the adaptive immune system He was interested to see what genes are turned on and off in T cells when they are activated When they are stimulated through their T cell receptor (TCR) alone or with a co-stimulatory signal He measured RNA
RNA is the intermediate between DNA and proteins DNA microarrays are a way of measuring specific RNAs
DNA microarrays allowed him to see hundreds and thousands of genes changing as he manipulated T cells He was able to build a catalog of all the genes that turned on or turned off when T cells were activated using different signals This catalog has been very helpful for understanding the mechanism of T cell activation This has gotten more interesting with the advent of immunotherapy
These were his 2 interests coming into medical school
One project focused on T cells , a critical part of the adaptive immune system
He was interested to see what genes are turned on and off in T cells when they are activated When they are stimulated through their T cell receptor (TCR) alone or with a co-stimulatory signal He measured RNA
When they are stimulated through their T cell receptor (TCR) alone or with a co-stimulatory signal
He measured RNA
DNA microarrays are a way of measuring specific RNAs
He was able to build a catalog of all the genes that turned on or turned off when T cells were activated using different signals
This catalog has been very helpful for understanding the mechanism of T cell activation
This has gotten more interesting with the advent of immunotherapy

How difficult was it to keep the RNA intact as you cataloged the generation of mRNA from DNA as this signal of gene expression?

There’s a lot of things there that are interesting; one of them is the instability of RNA

RNA is chemically unstable when compared to DNA and one has to be very careful to prevent its degradation This is a hurdle to overcome in working with RNA
For example, in his experiments with T cell stimulation, the cells are alive This maintains the RNA Only once the cells die does degradation start to happen The experiment is designed carefully to immediately add solutions to protect the RNA after the cells are killed at the end of the experiment So there’s not time for chemical degradation of the RNA
The other project Max worked on during his PhD was to develop a method to isolate RNA that’s stuck to the endoplasmic reticulum (ER) inside cells RNA encoding proteins that are secreted from cells or surface proteins, is sent to the ER These RNAs (and proteins) are important for diagnostic and therapeutic purposes He was interested in cataloging these RNAs There was a long procedure to purify the subset of RNA stuck to the ER It required a lot of work in a cold room (at 4 o C) to maintain the integrity of the RNA He did a lot of work on methods to stabilize RNA
This is a hurdle to overcome in working with RNA
This maintains the RNA
Only once the cells die does degradation start to happen
The experiment is designed carefully to immediately add solutions to protect the RNA after the cells are killed at the end of the experiment So there’s not time for chemical degradation of the RNA
So there’s not time for chemical degradation of the RNA
RNA encoding proteins that are secreted from cells or surface proteins, is sent to the ER
These RNAs (and proteins) are important for diagnostic and therapeutic purposes
He was interested in cataloging these RNAs
There was a long procedure to purify the subset of RNA stuck to the ER It required a lot of work in a cold room (at 4 o C) to maintain the integrity of the RNA
He did a lot of work on methods to stabilize RNA
It required a lot of work in a cold room (at 4 o C) to maintain the integrity of the RNA

How much insight could you get into noncoding sequences of genes (the part that doesn’t encode protein)?

With the technique Max was using at the time, they were focused on measuring the coding portion of the genes (the RNA transcripts that code for proteins)
They had to decide at the beginning of each experiment, which genes to measure Because they had to create a probe for each gene
Subsequently, this approach was used in some of the early work on long noncoding RNAs That was largely led by a former postdoc from Pat Brown’s lab who was there while Max was a grad student, named Howard Chang , who is professor at Stanford
Max finished his PhD in 3.5 years then he went on to begin his clinical rotations
Because they had to create a probe for each gene
That was largely led by a former postdoc from Pat Brown’s lab who was there while Max was a grad student, named Howard Chang , who is professor at Stanford

Max’s decision to specialize in radiation oncology [11:45]

Did you have a sense that you definitely wanted to be a clinician, which would mean not just finishing medical school, but then doing a residency?

Yes, he was set on doing a residency but wasn’t sure what the focus would be
Before med school he thought he would go to graduate school for a PhD and focus on research for his career
But when he was a junior in college, his father was diagnosed with lymphoma
His interactions with oncologist and the medical team revealed how little we knew about many things related to cancer
This convinced him to be on the doctor’s side to help people, not just on the patient side with his father
He wanted to, “ Help patients at the time of their life where it might be the worst time of their life as well as to try to move the field forward to improve treatments that while they worked somewhat obviously were not good enough. ”
He came to medical school with an inkling that oncology was where he wanted to be
What he didn’t realize early on was how many options in that there are He was looking at it mostly through the lens of medical oncology There is surgical oncology, radiation oncology, and other sorts of avenues
He was looking at it mostly through the lens of medical oncology
There is surgical oncology, radiation oncology, and other sorts of avenues

Max’s childhood

Max was born in Munich, Germany and came to the US when he was 11 years old
Initially his family moved to south Florida
When he was a sophomore in high school, they moved to Connecticut
He went to Harvard as an undergrad and lived in the Northeast until he came out to Stanford

Max’s clinical training

He considered surgical oncology; he liked the procedural aspects of it
Ultimately he decided on radiation oncology Initially he was leaning towards medical oncology, because that’s what he had seen through his dad’s journey He did a rotation in radiation oncology in large part because his wife (also a medical student at Stanford) did a year of research in the department of radiation oncology He really enjoyed radiation oncology from a patient care standpoint They would generally see less patients in a day than might be seen in medical oncology This was very attractive to him
Initially he was leaning towards medical oncology, because that’s what he had seen through his dad’s journey
He did a rotation in radiation oncology in large part because his wife (also a medical student at Stanford) did a year of research in the department of radiation oncology
He really enjoyed radiation oncology from a patient care standpoint
They would generally see less patients in a day than might be seen in medical oncology This was very attractive to him
This was very attractive to him

“ Radiation oncologists had a little more time in clinic to spend with each patients ”‒ Max Diehn

He also really liked the technology aspects
This was a field where they do a lot of imaging to see where tumors are in the patient’s body and then we have fancy robots that deliver the radiation very precisely
It seemed to combine his love of oncology and patient care with his interests in technology development
There was an opportunity to do work in the laboratory at a molecular level where not so many people were working on this compared to medical oncology

Was Max able to do any research during his residency?

Peter recalls discussing the last 2 years of medical school with Karl Deisseroth (in episode #191 ) the challenge of keeping his hand in the lab during his residency
Max did an internship in medicine at Stanford then 4 years of radiation oncology (residency), 5 years in total
He remembers trying to write some papers in the call room, to finish up work from his PhD
Radiation oncology residency programs have a research track for individuals who are interested in laboratory based research called the home and pathway In other fields this is called short tracking or fast tracking The idea is they can trade in some of the clinical training time for research time This gave him a postdoc time of about 2 years during his 4 years of radiation oncology residency During that postdoc time, he had clinic activities about half a day to a day a week and the other days were in the lab It actually was a really good preview of what his life would be like once he finished all the training and became a faculty member
In other fields this is called short tracking or fast tracking
The idea is they can trade in some of the clinical training time for research time
This gave him a postdoc time of about 2 years during his 4 years of radiation oncology residency
During that postdoc time, he had clinic activities about half a day to a day a week and the other days were in the lab It actually was a really good preview of what his life would be like once he finished all the training and became a faculty member
It actually was a really good preview of what his life would be like once he finished all the training and became a faculty member

A culture at Stanford that values research and physician scientists [17:00]

How did you decide to stay at Stanford?

Was the department of radiation oncology at Stanford a natural fit for the type of problems you wanted to solve on the research side?

The department at Stanford was one of the very first departments of radiation oncology in the US
Radiation oncology grew out of radiology , which is a diagnostic arm of radiology; where you do imaging to diagnose disease
Their first chairman, Henry Kaplan , was a very famous radiation oncology physician scientist He led the movement to develop radiation oncology as its own specialty In the ‘50s he worked on curing Hodkin’s disease with radiotherapy, some of the 1st successes in curing young patients with lymphoma
The radiation oncology department at Stanford has a history of strong interest in laboratory-based research Thanks to Henry Kaplan who was doing laboratory research at the time also while seeing patients
There were faculty in this department that were laboratory based Even just PhD’s who were fully laboratory based; this isn’t the case in many radiation oncology departments
Max really liked the research environment
He led the movement to develop radiation oncology as its own specialty
In the ‘50s he worked on curing Hodkin’s disease with radiotherapy, some of the 1st successes in curing young patients with lymphoma
Thanks to Henry Kaplan who was doing laboratory research at the time also while seeing patients
Even just PhD’s who were fully laboratory based; this isn’t the case in many radiation oncology departments

“ This was a place where I could see that the kind of research I wanted to do was valued ”‒ Max Diehn

There were mentors who were successful physician scientists This is important when you’re young, to have mentors that can show you how to overcome obstacles if you’re in trouble
This is important when you’re young, to have mentors that can show you how to overcome obstacles if you’re in trouble

The motivation to develop liquid biopsies [19:15]

When did he become interested in the idea of liquid biopsies?

Peter has been following liquid biopsies for about 6 years
Obviously Max has been working on it for a lot longer
For many people, liquid biopsies are still a bit of a black box
Max did not start his lab with the idea of focusing on liquid biopsies He got there by following the results from some experiments
He got there by following the results from some experiments

“ One of the keys to being a successful scientist… is following the data ”‒ Max Diehn

All research projects begin with a clinical need
He decided to specialize in treating lung cancer clinically
He was in the clinic 1 day a week and saw lung cancer patients exclusively
Treating early stage lung cancer patients left Max frustrated Stage 1 or stage 2 lung cancer, which means the cancer hasn’t spread (as far as they could tell); the table and figure below summarize the stages of cancer
Stage 1 or stage 2 lung cancer, which means the cancer hasn’t spread (as far as they could tell); the table and figure below summarize the stages of cancer

Figure 1. Stages of cancer. Image credit: National Cancer Institute

Figure 2. Cancer staging indicates its growth and spread. Image credit: Wikipedia

He could cure the majority of these patients with targeted, high dose radiation
But the cancer came back for 20-25% of these patients
After he finished treatment, in the 1st follow-up visit 3 months after radiation, he couldn’t tell who was cured and who would have the cancer come back
State-of-the-art at the time was to have patients come back for scans every 3, 6, 12 months
Max notes, “ It’s a very reactive approach and very unsatisfying, and this is something we went through with my dad .”

How small of a tumor can be detected with a CT scan?

It is hard to see things much smaller than one centimeter in diameter , maybe eight millimeters This will depend somewhat on the location in the body As a general rule of thumb, this is about a billion cells‒ a vast number of cancer cells
Peter notes, “ anything at a centimeter is not posing a direct threat to the organism ” With the rare exception of a badly placed brain tumor 1 cm in most cancers would be irrelevant A 10 cm x 10 cm tumor would be fatal
Max explains, “ whether a tumor is fatal or not, as you point out, greatly depends on the location. In certain areas, one can have a ton of tumor without it being fatal. And in other areas you have very little room for that. It’s a volumetric. It grows by the cube, by the radius multiplied by itself three times. And that is, of course, non-linear. It grows much faster. The number of cells in a mass are much more than just the linear increase in the diameter of the lesion. ”
Peter notes, this is the part that is hard to understand We do a good job with linear thinking We don’t think exponentially very well The point here is, we are flying blind from a diagnostic standpoint when the cancer cells number from 0 to 1 billion cells
Metastasis of cancer is ultimately what ends of killing patients; when the cancer spreads
If the cancer is localized, a surgeon or radiation oncologist can cure the patient
Once the cancer has spread to many places in the body, it cannot be removed or you can’t radiate it all
This will depend somewhat on the location in the body
As a general rule of thumb, this is about a billion cells‒ a vast number of cancer cells
With the rare exception of a badly placed brain tumor
1 cm in most cancers would be irrelevant
A 10 cm x 10 cm tumor would be fatal
We do a good job with linear thinking
We don’t think exponentially very well
The point here is, we are flying blind from a diagnostic standpoint when the cancer cells number from 0 to 1 billion cells

The problem is what’s called micrometastatic disease or microscopic disease that has spread elsewhere in the body, you could have dozens or even hundreds of these micrometastatic deposits in different organs; and these are all under 0.5 cm

The detection of micrometastatic disease is beyond the limits of our imaging methods; we can’t see them all
Max thought he could potentially get a handle on it if he had a test that could measure in the blood the contributions from all these dozens of micrometastatic deposits This may provide higher sensitivity This was his motivation to work on liquid biopsies
This may provide higher sensitivity
This was his motivation to work on liquid biopsies

“ I had no idea how I was going to do that when I started it ”‒ Max Diehn

History of blood-based cancer screening and understanding the predictive value of tests [25:30]

The history of blood-based tests for cancer

Peter notes, there are a couple areas where blood tests for specific proteins are being used for cancer screening; they’re not great PSA levels for prostate cancer CEA levels for colon cancer CA 19-9 levels for pancreatic cancer
When Max begin thinking about developing a blood-based test initially for lung cancer, he thought of these examples These examples above are all protein biomarkers Proteins made by cancer cells that are shed into the blood where they can be measured in a non-invasive fashion
PSA levels for prostate cancer
CEA levels for colon cancer
CA 19-9 levels for pancreatic cancer
These examples above are all protein biomarkers
Proteins made by cancer cells that are shed into the blood where they can be measured in a non-invasive fashion

Shortcomings of protein biomarkers

Max points out, “ A large issue with these protein biomarkers is that they are actually not that specific for cancer, meaning that these are also proteins that normal cells can make ” Cancers often make more of these proteins The problem is knowing how to interpret low levels of these proteins
Cancers often make more of these proteins
The problem is knowing how to interpret low levels of these proteins

Does the patient have a little bit of cancer or is it just the baseline of what their normal cells are making?

“ That lack of specificity… was the major Achilles’ heel of the protein-based biomarkers ”‒ Max Diehn

A lot of work had been previously done looking for biomarkers in lung cancer CEA was a good biomarker in some patients
Protein-based approaches were state-of-the art when Max began his research but their major weakness was a lack of specificity
CEA was a good biomarker in some patients

The difference between sensitivity and specificity in understanding the positive and negative predictive value of tests

Sensitivity is a synonym for the true positive rate , the likelihood of having a positive test when the patient has the actual condition

If you have 100 cancer patients [they all have cancer], sensitivity of a test is how many of them show a positive blood test
A perfect test would have 100% sensitivity, 100/ 100 cancer patients would test positive
Mammography might have an 85% sensitivity, which means 85 of 100 women with breast cancer would be identified with a mammogram This is giving a false negative result for 10 of the 100
It’s important to realize that there are no perfect tests
This is even more true when you start pushing the envelope to see smaller and smaller tumors It’s unrealistic to have a test that’s 100% sensitive No test will catch every cancer patient‒ this is hard for patients to understand and leads to a lot of frustration
This is giving a false negative result for 10 of the 100
It’s unrealistic to have a test that’s 100% sensitive
No test will catch every cancer patient‒ this is hard for patients to understand and leads to a lot of frustration

Peter’s anecdote on sensitivity and specificity

Peter’s former analyst, Bob Kaplan came up with a great thought experiment to explain this

If you drive a test to high sensitivity, the specificity must go down

If you wrote a letter to 1000 random women and told each they had breast cancer (with no additional knowledge), this is technically a test with 100% sensitivity because you have identified every woman in this group with breast cancer Let’s assume 50 of these women have breast cancer There are no false negatives in this group The problem is the specificity of this test is horrible There are so many false positives that the test is useless
Let’s assume 50 of these women have breast cancer
There are no false negatives in this group
The problem is the specificity of this test is horrible
There are so many false positives that the test is useless

This illustrates that you can have 100% sensitivity without any clinical utility

Max replies, “ That explains a lot of what happens in the research diagnostic research field in general. You try to push sensitivity, but that never is meaningful to report if you don’t also report on specificity .”
Specificity is the inverse of sensitivity, for patients that don’t have cancer it is the report that their test is negative

It’s the reporting of true negatives

Sensitivity and specificity are connected, you can push the sensitivity higher if you ignore the specificity

Current state of lung cancer and the need for better screening [32:45]

What is the current state of lung cancer?

“ It’s the number one cause of cancer death ”‒ Max Diehn

Outcomes in breast or prostate cancer are actually much better; we can cure a large fraction of these patients
The vast majority of patients with lung cancer, historically, cannot be cured This is why lung cancer is the #1 cause of cancer-related death
The incidence of lung cancer is going down in men and now also in women due to a decrease in smoking Secondly treatment have gotten better Screening is slowly getting better
While smoking is by far the largest risk factor for lung cancer, it’s not the only risk factor
This is why lung cancer is the #1 cause of cancer-related death
Secondly treatment have gotten better
Screening is slowly getting better

“ Anybody who has lungs can get lung cancer ”‒ Max Diehn

Lung cancer can develop as a result of: pollution, radon gas or other things in the environment, genetic associations
If we could magically make smoking go away, there would still be lung cancer

The 2 major types are small cell and non-small cell lung cancer

Non-small cell lung cancer involves adenocarcinoma , squamous cell carcinoma , and large cell Adenocarcinoma is the most common lung cancer in the non-small cell subtype When a non-smoker gets lung cancer, it’s usually adenocarcinoma Squamous cell carcinoma also makes up a significant fraction of non-small cell lung cancers
Carcinoid is its own category, it’s not a small cell lung cancer and it’s not considered non-small cell lung cancer either
Adenocarcinoma is the most common lung cancer in the non-small cell subtype
When a non-smoker gets lung cancer, it’s usually adenocarcinoma
Squamous cell carcinoma also makes up a significant fraction of non-small cell lung cancers

What genes are thought to be involved in the risk of lung cancer?

Asian women seem to have an elevated risk of lung cancer Potentially related to differences in estrogen and testosterone
Asian women seem to have an elevated risk of lung cancer Potentially related to differences in estrogen and testosterone
Potentially related to differences in estrogen and testosterone
This is a big mystery, we have not identified the genetic drivers of lung cancer in Asian women
It’s multifactorial‒ it’s due to multiple genetic variants that aggregate to elevate risk There may also be some further environmental factors that aren’t smoking that may interact with that genetic background to increase risk
There may also be some further environmental factors that aren’t smoking that may interact with that genetic background to increase risk

Radon is a big environmental exposure conferring risk

Particulate matter less than 2.5 microns is small enough to make its way into the most distal part of the lung
There is an increase in all-cause mortality associated with exposure to 2.5 micron particulate matter (PM 2.5) The graph below shows the size distribution of particulate matter in the atmosphere
The graph below shows the size distribution of particulate matter in the atmosphere

Figure 3. Size of various types of particulate matter in the atmosphere. Image credit: Wikipedia

More cases of lung cancer is associated with living in cities versus rural environments and this correlates with particulates in the ambient environment There is definitely an association in epidemiologic studies To prove it in a human is difficult
There is a biological rationale‒ particulates cause irritation and chronic inflammation
There are also chemicals in smog that along the lines of what’s in tobacco smoke, can be direct carcinogens
There is definitely an association in epidemiologic studies
To prove it in a human is difficult

What is the dose-response curve in pack-years of tobacco smoking for the relative increase in risk of lung cancer?

Studies suggest there is increased risk as you go from zero to higher levels
Risk plateaus around 20-30 pack-years; this may be saturating risk
A pack-year is not an ideal clinical variable because you can’t measure it in a completely unbiased way In reality, patients smoke different amounts of time and they don’t always remember perfectly
In reality, patients smoke different amounts of time and they don’t always remember perfectly

What does the data say about secondhand smoke?

This is very difficult to measure, even harder than pack-years
Some professions are exposed to a lot of smoking‒ waitresses/ waiters, stewardesses/ stewards There are increases in lung cancer in these groups epidemiologically
There are increases in lung cancer in these groups epidemiologically

How to quantitate secondhand smoke exposure and determine when one’s risk is significantly elevated is a major problem

This is an areas where it would be very useful to have a biomarker that could be quantitatively measured, to actually measure one’s exposure
Secondhand smoke is not a criteria for patient eligibility for cancer screening by low-dose CT This criteria is focused on people who have the highest risk in the population You want to detect at least 0.5% of the cancer you are screening for because otherwise you will screen lots and lots of patients who will never get cancer This is when specificity becomes a problem Peter notes, “ You then need to know the prevalence or the pretest probability to know how to interpret the result. Without the prevalence, you can’t impute the positive and negative predictive value, which is what tells you when you have a positive, how confident are you it’s positive, and similarly, when you have a negative, how confident are you there? ” (we’ll come back to this)
Though exposure to pollutants and secondhand smoke are known to increase the risk of lung cancer, there are public health reasons why these are not the screening criteria This can be frustrating for patients in these categories
This criteria is focused on people who have the highest risk in the population
You want to detect at least 0.5% of the cancer you are screening for because otherwise you will screen lots and lots of patients who will never get cancer This is when specificity becomes a problem Peter notes, “ You then need to know the prevalence or the pretest probability to know how to interpret the result. Without the prevalence, you can’t impute the positive and negative predictive value, which is what tells you when you have a positive, how confident are you it’s positive, and similarly, when you have a negative, how confident are you there? ” (we’ll come back to this)
This is when specificity becomes a problem
Peter notes, “ You then need to know the prevalence or the pretest probability to know how to interpret the result. Without the prevalence, you can’t impute the positive and negative predictive value, which is what tells you when you have a positive, how confident are you it’s positive, and similarly, when you have a negative, how confident are you there? ” (we’ll come back to this)
This can be frustrating for patients in these categories

Low-dose CT scans: an important tool for managing lung cancer but with limitations [42:00]

Low-dose CT has been one of the major changes in the last 10 years in how lung cancer is managed
There have been a lot of efforts to develop a screening test for lung cancer, given that it is the number one cause of cancer deaths There have been a lot of failures Initial studies focused on chest x-rays; there was little benefit due to low resolution
The National Lung Screening Trial took patients and randomized them to get either a low-dose CT scan or a chest x-ray There was a relative reduction of risk of about 20% in the patients who got the low-dose CT scan This was viewed as a win for using low-dose CT as a screening test The absolute risk reduction was small, in the single digit % Let’s say it was 5%, then the NNT (# needed to treat) would be 20; you could save a life for every 20 people Peter thinks this sounds good Max notes the value is actually lower than that, maybe 0.5-1%
A a low-dose CT scan is a more high resolution way of looking at the lungs; it can see smaller nodules than a chest x-ray
The value of a low-dose CT scan gets complicated because you think you either see a nodule or you don’t, but it’s not that easy A CT scan is complicated to read Interpretation of the scan and what is considered positive will affect the sensitivity, specificity, and the NNT
The way low-dose CT scans are read now is different than when the initial study was done because of a high risk of false positives
Low-dose CT is not Max’s area of expertise
There have been a lot of failures
Initial studies focused on chest x-rays; there was little benefit due to low resolution
There was a relative reduction of risk of about 20% in the patients who got the low-dose CT scan This was viewed as a win for using low-dose CT as a screening test
The absolute risk reduction was small, in the single digit % Let’s say it was 5%, then the NNT (# needed to treat) would be 20; you could save a life for every 20 people Peter thinks this sounds good Max notes the value is actually lower than that, maybe 0.5-1%
This was viewed as a win for using low-dose CT as a screening test
Let’s say it was 5%, then the NNT (# needed to treat) would be 20; you could save a life for every 20 people
Peter thinks this sounds good
Max notes the value is actually lower than that, maybe 0.5-1%
A CT scan is complicated to read
Interpretation of the scan and what is considered positive will affect the sensitivity, specificity, and the NNT

Even though this test is a major home-run for screening (because we can save cancer deaths), it’s not a perfect test

The estimated amount of radiation in a low-dose CT is 1.5 mSv This is significantly less than scans used routinely for patients who already have cancer This is an important concern because radiation can cause cancer; so the question becomes, is the risk worth it The risk of causing cancer is much, much lower for a low-dose CT scan
This is significantly less than scans used routinely for patients who already have cancer
This is an important concern because radiation can cause cancer; so the question becomes, is the risk worth it
The risk of causing cancer is much, much lower for a low-dose CT scan

What would be your personal threshold for how much radiation you would want to receive a year from imaging?

Peter points out, “ The NRC says that we shouldn’t be exposed to more than 50 millisieverts in a year. Is that like saying you shouldn’t drink more than 10 drinks in a day? ”
Being at sea level probably exposes you to 1-2 millisieverts a year If you live in Colorado, you could probably double that
If you live in Colorado, you could probably double that

Do you start to worry at 20 millisieverts in a year? Which is obviously pretty easy to do with a PET CT scan.

Max notes, they don’t routinely track this at the individual patient level
The risk is really low and harm is actually very difficult to quantitate It’s difficult to quantitate exactly how much risk of, let’s say, a second malignancy there is, a future malignancy there is, from a given protocol of a CT scan or whatnot You could never get the data to randomize all those things
It’s difficult to quantitate exactly how much risk of, let’s say, a second malignancy there is, a future malignancy there is, from a given protocol of a CT scan or whatnot You could never get the data to randomize all those things
You could never get the data to randomize all those things

The calculus of medical practitioners who order imaging studies is‒ will this test help manage the patient, will it benefit the patient

If the scans won’t change the course of treatment, then they shouldn’t be ordered
The way practitioners think about scans is‒ are they worth doing?
Some patients who have lung cancer and get radiation therapy receive very high doses of radiation If they were not a cancer patient, one would never want those amount of radiation But Max points out, “ The vast majority of those patients will not get a cancer from their imaging or radiation treatment; and their cancer, if we leave it untreated, will kill them. ”
This comes back to the need to catch cancer early, to have a chance at a cure
They don’t have patients wear radiation-detection badges to track their exposure, as if beyond a certain point no further imaging would be done It’s always decided on a case-by-case basis
If they were not a cancer patient, one would never want those amount of radiation
But Max points out, “ The vast majority of those patients will not get a cancer from their imaging or radiation treatment; and their cancer, if we leave it untreated, will kill them. ”
It’s always decided on a case-by-case basis

Using liquid biopsies to identify circulating tumor cells [47:00]

Liquid biopsy research began with collecting blood samples from lung cancer patients

A hypothetical cancer patient

A patient had either a resection and/or radiation following their NED (no evidence of disease)
Let’s say the recurrence rate is 25% So 25 of 100 of these people are going to have a recurrence of their cancer
We know that the sooner we catch the recurrence, the better the odds are for treatment There is a lower tumor burden There is a lower burden of mutations This gives greater odds for success of therapy
If this were 200 years ago, we’d be hosed because we’d have to wait until the patient was coughing up blood
But even with high-resolution CT scans, at best we have to wait until there are at least a billion cancer cells
So 25 of 100 of these people are going to have a recurrence of their cancer
There is a lower tumor burden
There is a lower burden of mutations
This gives greater odds for success of therapy

How did Max think of looking for a different type of biopsy, a liquid biopsy?

“ I was frustrated by not being able to diagnose a recurrence earlier ”‒ Max Diehn

The inability to predict which of his patients was going to have their cancer recur was an unmet need that he wanted his lab to work on
There are 2 ways to go about this 1 – Build a mouse model (a preclinical model) to try to grow tumors in animals then see if you could have a hypothesis for what might be a good biomarker Test it in the mouse and then go back to the human 2 – Translational research, do the research in the human; it’s the final model This is where he began They work directly on blood samples If they find something, it should be directly applicable
He began by just collecting blood samples He didn’t know what he was actually going to do with them
1 – Build a mouse model (a preclinical model) to try to grow tumors in animals then see if you could have a hypothesis for what might be a good biomarker Test it in the mouse and then go back to the human
2 – Translational research, do the research in the human; it’s the final model This is where he began They work directly on blood samples If they find something, it should be directly applicable
Test it in the mouse and then go back to the human
This is where he began
They work directly on blood samples
If they find something, it should be directly applicable
He didn’t know what he was actually going to do with them

Was this funded by the NIH, a R01 ?

As a new faculty member, when you get that first job, you get startup funds This provides the money to kickstart your research because you don’t have any grants yet This is how he funded this initial work
One of the things with academic research and funding is that oftentimes you have to show it already works before you can get funding for it This is a very common story because grant funding is limited Organizations that fund research are often quite conservative if the idea sounds great, but you have no proof that it works at all‒ they aren’t going to give you money
Max used his startup funds to begin collecting blood and started thinking about what biomarkers would be good He collected blood both prior to resection/ radiation treatment and in follow-up visits after treatment
He read the literature on protein biomarkers They didn’t seem very useful
He spent a year looking at circulating tumor cells (CTCs) , which is one subtype of the liquid biopsy field These are intact cancer cells that can circulate in the blood of patients
These cells are very complicated to measure for a variety of reasons They are not very abundant so they’re difficult to purify In a good purification there might still only be 1% or less of the purified sample
This provides the money to kickstart your research because you don’t have any grants yet
This is how he funded this initial work
This is a very common story because grant funding is limited
Organizations that fund research are often quite conservative if the idea sounds great, but you have no proof that it works at all‒ they aren’t going to give you money
He collected blood both prior to resection/ radiation treatment and in follow-up visits after treatment
They didn’t seem very useful
These are intact cancer cells that can circulate in the blood of patients
They are not very abundant so they’re difficult to purify
In a good purification there might still only be 1% or less of the purified sample

Finding circulating tumor cells is like finding a needle in a haystack

Further with intact cells, they need to be processed on the same day or within 1-2 hours of the blood draw This makes it very difficult to build up biobanks of frozen samples to provide a large enough cohort to study
This makes it very difficult to build up biobanks of frozen samples to provide a large enough cohort to study

The last thing he did were some control experiments

These are critical
You always want to ask, is the thing I’m working on really working?
They drew blood from healthy individuals who don’t have cancer They found circulating tumor cells even though these patients didn’t have cancer This led him to realize they were picking up something else; there was a lack of specificity
They found circulating tumor cells even though these patients didn’t have cancer
This led him to realize they were picking up something else; there was a lack of specificity

He realized looking for circulating tumor cells wasn’t going to make it into the clinic anytime soon

This led him to reevaluate and ask, what else could they do?

Did you find any proteins that were expressed by lung cancer cells that were unique to them?

No
They did not screen themselves, but other people had
Mass spectrometry can be used to analyze the proteins in a blood sample; those studies did not find convincing markers unique to lung cancer cells
Some markers were elevated in some patients CEA was one marker that can be elevated in lung cancer patients and can even be a marker for recurrence (like PSA can be in prostate cancer) But there is a problem with specificity
Many studies of protein biomarkers may be an artifact of the cohort or elevated due to something else going on There was problems with reproducibility Often the lack of reproducibility is not published You need a very large study to definitively prove something doest work, and often this is not worth the time and cost
CEA was one marker that can be elevated in lung cancer patients and can even be a marker for recurrence (like PSA can be in prostate cancer)
But there is a problem with specificity
There was problems with reproducibility
Often the lack of reproducibility is not published
You need a very large study to definitively prove something doest work, and often this is not worth the time and cost

The selection bias for publishing positive results

Peter notes the publication bias this creates Positive results are published more When scientists fail to reproduce the work of others, this is often not published Peter comments on the system, “ That means that one finding gets disproportionately propagated as a positive finding, when in reality four labs have now found it to be highly unlikely, and that doesn’t get published .”
Max agrees this is a problem and points out, “I n general, we don’t have a system where one lab could know that three other labs had done the same thing. Maybe, eventually, you find out over dinner at a conference or something, right? But because there is no place to look up to say what other people have done, there is no way to know. It would be great if we could fix it. It’s a very difficult problem, of course, because of many factors, including the economic parts of limited resources and time. ” If a grad student were assigned to publish such a study, it would take 6 months of their time That’s 6 months they’re not spending on their main project
Reproducing published works is important to know that something really works
Positive results are published more
When scientists fail to reproduce the work of others, this is often not published
Peter comments on the system, “ That means that one finding gets disproportionately propagated as a positive finding, when in reality four labs have now found it to be highly unlikely, and that doesn’t get published .”
If a grad student were assigned to publish such a study, it would take 6 months of their time
That’s 6 months they’re not spending on their main project

It’s really helpful to see multiple studies from multiple groups, ideally with multiple methods, that find the same thing

This is where reading the literature can show if the finding is robust

Back to circulating tumor cells‒ for patients that have stage II lung cancer, how much blood does Max sample before resection?

Around 10-20 ml, a few tablespoons of blood

What’s the maximum number of circulating tumor cells (CTCs) would they find in these samples?

Often it’s zero for a stage I-II cancer patient This means the sensitivity isn’t very good
In stage IV patients they saw quite a lot, some patients had massive amounts
Occasionally he would see early-stage patients with a lot of CTCs Caroline Dive from University of Manchester did a lot of these studies Her group took blood from the draining veins at the time of the surgery (the lobectomy for lung cancer) and did circulating tumor cell assays They found evidence that they could find more signals, meaning that sensitivity was better in that context The hypothesis was that if you’re measuring the blood outflow of the tumor, you’re going to catch more cells Once CTCs hit the systemic circulation, they get diluted all over the entire body
This means the sensitivity isn’t very good
Caroline Dive from University of Manchester did a lot of these studies Her group took blood from the draining veins at the time of the surgery (the lobectomy for lung cancer) and did circulating tumor cell assays They found evidence that they could find more signals, meaning that sensitivity was better in that context The hypothesis was that if you’re measuring the blood outflow of the tumor, you’re going to catch more cells Once CTCs hit the systemic circulation, they get diluted all over the entire body
Her group took blood from the draining veins at the time of the surgery (the lobectomy for lung cancer) and did circulating tumor cell assays
They found evidence that they could find more signals, meaning that sensitivity was better in that context
The hypothesis was that if you’re measuring the blood outflow of the tumor, you’re going to catch more cells
Once CTCs hit the systemic circulation, they get diluted all over the entire body

If you had 2 stage II patients, one is typical and has no CTCs while the second patient has a reasonably high burden of CTCs, would you predict the second patient would have a recurrence over the first patient?

Yes and there’s evidence of that Studies show when you see high levels of CTCs, this is a bad thing; patients are at higher risk of recurrence One complication in the CTC field is that even non-cancer patients can have circulating cells that look like CTCs using the markers available The marker commonly used to identify white blood cells is CD45 White blood cells need to be differentiated from CTCs A marker for CTCs should be expressed in cancer cells but not white blood cells One marker is cytokeratin , a structural protein specific to epithelial cells
Yes and there’s evidence of that
Studies show when you see high levels of CTCs, this is a bad thing; patients are at higher risk of recurrence
One complication in the CTC field is that even non-cancer patients can have circulating cells that look like CTCs using the markers available
The marker commonly used to identify white blood cells is CD45 White blood cells need to be differentiated from CTCs
A marker for CTCs should be expressed in cancer cells but not white blood cells One marker is cytokeratin , a structural protein specific to epithelial cells
White blood cells need to be differentiated from CTCs
One marker is cytokeratin , a structural protein specific to epithelial cells
But, Max’s group and others have found healthy patients can also have cells that have cytokeratin expression but no white blood cell marker expression in circulating cells
Other people have gone further to do single cell sequencing of those to show that their genomes are actually normal and not mutated like cancer cells are

It appears that there can be epithelial cells circulating that are not cancer cells; this makes it complicated to look for the presence of tumor cells

This indicates that looking for CTCs is not a good cancer screening tool
Maybe this approach holds some promise for determining the course of adjuvant therapy in a patient If they see a high number of CTCs then they may want to give the patient additional chemotherapy because of the increased risk of recurrence
This motivated Max to start work on liquid biopsies
If they see a high number of CTCs then they may want to give the patient additional chemotherapy because of the increased risk of recurrence

There is still a lot of work going on in the circulating tumor cell field, but more development is required to use it clinically

Liquid biopsy research moves from circulating tumor cells to cell-free DNA [1:03:00]

After some work with CTCs, Max moved on to look at cell-free DNA This is DNA found in circulation, in the blood plasma, outside of cells It’s called cell-free because it’s outside of cells; of course it originally comes from cells but now it’s circulating by itself
Max became interested in this field after reading about prenatal diagnostics Work by a number of people like Dennis Lo (in Hong Kong) and Steve Quake (in the Stanford Engineering Department) have shown they can detect DNA from the fetus in the pregnant mother’s blood
This is DNA found in circulation, in the blood plasma, outside of cells
It’s called cell-free because it’s outside of cells; of course it originally comes from cells but now it’s circulating by itself
Work by a number of people like Dennis Lo (in Hong Kong) and Steve Quake (in the Stanford Engineering Department) have shown they can detect DNA from the fetus in the pregnant mother’s blood

Max thought if this could be done in pregnancy, then why not in cancer patients

Peter asks for clarification, “ When you take a tube of blood and you spin it, you get all of the cellular matter below the buffy coat, and then the plasma is above. Is the cell-free DNA in the clear plasma and not stuck within the cellular material? ”

Correct, the cell-free DNA (cfDNA) is in the plasma, and that’s a critical technical aspect of isolating it There may be more cfDNA mixed in the cellular compartment, but it would be harder to extract This might be a future research direction
There are low levels of cfDNA in the plasma In a healthy person, there’s 1-5 ng/mL of cfDNA in the plasma
This cfNDA is double-stranded DNA, but rather molecules of around 170 base pairs This is the length of DNA you would expect if the DNA were wrapped around the core histones Histones are the proteins that package DNA in our cells They’re a scaffold that lets the long lengths of DNA compact to fit into the nucleus of a 10 micron cell Shown in the figure below, 8 histones make up 1 nucleosome core particle that winds up about 167 bp of DNA Each nucleosome has 147-167 bp of DNA wound around it The DNA that stretches between nucleosomes is called linker DNA
There may be more cfDNA mixed in the cellular compartment, but it would be harder to extract This might be a future research direction
This might be a future research direction
In a healthy person, there’s 1-5 ng/mL of cfDNA in the plasma
This is the length of DNA you would expect if the DNA were wrapped around the core histones Histones are the proteins that package DNA in our cells They’re a scaffold that lets the long lengths of DNA compact to fit into the nucleus of a 10 micron cell Shown in the figure below, 8 histones make up 1 nucleosome core particle that winds up about 167 bp of DNA Each nucleosome has 147-167 bp of DNA wound around it The DNA that stretches between nucleosomes is called linker DNA
Histones are the proteins that package DNA in our cells
They’re a scaffold that lets the long lengths of DNA compact to fit into the nucleus of a 10 micron cell
Shown in the figure below, 8 histones make up 1 nucleosome core particle that winds up about 167 bp of DNA
Each nucleosome has 147-167 bp of DNA wound around it
The DNA that stretches between nucleosomes is called linker DNA

Figure 4. DNA is wound on nucleosome core particles (composed of histone proteins) to allow it to compact into chromatin; linker DNA is the region of DNA between nucleosomes. Image credit: Nature Reviews Molecular Cell Biology 2018

There is a high level of activity of enzymes that chew up DNA in our blood and extracellular fluid So cfDNA is only in the blood temporarily and it’s temporarily protected by these histones So the DNA that’s bound to the histones is present at a higher frequency in the blood than DNA that’s between the histones [linker DNA] Histones are like pearls on a string, shown in the diagram above And the DNA region between 2 pearls (linker DNA) is relatively depleted from the blood because of enzymes that chew up DNA It’s because of these enzymes that there are low levels of DNA in the blood There are lots of cell deaths, so you might imagine higher levels of DNA in our blood But we have enzymes to degrade DNA High concentrations of DNA become a snot-like substance; that would probably be bad to have in your blood
So cfDNA is only in the blood temporarily and it’s temporarily protected by these histones
So the DNA that’s bound to the histones is present at a higher frequency in the blood than DNA that’s between the histones [linker DNA]
Histones are like pearls on a string, shown in the diagram above
And the DNA region between 2 pearls (linker DNA) is relatively depleted from the blood because of enzymes that chew up DNA It’s because of these enzymes that there are low levels of DNA in the blood There are lots of cell deaths, so you might imagine higher levels of DNA in our blood But we have enzymes to degrade DNA High concentrations of DNA become a snot-like substance; that would probably be bad to have in your blood
It’s because of these enzymes that there are low levels of DNA in the blood
There are lots of cell deaths, so you might imagine higher levels of DNA in our blood
But we have enzymes to degrade DNA
High concentrations of DNA become a snot-like substance; that would probably be bad to have in your blood

Where does this cell-free DNA come from?

This is a mystery
This is difficult to study because most, if not all tissues contribute to the pool of cfDNA

It’s very difficult to study release of something from everywhere

cfDNA is chopped up in these small histone-bound fragments
Historically most of the reviews and textbooks argue that cfDNA comes from apoptosis Cells that undergo apoptosis have their DNA chopped up as part of the process
Apoptosis is a type of cell death It’s a controlled, programmed cell death that gets rid of cells in a tidy way This is a type of cell suicide that can get activated It’s critical for our development because sometimes cells have to die to make room for other cells. Apoptosis is part of our normal developmental homeostasis It also is critical for getting rid of sick cells
DNA inside a cell is chopped up during apoptosis
DNA could also get released from cells through a non-apoptosis process, and it would likely become chopped up too
Necrosis is another way a cell can die and release some of its contents into the blood
Cells that undergo apoptosis have their DNA chopped up as part of the process
It’s a controlled, programmed cell death that gets rid of cells in a tidy way
This is a type of cell suicide that can get activated
It’s critical for our development because sometimes cells have to die to make room for other cells.
Apoptosis is part of our normal developmental homeostasis
It also is critical for getting rid of sick cells

Zeroing-in on circulating tumor DNA in cell-free DNA [1:10:45]

How do you quantify the amount of cell-free DNA (cfDNA) you would get from 10 CC of blood? And how are you distinguishing what is potentially cell-free DNA from a normal cell versus cell-free DNA that is from the cancer cell?

There is a laboratory procedure for isolating DNA from the plasma (the noncellular component of blood) This has been optimized for low concentrations of DNA The concentration of cfDNA is really low, in the single-digit ng/mL Often in the lab we work with mg of DNA so working with ng (1000x less) quantities was one of the early challenges they had to overcome Recall the amount of cfDNA from cancer cells is less than 0.01 or 0.001% of what’s isolated from the blood [in the early stages of cancer]
The most common method to quantify the amount of DNA uses a fluorescent dye that binds to the DNA and gives a specific fluorescence This fluorescence can bread in a machine You use a standard curve to read off how much DNA you have
The more complicated thing they would do next is quantitative PCR This is an enzymatic method to amplify DNA Again, you use a standard curve where the amount of DNA is known and compare to the amplification curve of your sample DNA to determine how much DNA is in the sample
Most of the time they use the simple method‒ add dye to the DNA sample, put it in the machine, and get the result It’s really fast
You would get 10 mL of plasma from 20 mL of blood in most individuals This would have between 1-5 ng/mL of cfDNA in healthy individuals So that would mean a total of 10-50 ng cfDNA
In some patients the level of cfDNA can be much higher (100’s ng/mL) because there’s a lot of cell death happening in the body Advanced cancer patients Patients who have trauma Patients with infection
This has been optimized for low concentrations of DNA
The concentration of cfDNA is really low, in the single-digit ng/mL
Often in the lab we work with mg of DNA so working with ng (1000x less) quantities was one of the early challenges they had to overcome Recall the amount of cfDNA from cancer cells is less than 0.01 or 0.001% of what’s isolated from the blood [in the early stages of cancer]
Recall the amount of cfDNA from cancer cells is less than 0.01 or 0.001% of what’s isolated from the blood [in the early stages of cancer]
This fluorescence can bread in a machine
You use a standard curve to read off how much DNA you have
This is an enzymatic method to amplify DNA
Again, you use a standard curve where the amount of DNA is known and compare to the amplification curve of your sample DNA to determine how much DNA is in the sample
It’s really fast
This would have between 1-5 ng/mL of cfDNA in healthy individuals
So that would mean a total of 10-50 ng cfDNA
Advanced cancer patients
Patients who have trauma
Patients with infection

How do you distinguish cfDNA from a normal cell (presumably the majority of cfDNA) versus a cancer cell?

In the cancer field they focus on unique molecular properties of cancer cells that normal cells generally don’t have

These are the mutations that cause cancer; they are only present in cancer cells, not normal cells

Cancer is a disease that is caused by mutation mistakes in the DNA of a normal cell that accumulate and ultimately lead to that cell, not being responsive to cues, to stop growing or to kill itself and just to grow uncontrollably

“ Those mutations… serve as really exquisitely specific markers of the cancer cell ”‒ Max Diehn

These mutations are attractive biomarkers because 1) you’re tracking the cause of the disease
2) These mutations are highly specific biomarkers Compare to PSA , it can be made by normal prostate cells and by prostate cancer cells But the mutations in the prostate cancer cell would not be present in normal prostate cells
They look for mutations in these short DNA molecules that make up cfDNA
Most of the cfDNA molecules come from normal cells and don’t have mutations
There is a small subset of cfDNA molecules that do contain mutations In a patient with advanced lung cancer 1% of their cfDNA may come from cancer cells This is still a small amount In early stage patients, the amount of cfDNA from cancer cells is less than 0.01 or 0.001% of what’s isolated
Compare to PSA , it can be made by normal prostate cells and by prostate cancer cells
But the mutations in the prostate cancer cell would not be present in normal prostate cells
In a patient with advanced lung cancer 1% of their cfDNA may come from cancer cells
This is still a small amount
In early stage patients, the amount of cfDNA from cancer cells is less than 0.01 or 0.001% of what’s isolated

How is such a small quantity of cfDNA from cancer cells detected?

Next-generation sequencing is the most common method used This is a high throughput molecular methods for sequencing DNA It can identify the actual sequence of the bases (the AGTC bases) that make up DNA It can provide the exact sequence for hundreds of millions of molecules or billions of molecules of DNA in parallel in one experiment
They get millions of DNA sequences that are all about 170 bases in length
They compare those sequences to the patient’s germline DNA (from a normal cell, purified from a buccal swab or the cellular compartment of the blood)
They also start sequencing the actual tumor and compare this sequence to data from the patient’s cfDNA
This is a high throughput molecular methods for sequencing DNA
It can identify the actual sequence of the bases (the AGTC bases) that make up DNA
It can provide the exact sequence for hundreds of millions of molecules or billions of molecules of DNA in parallel in one experiment

Using analysis of cfDNA to see which patients were cured of cancer after treatment

They have a piece of the patient’s tumor, sequence it, and learn the mutations it contains They compare the DNA sequence from the tumor to the DNA sequence from the patient’s normal cells
For example, they identify 10 mutations present in the patient’s tumor but not in their healthy cells
Now they look in the database of sequences from the patient’s cfDNA for any of those 10 mutations
They compare the DNA sequence from the tumor to the DNA sequence from the patient’s normal cells

That turns out to be exquisitely specific and sensitive for the presence of the cancer. If we see even one or a handful of molecules that carry one of these mutations that… means there’s still cancer cells in the body.

The cfDNA fragments are very short, 170 bases long
99.9% of the cfDNA lines up to some segment of the patient’s germline DNA [from normal cells]; but no 2 of these segments are the same either If you could lay their entire germline genome out, it would span miles and miles and miles Peter remarks, “ So you have a whole bunch of 170 base segments that line up somewhere on that problem. That’s a pretty interesting computational problem to me because 170 is not that big .” The cancer is only going to have maybe 100 mutations in the coding regions of the DNA Having more than 10,000+ mutations in the whole genome would be unusual
If you could lay their entire germline genome out, it would span miles and miles and miles
Peter remarks, “ So you have a whole bunch of 170 base segments that line up somewhere on that problem. That’s a pretty interesting computational problem to me because 170 is not that big .”
The cancer is only going to have maybe 100 mutations in the coding regions of the DNA
Having more than 10,000+ mutations in the whole genome would be unusual

This process of finding mutations in cfDNA strikes Peter ase almost improbable given the few number of mutations that exist in the cancer… they would be easy to miss

Max agrees, this is one of the major challenges in the field
His first paper in this field, back in 2014 addressed this problem Published in Nature Medicine , An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage
Published in Nature Medicine , An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage

Figure 5. Identifying mutations in a patient’s tumor and then looking for them in cfDNA to see if the cancer is still present. Image credit: Nature Medicine 2014

History of work with cfDNA

cfDNA was first observed in the mid 1900s by a French group
They could only find this in very advanced cancer patients because they were using very insensitive methods
Even then they had the foresight to think this could potentially be useful
Not much progress was made in the next 50-60 years because the amount of cfDNA was so small and methods were not available yet to measure it
From 2005-2010 very sensitive PCR methods were developed to look for single mutations
For example, you look for a p53 mutation The problem that arises is, how do you look for 1 mutation when there is less cancer genome present in the blood than normal/ germline genome? There are lots of negative samples: it’s possible the mutation isn’t present in that particular blood draw
The way they overcame this problem was fairly simple They used next-generation sequencing technology to get the sequence for millions of molecules Instead of looking for 1 mutation, they looked for dozens of mutations from the patient’s cancer in parallel Compare looking for 1 to looking for 10 This actually increases the sensitivity by 10-fold (a whole order of magnitude)
The problem that arises is, how do you look for 1 mutation when there is less cancer genome present in the blood than normal/ germline genome?
There are lots of negative samples: it’s possible the mutation isn’t present in that particular blood draw
They used next-generation sequencing technology to get the sequence for millions of molecules
Instead of looking for 1 mutation, they looked for dozens of mutations from the patient’s cancer in parallel Compare looking for 1 to looking for 10 This actually increases the sensitivity by 10-fold (a whole order of magnitude)
Compare looking for 1 to looking for 10
This actually increases the sensitivity by 10-fold (a whole order of magnitude)

You don’t have to see all 10 mutations to know the cancer is there; you just need to see 1 mutation

These could be mutations in a coding or noncoding region of the genome
What is beneficial is to have what’s called a truncal or clonal mutation This is a mutation present in every cancer cell
As cancer develops, it gets the first mutation then more and ultimately becomes cancer It can have 10,000 mutations; these accumulate over time Many of these mutations are present in all of the cells of the tumor Even though they’re not driving the cancer, just because of the evolutionary history of the cancer, as cells divide these mutations are passed on to their progeny Once the cell gets that final mutation that allows it to keep dividing forever, it keeps all those other 10,000 mutations
This is a mutation present in every cancer cell
It can have 10,000 mutations; these accumulate over time
Many of these mutations are present in all of the cells of the tumor
Even though they’re not driving the cancer, just because of the evolutionary history of the cancer, as cells divide these mutations are passed on to their progeny
Once the cell gets that final mutation that allows it to keep dividing forever, it keeps all those other 10,000 mutations

It doesn’t matter which mutation they track, they just want a mutation that’s present in all the tumor cells

Peter notes, “ It’s interesting because driver mutations have surprising heterogeneity, right? ”
Talking to Steve Rosenberg about this on the podcast last year , he was amazed at the nonoverlap of driver mutations A lot of people have mutations in p53, but it wasn’t necessarily the driver
This also gets back to the immunology‒ how often are these mutations inducing antigens that are recognized by the immune system?
A lot of people have mutations in p53, but it wasn’t necessarily the driver

Cell-free RNA and Max’s vision for cancer detection from a blood sample [1:22:00]

Cell-free RNA

There is also cell-free RNA (cfRNA)
This field is much more nascent than the DNA field

What advantages would cfRNA offer given the instability of RNA?

Max doesn’t think it has advantages in looking for mutations

Max’s goal vision for cancer analysis from a blood sample

They want to determine as much as possible about a patient’s cancer from a blood draw Even to the point where they don’t need to do a biopsy This is currently science fiction But if you extend their work decades into the future, that’s where they’re trying to go
To get there they need to be able to measure things other than mutations
Mutations are critical, but they’re only a small part of the puzzle
Measuring RNA could tell you about which genes are on and off in the cancer theoretically That is critically important too
In the immunotherapy field, there are some markers expressed in the tumor cells that can basically hide the cancer from the immune system (like PD-L1 ) This marker is not a mutational process in the majority of patients It’s expressed from the promoter ; expression is turned on in response to signaling or epigenetic changes
Even to the point where they don’t need to do a biopsy
This is currently science fiction
But if you extend their work decades into the future, that’s where they’re trying to go
That is critically important too
This marker is not a mutational process in the majority of patients
It’s expressed from the promoter ; expression is turned on in response to signaling or epigenetic changes

You can’t tell by looking at the sequence in DNA of PD-L1 whether the tumor has this marker

This is not a mutation marker
They want to measure something in the blood to distinguish between an adenocarcinoma and a squamous cell carcinoma or between a lung cancer and a breast cancer This could be done with cfDNA but using cfRNA might be better
This could be done with cfDNA but using cfRNA might be better
The use of cfRNA is fascination; they haven’t published their work yet but will soon

“ In the future, RNA will be a big part of the liquid biopsy puzzle that doesn’t replace DNA, it offers complimentary pieces of information ”‒ Max Diehn

Is there a difference between cell-free DNA and circulating tumor DNA or are these terms interchangeable? [1:24:00]

They are different
Generally cfDNA refers to the total DNA in the circulation That includes both the healthy cell-derived DNA and the cancer cell-derived DNA
Circulating tumor DNA is a subfraction of the cfDNA, maybe 1% of the cfDNA is from the tumor
That includes both the healthy cell-derived DNA and the cancer cell-derived DNA

Methylation patterns and other informative signatures found in DNA [1:24:30]

What other signatures are helpful here?

They can look at the DNA methylation patterns in these 170 bp DNA fragments in circulation
They can also look at chemical modifications of certain nucleotides (modified by enzymes)
This is different in different types of cells
Different cells, based on what tissue they are from will have different patterns of methylation
DNA methylation influences which genes are turned on and off You could envision that in a lung cell that needs to have the genes on for surfactant production or making alveoli or whatever, but it doesn’t need the to have the genes on to make fat because it’s not doing that (it’s not a fat cell) Some genes are methylated and turned off in different cells, in different ways
You could envision that in a lung cell that needs to have the genes on for surfactant production or making alveoli or whatever, but it doesn’t need the to have the genes on to make fat because it’s not doing that (it’s not a fat cell)
Some genes are methylated and turned off in different cells, in different ways

DNA methylation patterns can be exquisite marks of the origin of the DNA

Further, the methylation profile of 2 lung cancers is more similar than their mutation profile

Peter’s takeaway: the methylation profile of 2 lung cancers is more similar than their mutation profile

The methylation profile could be used as a pan screen to identify a type of cancer

This is a different scenario than discussed earlier, different from wanting to know if the patient has been cured
The sensitivity of the methylation-based approaches are significantly inferior to that of the mutation-based approaches
But from a practical standpoint, the methylation approaches have some advantages for the screening question The sensitivity and the practicality aren’t aligned You’d want the most sensitive assay possible for screening because the tumors are tiny This is not the current state in the field Currently mutation-based methods are much more sensitive
The sensitivity and the practicality aren’t aligned
You’d want the most sensitive assay possible for screening because the tumors are tiny This is not the current state in the field Currently mutation-based methods are much more sensitive
This is not the current state in the field
Currently mutation-based methods are much more sensitive

Mutation-based methods of liquid biopsies [1:26:30]

Max recently published a new method, their 3rd-generation mutation-based method

Published in Nature Biotechnology in December 2021, Enhanced detection of minimal residual disease by targeted sequencing of phased variants in circulating tumor DNA
It’s 100x more sensitive than than their prior methods (and other methods in the field)

Now they can get down to 1 part in a million

With prior methods they were bottoming out at about 0.01%, so 1 in 10,000
This 3rd-generation mutation-based method is a 2-log improvement in their diagnostic assay
Now they are looking at samples left over from running their 1st-generation assay that came back negative‒ asking which are false-negatives? How many of these samples can they now detect as positive? It wasn’t that there was no ctDNA, they just couldn’t detect it before
With the methylation-based assays, the data is still not very mature to know exactly what their sensitivity is It’s probably closer to 1 in 1000, so 0.1% sensitivity
How many of these samples can they now detect as positive?
It wasn’t that there was no ctDNA, they just couldn’t detect it before
It’s probably closer to 1 in 1000, so 0.1% sensitivity

Peter’s takeaway on mutation-based methods

With the mutation-based methods, you know what you’re looking for
If you’re doing a pan screen for someone who doesn’t have cancer, you can’t know what mutation to look for
A test like GRAIL looks at methylation patterns of cfDNA; they’re doing a pan screen They take 10-20 ml of blood and look for all the cfDNA to identify if there’s cancer or non-cancer there If they find cancer, they want to predict the origin of the tissue The current sensitivity for all stages is probably 50% with a specificity of > 99% But if you’re trying to catch an early cancer (stage 1), the sensitivity is maybe 20%
They take 10-20 ml of blood and look for all the cfDNA to identify if there’s cancer or non-cancer there
If they find cancer, they want to predict the origin of the tissue
The current sensitivity for all stages is probably 50% with a specificity of > 99%
But if you’re trying to catch an early cancer (stage 1), the sensitivity is maybe 20%

How good are these tests?

We want a high specificity 95-99% and we want high sensitivity
Max takes issue with studies that report sensitivity across stages because there is a bias to include more stage IV patients You could include a single stage I patient and 99 stage IV patients and report a sensitivity of xyz for stages I-IV
It’s important to report sensitivity for each stage Within stage I it should be further broken down into stage IA, IB, and even stage IA1 and IA2 In lung cancer, actually the sensitivity for stage I lung cancer in the GRAIL data that’s been presented is actually 5% or less
Peter points out, if the tumor is bigger than stage I, you don’t need a liquid biopsy because it can be seen with a low-dose CT scan… in theory, the only reason you’d care about a liquid biopsy is if we’re talking about fewer than a billion cells
You could include a single stage I patient and 99 stage IV patients and report a sensitivity of xyz for stages I-IV
Within stage I it should be further broken down into stage IA, IB, and even stage IA1 and IA2
In lung cancer, actually the sensitivity for stage I lung cancer in the GRAIL data that’s been presented is actually 5% or less

Max sees an advantage in the blood-based tests because this is a way to get more patients screened

Understanding the sensitivity and specificity of a diagnostic test [1:30:30]

If a test has 50% sensitivity and 99.5% specificity for detecting stage I cancer in an unbiased sample, that would be pretty good

How Peter explains it to his patients

Say you’re 45, not a smoker, your family history is negative, and your colonoscopy is negative You’re pretest probability of having cancer is 1%
Peter’s calculator will allow you to plug in sensitivity, specificity, and prevalence, and it will spit out positive and negative predictive values
You’re pretest probability of having cancer is 1%

⇒ Download the calculator here

⇒ Further discussion in AMA #25

EXAMPLE:

For example with a test that has 55% sensitivity, 99% specificity and the prevalence is 1% the negative predictive value goes to 99.5%
But this is only a little bit better than what was known without the test With a 1% prevalence, this test moved you from 99% to 99.5% negative predictive value
Also note that the positive predictive value is 40% The positive predictive value is the chance that the positive result is not a false positive
With a 1% prevalence, this test moved you from 99% to 99.5% negative predictive value
The positive predictive value is the chance that the positive result is not a false positive

Peter never uses these tests in isolation for his patients

Are these liquid biopsies useful without the mutational information that makes them so sensitive?

Max agrees, these are excellent points
Not only is the sensitivity and specificity of tests hard to understand for patients, but it’s not well understood by a lot of researchers and physicians
50% sensitivity and 99% specificity sounds great, evening ignoring the fact that stage I sensitivity is < 5%
Even at a 50% sensitivity for a patient that has no risk of cancer (less than 1% of them will have cancer) this test isn’t really moving the needle significantly It has a high risk of false positives because the positive predictive values isn’t so good False positive results lead to a lot of testing anxiety If follow-up tests can’t find anything, the patients are worried for years that something might be brewing
It has a high risk of false positives because the positive predictive values isn’t so good
False positive results lead to a lot of testing anxiety
If follow-up tests can’t find anything, the patients are worried for years that something might be brewing

“ I really strongly feel that we as the field need to do studies that prove cancer-specific survival benefits of these tests ”‒ Max Diehn

Currently this is not on the roadmap
There are many commercial efforts but they’re not planning large randomized trials (RCTs) like were done for low-dose CT scans Why not? Because they take years to complete and are expensive It’s not attractive from an investment standpoint
Why not?
Because they take years to complete and are expensive
It’s not attractive from an investment standpoint

We need to do RCTs on these liquid biopsies to learn if they are good enough to save lives

“ We really want to be sure that we are helping our patients and we’re not just adding costs to the healthcare system ”‒ Max Diehn

RCTs will help us do this

The benefit of liquid biopsies is complicated and this is why we need randomized trials

Consider stage I lung cancer patients, there’s 2 types of stage I lung cancer 1) Where the cancer cells have not left the lung; the surgeon removes the tumor and the patient is cured 2) This type looks the same but there are microscopic tumor cells in the liver and brain that we don’t know about
In order for a screening test to be useful, it has to catch a significant fraction of the 1st type of stage I, the type without micrometastases (which the surgeon can cure)
The surgeon cannot cure the 2nd type They can only cure what they can see They don’t know about the micrometastases in the liver, brain, etc.
1) Where the cancer cells have not left the lung; the surgeon removes the tumor and the patient is cured
2) This type looks the same but there are microscopic tumor cells in the liver and brain that we don’t know about
They can only cure what they can see
They don’t know about the micrometastases in the liver, brain, etc.

That is why simply saying a test increases the number of patients diagnosed at earlier stages, doesn’t automatically prove that that test will save lives

You have to look at the granular level and consider there are different kinds of stage In patients
You can envision a test that mainly catches the patient with micrometastatic disease We have hints that the ctDNA assays preferentially catch those patients
The way to prove that tests are not falling into that pitfall is to do randomized studies
We have hints that the ctDNA assays preferentially catch those patients

“ We don’t know yet if they’re going to save lives ”‒ Max Diehn

People argue it isn’t unethical to use these tests because if we find out 5 or 10 years from now that they save lives and we didn’t use them, that’s 5-10 years of patients who didn’t have its benefit
The problem with that argument (which sounds ethically reasonable) is this means we would never test anything new developed in medicine, because there’s always a possibility that this new thing could improve patient outcomes
Peter notes, “ I think that’s why the FDA has taken the posture because right now there are only 4 approved liquid biopsies ” Guardant and 3 others
Max points out these tests are for very different things, for genotype These are for identifying what mutations are present in patients with advanced disease
Guardant and 3 others
These are for identifying what mutations are present in patients with advanced disease

Existing clinical liquid biopsy tests and their limitations [1:37:30]

Guardant

The Guardant assay was one of the first in the field
They use next-generation sequencing to measure circulating tumor DNA (ctDNA) Similar to what Max’s group does
Their initial test (that’s FDA approved), was developed for a very specific clinical problem You have a patient with metastatic cancer, let’s say lung cancer The cancer is in multiple parts of the body, and you want to identify what mutations that patient has This is actionable because in lung cancer, there are drugs that work for patients with certain mutations Historically, a biopsy was needed to get a piece of the tumor to identify the mutations
Guardant (and Max and others) have shown that in many patients with advanced disease 1% of the ctDNA is from the cancer And that’s enough to identify the mutations No need for a biopsy
Similar to what Max’s group does
You have a patient with metastatic cancer, let’s say lung cancer
The cancer is in multiple parts of the body, and you want to identify what mutations that patient has
This is actionable because in lung cancer, there are drugs that work for patients with certain mutations Historically, a biopsy was needed to get a piece of the tumor to identify the mutations
Historically, a biopsy was needed to get a piece of the tumor to identify the mutations
And that’s enough to identify the mutations
No need for a biopsy

This is useful in patients who can’t have a biopsy or where they had a biopsy, but all the tissue is used up; or where the biopsy missed the tumor (which happens a lot)

For example, the liquid biopsy may identify that the patient does have an EGFR mutation, so they should be treated with a EGFR tyrosine kinase inhibitor Max notes, “ That is the easiest problem in the liquid biopsy space, because you’re in stage four patients that are known to have cancer. They have… a high tumor burden. And so, the tests are optimized for that; they work well for that .”
Max notes, “ That is the easiest problem in the liquid biopsy space, because you’re in stage four patients that are known to have cancer. They have… a high tumor burden. And so, the tests are optimized for that; they work well for that .”

Liquid biopsies have an 80% agreement with a traditional biopsy; they’re a good test for this problem

Liquid biopsies are not designed for early cancer screening or detection

They are not designed for situations where the levels ctDNA are many logs lower

They’re not designed for this question of is the patient cured or not after treatment

Again where the levels of tumor DNA are much lower, the FDA approved tests don’t detect this well

Is this true of the other FDA-approved tests, CellSearch and FoundationOne?

Yes
Foundation has a liquid biopsy test, very similar to what we developed in Guardant and they do the same thing now
CellSearch is actually a circulating tumor cell assay It was FDA approved before any ctDNA assay We had talked about circulating tumor cells at the beginning of the liquid biopsy field Circulating tumor cells were the focus of the field in the early 2000s, and so that’s why that was the first test
It was FDA approved before any ctDNA assay
We had talked about circulating tumor cells at the beginning of the liquid biopsy field
Circulating tumor cells were the focus of the field in the early 2000s, and so that’s why that was the first test

“ No one really uses it because it doesn’t provide you actionable information ”‒ Max Diehn

Your doctor runs that test; whatever the result is doesn’t impact how the patient is treated

There is a difference between FDA approval and actually clinical usefulness

A test like GRAIL is not FDA approved yet its use is still permitted

This is a complicated area in diagnostics
There are 2 ways the US government regulates tests
1) FDA approval This focuses on approving drugs and diagnostic tests The FDA evaluates them carefully, gives approval, and allows companies to market them if they pass the bar of what they need to show
2) Set up the assay in a lab that is compliant with the CLIA Act (Clinical Laboratory Improvement Amendments) This is an easier way to get assays to patients This relies on an act of Congress that focuses on regulating laboratories that do diagnostic tests It is independent of the FDA Labs get this designation and are inspected every year of 2 to make sure they’re doing things appropriately These labs now have the blessing to develop any test they want and offer them to patients They’re supposed to run controls but the FDA never reviews them The laboratory is regulated but not the individual test This is what Guardant did before they got their FDA approval This is what GRAIL is doing
This focuses on approving drugs and diagnostic tests
The FDA evaluates them carefully, gives approval, and allows companies to market them if they pass the bar of what they need to show
This is an easier way to get assays to patients
This relies on an act of Congress that focuses on regulating laboratories that do diagnostic tests
It is independent of the FDA
Labs get this designation and are inspected every year of 2 to make sure they’re doing things appropriately
These labs now have the blessing to develop any test they want and offer them to patients
They’re supposed to run controls but the FDA never reviews them
The laboratory is regulated but not the individual test
This is what Guardant did before they got their FDA approval
This is what GRAIL is doing

This allows laboratories to sell tests to patients and providers long before they get FDA approval

Peter notes, there are 30 other companies out there doing this right now and from a technology standpoint, the holy grail is a blood test that will screen patients (who don’t think they have cancer) with enough sensitivity and specificity to identify cancer If it were colon cancer, the patient would then get a colonoscopy to identify the tiny adenomatous polyp in early stage I They would then get the smallest partial colectomy and be cured Or a woman who had a negative mammogram 6 months ago could be identified with a blood test then go get a diffusion weighted image MRI to identify the cancer
If it were colon cancer, the patient would then get a colonoscopy to identify the tiny adenomatous polyp in early stage I
They would then get the smallest partial colectomy and be cured
Or a woman who had a negative mammogram 6 months ago could be identified with a blood test then go get a diffusion weighted image MRI to identify the cancer

“ What would really be amazing is if we had the confidence in these things that even if they were never showing up on imaging, you could treat them ”‒ Peter Attia

For something like pancreatic adenocarcinoma this type of early detection would save lives The 5 year survival rate of even a 1 cm pancreatic adenocarcinoma has a 25% five year survival is only 25% This is a death sentence of a cancer, even at stage 1 So waiting until you can see a billion cells on a CT or MRI is not going to work Peter notes, “ It’s either going to be much earlier detection or treatments that actually work… today we have neither of the above ”
The 5 year survival rate of even a 1 cm pancreatic adenocarcinoma has a 25% five year survival is only 25% This is a death sentence of a cancer, even at stage 1 So waiting until you can see a billion cells on a CT or MRI is not going to work
Peter notes, “ It’s either going to be much earlier detection or treatments that actually work… today we have neither of the above ”
This is a death sentence of a cancer, even at stage 1
So waiting until you can see a billion cells on a CT or MRI is not going to work

The future of liquid biopsies [1:44:00]

Is there a path to this sci-fi world where using blood alone can detect cancer in patients 3-4 years earlier than what’s clinically possible now?

Max is hopeful and notes, “ It’s not going to come in one step, but we are already taking steps towards that, sort of building the bricks on top of each other ”
The easier application of this is in patients who already have known cancer We have super sensitive tests we can run after treatment to see whose cured or not These detect the state that’s called minimal residual disease, meaning microscopic cells that are residual after your treatment and that have currently at least very high, positive, predictive value If you detect ctDNA, that patient is very likely to have the cancer grow back
There are clinical trials underway now to see how this can guide therapy
Max has launched a couple trials recently at Stanford with early stage lung cancer patients After their standard care (chemo &/or radiation) they get a blood test If it’s positive, those patients get immunotherapy Cancer cannot be detected in these patients, but the blood test indicates microscopic cells are left behind This is the moment in the patient’s lifetime with their cancer, that they will have the least number of tumor cells, the fewest mutations, the lowest heterogeneity At this moment, they have the lowest chance of resistance
We have super sensitive tests we can run after treatment to see whose cured or not
These detect the state that’s called minimal residual disease, meaning microscopic cells that are residual after your treatment and that have currently at least very high, positive, predictive value If you detect ctDNA, that patient is very likely to have the cancer grow back
If you detect ctDNA, that patient is very likely to have the cancer grow back
After their standard care (chemo &/or radiation) they get a blood test
If it’s positive, those patients get immunotherapy Cancer cannot be detected in these patients, but the blood test indicates microscopic cells are left behind This is the moment in the patient’s lifetime with their cancer, that they will have the least number of tumor cells, the fewest mutations, the lowest heterogeneity At this moment, they have the lowest chance of resistance
Cancer cannot be detected in these patients, but the blood test indicates microscopic cells are left behind
This is the moment in the patient’s lifetime with their cancer, that they will have the least number of tumor cells, the fewest mutations, the lowest heterogeneity
At this moment, they have the lowest chance of resistance

Max is hopeful they will be able to cure more patients with this type of follow-up

It is known that giving even the same drug in adjuvant therapy is more effective than when it’s used for treating metastatic cancer

The value of early detection and early treatment

“ Early detection absolutely matters … less tumor burden, less heterogeneity equals better outcomes ”‒ Peter Attia

Peter notes this is an important point that gets lots on many people in the field, the value of early detection He’s never seen it refuted Early detection equals better outcomes
He’s never seen it refuted
Early detection equals better outcomes

Why cancers respond better to treatment initially

Cancers become resistant because of genetic heterogeneity
All cancer cells in the tumor are not the same; they already have mutations
Max explains, “ The more cancer you have, the higher chance that one of those mutations will make the cancer resistant to whatever therapy you’re using. So this idea of guiding adjuvant therapy based on ctDNA MRD I think is the first step towards the future that you’re envisioning. ”

Using liquid biopsy after cancer treatment will be useful for guiding adjuvant therapy for some cancers

Peter notes, it won’t be useful in pancreatic cancer because everybody’s probably going to need adjuvant therapy
It will certainly be relevant in lung cancer, breast cancer, colorectal cancer Maybe even in prostate cancer, although the treatment is pretty bad These are the big 5 in cancer Max adds bladder cancer (the 6th or 7th most common cancer) and melanoma
Maybe even in prostate cancer, although the treatment is pretty bad
These are the big 5 in cancer
Max adds bladder cancer (the 6th or 7th most common cancer) and melanoma

“ It’s going to be useful in the cases where a minority of patients develops recurrence ”‒ Max Diehn

A situation like pancreatic cancer, where the vast majority of most stage I patients will develop recurrence, that’s not the place to start developing this kind of a test Because the tests aren’t ever going to be a 100% perfect No test gives you 75% chance of recurrence This test is useful where there’s a subset of patients that recur For cancers where the vast majority of stage I patients develop recurrence (such as pancreatic cancer), this is not the place to develop this kind of test
A situation like pancreatic cancer, where the vast majority of most stage I patients will develop recurrence, that’s not the place to start developing this kind of a test Because the tests aren’t ever going to be a 100% perfect No test gives you 75% chance of recurrence
This test is useful where there’s a subset of patients that recur
For cancers where the vast majority of stage I patients develop recurrence (such as pancreatic cancer), this is not the place to develop this kind of test
Because the tests aren’t ever going to be a 100% perfect
No test gives you 75% chance of recurrence
When we know the probability of recurrence is high enough, everyone gets adjuvant therapy
Tests are never going to be 100% perfect No test give you a 75% chance of recurrence Stage III colon cancer is another example where we don’t need the test, everyone gets treated Stage I and stage II cancers are a different story
Tests are never going to be 100% perfect
No test give you a 75% chance of recurrence
Stage III colon cancer is another example where we don’t need the test, everyone gets treated
Stage I and stage II cancers are a different story

Peter’s takeaway: we’ve identified one problem that can be addressed with liquid biopsies, identifying the high risk patient who needs adjuvant therapy. Great progress has been made with lung cancer.

What is the state of the art for this problem in breast, prostate, and colorectal cancer?

There are active studies underway
Colorectal cancer might be the furthest along
There are some Medicare approved tests using ctDNA from a company called Natera for colorectal cancer This was recently approved even though proof of benefit hasn’t been shown yet
There are trials in breast cancer to do similar things
This was recently approved even though proof of benefit hasn’t been shown yet

Challenges presented by breast and prostate cancer

1) They have very low levels of ctDNA

There’s difference between tumor types and how much ctDNA they shed

These require very, very sensitive mutation-based methods
Studies are ongoing in breast cancer, Max is not aware of any for prostate cancer
Peter notes, “ Prostate seems to be a very privileged site. Even the GRAIL test… doesn’t screen effectively for prostate because you’re just not getting enough cell-free DNA in the circulation. ”
The development plan for GRAIL was initially focused on breast cancer but they pivoted to other cancers because the early results in breast didn’t go well

Who is the next group of patients Max wants to develop a liquid biopsy for?

Would it be patients who have completed adjuvant therapy (or who were deemed low enough risk to not need adjuvant therapy) but are still higher risk in the population in whom you’re screening for recurrence?

We need results from these early studies to provide proof of principle that this approach works

This would be the next logical step
Max can envision a future where patients at higher risk of recurrence are tested repeatedly, say every 3 months They don’t get adjuvant therapy until the test is positive
They don’t get adjuvant therapy until the test is positive

Maybe this can minimize missing patients and still catch them months before they have clinical recurrence

This would be better because you’re not giving toxic treatment to a big chunk of patients, only those who test positive
A 2nd approach Max is excited about uses the same set of trials but looks at it the other way around

Test patients to avoid overtreatment

One big problem in adjuvant therapy (breast cancer is a good example) is the number needed to treat (NNT) is quite high Most patients are cured by the surgeon and don’t have micrometastatic disease
Second, the the aren’t perfect, some patients with micrometastatic disease don’t benefit
Repeated liquid biopsy testing could help with the overtreatment situation and help treat only patients where there is substantial evidence they still have cancer cells in their body
Most patients are cured by the surgeon and don’t have micrometastatic disease

How we get to the panacea of cancer screening [1:52:00]

We need a different approach to move to the panacea of cancer screening using a blood test

Imagine if every person got a test from their doctor each year which could tell them yes/no they have cancer from 10-20 mL of blood
Further, this test could tell them the organ of the cancer’s origin This would direct further diagnostics
This would direct further diagnostics

Thinking of mutation analysis, is it possible to develop a robust database of mutations for every given tumor?

Over a year ago Max’s group published a paper where they developed a mutation based lung cancer screening method using ctDNA called Lung CLiP CLiP stands for c ancer l ikelihood i n p lasma Published in Nature in 2020, Integrating genomic features for noninvasive early lung cancer detection
This method is purely based on mutations
How it works‒ sequence the plasma and the white blood cells of the patients you’re trying to screen
1) Sequence both the cfDNA, cell-free DNA, and the leukocyte DNA They’ve learned (particularly in older individuals) that there are mutations present in the leukocytes often that have been acquired through age through a process called clonal hematopoiesis So those mutations that are in the white blood cells end up showing up in the plasma most of the time because the white blood cells die in the circulation and release their DNA and that’s part of the cfDNA These mutations are not coming from the cancer; so you want to get rid of these from the testing process
2) Subtract the things that are in the white blood cells
3) Then use a machine learning algorithm that looks at the mutations that are left in the plasma and looks at things like the length of the cell-free DNA molecule (in base pairs, bp) Max and others have shown, that the cancer derived cfDNA molecules are a little bit shorter than that 170 bp 80-90% of the cfDNA in the blood comes from white blood cells, 10-20% comes from solid organs and cancer Probably DNA from the cancer has to traverse longer to get into the blood; cells could die in the tissue and it takes the cfDNA a while to get into the blood where it is exposed to enzymes that cleave the DNA (discussed earlier) It may also be something about the way the histones and chromatin is configured that contributes to the length of DNA
CLiP stands for c ancer l ikelihood i n p lasma
Published in Nature in 2020, Integrating genomic features for noninvasive early lung cancer detection
They’ve learned (particularly in older individuals) that there are mutations present in the leukocytes often that have been acquired through age through a process called clonal hematopoiesis
So those mutations that are in the white blood cells end up showing up in the plasma most of the time because the white blood cells die in the circulation and release their DNA and that’s part of the cfDNA
These mutations are not coming from the cancer; so you want to get rid of these from the testing process
Max and others have shown, that the cancer derived cfDNA molecules are a little bit shorter than that 170 bp
80-90% of the cfDNA in the blood comes from white blood cells, 10-20% comes from solid organs and cancer
Probably DNA from the cancer has to traverse longer to get into the blood; cells could die in the tissue and it takes the cfDNA a while to get into the blood where it is exposed to enzymes that cleave the DNA (discussed earlier) It may also be something about the way the histones and chromatin is configured that contributes to the length of DNA
It may also be something about the way the histones and chromatin is configured that contributes to the length of DNA

They look at the mutation present, what gene is it in, how long is the cfDNA fragment and the machine learning model tells them if they have seen this mutation before in lung cancer and the model ultimately determines a probability that that blood sample was from a patient with lung cancer or not

Max envisions this mutation based screening approach where you’re not actually using a catalog of mutations because that won’t work because every cancer is unique and you can have a mutation in any gene in any position
The machine learning is used to predict what would be a cancer mutation or not
In the follow-up work, he is trying to add more features The more features you have that link with cancer, the better it is Could this be combined with methylation analysis? Can it be combined with other analytes or approaches?
The more features you have that link with cancer, the better it is
Could this be combined with methylation analysis?
Can it be combined with other analytes or approaches?

Max’s takeaway: “ We’re going to move forward to try to ultimately develop the best screening test and it very likely will be a combination of not just methylation or just mutation or just what’s called fragmentomics (which is this size of the molecules or their distribution). It likely will be a combo method. That’s still very much in the early research phase.”

What might be the theoretical limits of this?

Max doesn’t know
One critical experiment (currently underway), is to take the 3rd generation method he mentioned (that has the 1 in a million sensitivity) and apply it to a large cohort of stage I patients and see how many of those patients they can detect dtDNA in, when they know the mutation This is not a screening question but a preamble to it where they can definitively quantitate ctDNA in these patients in stage I This experiment will tell how much ctDNA patients are shedding during stage I lung cancer This will show the specificity of this approach
Max is hopeful they can push it to above the 5% sensitivity discussed earlier
Their study from a couple years ago showed they could get about 20% sensitivity This included stage I non-small cell lung cancer in a validation cohort The specificity was about 98%
This is not a screening question but a preamble to it where they can definitively quantitate ctDNA in these patients in stage I
This experiment will tell how much ctDNA patients are shedding during stage I lung cancer
This will show the specificity of this approach
This included stage I non-small cell lung cancer in a validation cohort
The specificity was about 98%

Back of the envelope math

If someone has a pretest probability of 1%, you’re going to need 80% sensitivity and 99.5% specificity to say that a test matters You need a high enough positive predictive value to do something with Just as important is the negative predictive value; this needs to be high enough to reassure that disease is not present
Max replies, “ It’s possible we can’t get there with this approach. If you had asked this question in the fifties with some of the protein biomarker work, which was exciting, people would’ve said, well, okay, we’re not there but maybe in 10 years we will .”
Currently we don’t know the real distribution of ctDNA because most of the experiments used insensitive methods where most of them are negative
Max is hopeful we can get to at least 50%
Peter agrees, 50-75% detection in early stage I cancer would be a big step forward If you can keep the specificity > 99%
Some liquid biopsy groups have gone with 80 percent specificity Because if we advertise it has a 99% specificity and a 20% sensitivity, but now you drop the specificity to 80, then your sensitivity management goes up by 20% because just the dumb luck part
You need a high enough positive predictive value to do something with
Just as important is the negative predictive value; this needs to be high enough to reassure that disease is not present
If you can keep the specificity > 99%
Because if we advertise it has a 99% specificity and a 20% sensitivity, but now you drop the specificity to 80, then your sensitivity management goes up by 20% because just the dumb luck part

The problem is the negative predictive value goes out the window

The foreseeable future

There are 2 main ways to apply these types of liquid biopsies
1) In cancers that currently lack a screening test, where the bar of success is much lower
Most screening methods aren’t perfect Mammography, low-dose CT Colonoscopy is a bit of an exception, it’s better But these screening methods have been shown to save lives for lung cancer and colon cancer Breast cancer screening is more controversial
The ideal case would be pancreatic cancer (for reasons discussed earlier)
There are other cancers we don’t screen for
Mammography, low-dose CT
Colonoscopy is a bit of an exception, it’s better
But these screening methods have been shown to save lives for lung cancer and colon cancer
Breast cancer screening is more controversial

How this test compares to current screening tests will be important

Currently, the liquid biopsies aren’t as good as those tests

The use of liquid biopsies should start around some practical consideration or health system consideration

We know only a tiny minority of patients who are eligible for low-dose CT currently get one, in the US People are concerned about radiation exposure and don’t get this test even though Medicare pays for it Maybe the problem is the difficulty in booking an appointment or access to the imaging facility
If the patient is already at the doctor and having their hemoglobin A1c tested or whatever, it’s easy to draw another couple tubes of blood for a liquid biopsy This is not the dream application of liquid biopsies but it is helpful
We still need to prove again that these tests decrease cancer-specific death
People are concerned about radiation exposure and don’t get this test even though Medicare pays for it
Maybe the problem is the difficulty in booking an appointment or access to the imaging facility
This is not the dream application of liquid biopsies but it is helpful

Do you know of any companies out there that are in pursuit of this dream scenario where you have a high sensitivity and high specificity test for early stage I cancer detection?

All companies working on this are trying to improve there tests constantly
They’re being run in the CLiA environment and it’s much easier to change the test than if it’s under FDA approval
Max’s group is working hard on this

It’s too early to know how much they’re going to push the bar over what Max showed in that prior paper or what GRAIL showed in their recent papers

How much money is the NIH putting into this?

Great question, Max has never seen numbers for how much funding they are putting into liquid biopsy research It has increased dramatically When he started working in this area (around 2012, their first publication was in 2014) and went to meetings, he would be the only person talking about ctDNA There field was focused on diagnostic imaging and circulating tumor cells Now the vast majority of presentations are focused on liquid biopsies In the NIH study section he sits on (focused on cancer biomarkers), he sees lots and lots of liquid biopsy work submitted and often scoring high
It has increased dramatically
When he started working in this area (around 2012, their first publication was in 2014) and went to meetings, he would be the only person talking about ctDNA There field was focused on diagnostic imaging and circulating tumor cells
Now the vast majority of presentations are focused on liquid biopsies
In the NIH study section he sits on (focused on cancer biomarkers), he sees lots and lots of liquid biopsy work submitted and often scoring high
There field was focused on diagnostic imaging and circulating tumor cells

“ There’s a lot of interest at NCI and NIH; they see the value just like many of us do ”‒ Max Diehn

Selected Links / Related Material

Foresight Diagnostics, developing novel liquid biopsy tests : Foresight Diagnostics | [1:00]

CyberMed, biotech specializing in discovery of biomarkers : CiberMed | [1:00]

Episode of The Drive with Karl Deisseroth : #191 – Revolutionizing our understanding of mental illness with optogenetics | Karl Deisseroth M.D., Ph.D. | Host Peter Attia, The Peter Attia Drive Podcast (January 17, 2022) | [15:30]

Method to identify cancer mutations in cfDNA : An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage | Nature Medicine (AM Newman et al. 2014) | [1:19:00, 2:03:00]

Episode of The Drive with Steve Rosenberg : #177 – Steven Rosenberg, M.D., Ph.D.: The development of cancer immunotherapy and its promise for treating advanced cancers | Host Peter Attia, The Peter Attia Drive Podcast (September 27, 2021) | [1:21:30]

3rd-Generation detection of circulating tumor DNA : Enhanced detection of minimal residual disease by targeted sequencing of phased variants in circulating tumor DNA | Nature Biotechnology (Dm Kurtz et al. 2021) | [1:26:45]

AMA 25 of The Drive on cancer screening : #170 – AMA #25: Navigating the complexities and nuances of cancer screening | Host Peter Attia, The Peter Attia Drive Podcast (July 26, 2021) | [1:31:15]

Effect of low-dose CT screening for lung cancer on mortality, RCTs : Reduced Lung-Cancer Mortality with Volume CT Screening in a Randomized Trial | New England Journal of Medicine (HD de Koning et al. 2020) | [1:34:00]

CLiP, lung cancer screening using ctDNA : Integrating genomic features for noninvasive early lung cancer detection | Nature (JJ Chabon et al . 2020) | [1:53:00, 1:58:15, 2:02:30]

GRAIL’s Galleri multi-cancer early detection test, recent results : The PATHFINDER Study: Assessment of the Implementation of an Investigational Multi-Cancer Early Detection Test into Clinical Practice | Cancers (JA Borgia et al. 2021) | [2:02:45]

Max’s lab website : Diehn Lab | Stanford Medicine

Max’s faculty profile : Maximilian Diehn, MD, PhD | Stanford Profiles

3.5 min video on liquid biopsies vimeo

Recent news about publications by Max’s lab : In the News | Stanford Medicine: Diehn Lab

Liquid biopsy tests are currently offered by :

GRAIL , a pan screen of methylation patterns of cfDNA | not FDA approved
Guardant uses next-generation sequencing to measure ctDNA | FDA approved
Foundation offers a test similar to Guardant | FDA approved
CellSearch is a circulating tumor cell assay, approved before any ctDNA assay | FDA approved]
Natera test detects ctDNA from colorectal cancer | not FDA approved

Max’s recent work on epigenetic expression inference from cell-free DNA-sequencing (EPIC-seq) to classify lung cancer subtypes : Inferring gene expression from cell-free DNA fragmentation profiles | Nature Biotechnology (MS Esfahani et al. 2022)

Max’s review of liquid biopsy analysis of ctDNA for predicting cancer relapse : Detecting Liquid Remnants of Solid Tumors: Circulating Tumor DNA Minimal Residual Disease | Cancer Discovery (EJ Moding et al. 2021)

Max’s review of the technology used for liquid biopsy detection of cancer : Detection and Diagnostic Utilization of Cellular and Cell-Free Tumor DNA | Annual Reviews of Pathology (JC Dudley and M Diehn 2021)

People Mentioned

Patrick (Pat) Brown ( Emeritus Professor of Biochemistry at Stanford and founder of Impossible Foods ) [5:45]
Howard Chang (Professor of Cancer Research and Genetics at Stanford) [11:30]
Karl Deisseroth (Professor of Bioengineering and of Psychiatry and Behavioral Sciences at Stanford) [15:30]
Henry Kaplan Chair of Stanford’s Radiology department from 1948-1972 and director of Stanford’s Cancer Biology Research Laboratory in 1975) [17:45]
Caroline Dive (Professor of Cancer Pharmacology at the University of Manchester, Deputy Director of the Cancer Research UK (CRUK) Manchester Institute) [59:00]
Dennis Lo (Director of the Li Ka Shing Institute of Health Sciences and Professor of Medicine at the Chinese University of Hong Kong) [1:03:30]
Stephen (Steve) Quake (Professor of Bioengineering and Applied Physics at Stanford University) [1:03:30]
Steven Rosenberg (Chief of surgery and head of tumor immunology at NCI) [1:21:30]

Max Diehn earned his Artium Baccalaureus (AB) in biochemical sciences at Harvard in 1997. He earned his MD PhD at Stanford University in 2004, focusing on Biophysics. He completed an internship in internal medicine at Stanford in 2005 and his residence in radiation oncology at Stanford in 2009. He is a board certified radiation oncologist and Professor of Radiation Oncology at Stanford. [ Stanford|Profiles ]

The overarching research goal of the Diehn lab is to develop and translate novel diagnostic assays and therapies to improve personalized treatment of cancer patients. Their main focus is on the development and application of liquid biopsy technologies for non-invasive detection, monitoring, and characterization of human cancers, with a special emphasis on lung cancer. Diehn’s group is an internationally recognized leader in this field and they collaborate with investigators from around the world to apply our assays to diverse clinical problems and patient cohorts. In parallel he also studies primary and acquired resistance to anti-cancer therapies, including radiotherapy, immunotherapy, and targeted agents. He has a particular interest in overcoming resistance mediated by the NRF2 (NFE2L2)/KEAP1 pathway. His research projects always begin by identifying an unmet need in the clinical management of cancer patients. His lab uses next generation sequencing, bioinformatics, genome editing, high throughput screening, and preclinical animal models to address their research goals. Discoveries from his group are currently being tested in multiple clinical trials at Stanford and elsewhere to attempt to translate them into the clinic. [ Diehn Lab ]

Twitter: @max_diehn

Transcript

Show transcript