Experimental Blood Test Detects Cancer Up To Four Years Before Symptoms Appear (scientificamerican.com) 80
An anonymous reader quotes a report from Scientific American: For years scientists have sought to create the ultimate cancer-screening test -- one that can reliably detect a malignancy early, before tumor cells spread and when treatments are more effective. A new method reported today in Nature Communications brings researchers a step closer to that goal. By using a blood test, the international team was able to diagnose cancer long before symptoms appeared in nearly all the people it tested who went on to develop cancer. [...] Kun Zhang, a bioengineer at the University of California, San Diego, and a co-author of the study, and his colleagues began collecting samples from people before they had any signs that they had cancer. In 2007 the researchers began recruiting more than 123,000 healthy individuals in Taizhou, China, to undergo annual health checks -- an effort that required building a specialized warehouse to store the more than 1.6 million samples they eventually accrued. Around 1,000 participants developed cancer over the next 10 years.
Zhang and his colleagues focused on developing a test for five of the most common types of cancer: stomach, esophageal, colorectal, lung and liver malignancies. The test they developed, called PanSeer, detects methylation patterns in which a chemical group is added to DNA to alter genetic activity. Past studies have shown that abnormal methylation can signal various types of cancer, including pancreatic and colon cancer. The PanSeer test works by isolating DNA from a blood sample and measuring DNA methylation at 500 locations previously identified as having the greatest chance of signaling the presence of cancer. A machine-learning algorithm compiles the findings into a single score that indicates a person's likelihood of having the disease. The researchers tested blood samples from 191 participants who eventually developed cancer, paired with the same number of matching healthy individuals. They were able to detect cancer up to four years before symptoms appeared with roughly 90 percent accuracy and a 5 percent false-positive rate.
Zhang and his colleagues focused on developing a test for five of the most common types of cancer: stomach, esophageal, colorectal, lung and liver malignancies. The test they developed, called PanSeer, detects methylation patterns in which a chemical group is added to DNA to alter genetic activity. Past studies have shown that abnormal methylation can signal various types of cancer, including pancreatic and colon cancer. The PanSeer test works by isolating DNA from a blood sample and measuring DNA methylation at 500 locations previously identified as having the greatest chance of signaling the presence of cancer. A machine-learning algorithm compiles the findings into a single score that indicates a person's likelihood of having the disease. The researchers tested blood samples from 191 participants who eventually developed cancer, paired with the same number of matching healthy individuals. They were able to detect cancer up to four years before symptoms appeared with roughly 90 percent accuracy and a 5 percent false-positive rate.
Too sensative? (Score:3, Interesting)
Our bodies contain cancer cells at all times. Normally, our immune systems seek and destroy these cells. It's only when the cancer cells begin to overwhelm the body's immune system that they start to grow out of control, forming tumors. If a test is too sensitive, it might lead to unnecessary treatments, some of which could itself harm a person's health.
Re: (Score:3)
I would trust doctors to instead use the test only as an indication to see if further tests (such as expensive CT scans) are necessary.
That is, if this blood test doesn't become unattainable in comparison.
Re: (Score:3, Insightful)
If so, we can prevent a repeat of it infesting the White House.
If there's one thing destroying America faster than COVID, it's the fucking political civil war.
That didn't start with Trump. And it sure as shit won't end with Trump. In fact, Nothing Will Change is now a campaign slogan.
Go ahead. Keep slinging that shit and see how it helps Americans or the country.
Re: (Score:2)
In fact, Nothing Will Change is now a campaign slogan.
Well clearly nothing will change until Trump is gone and an adult is put back in charge.
Adult? Dementia Joe will have the mind of a child soon. And people still believe he is the best American for the job. Pathetic.
Greed N. Corruption is in charge. Always has been. Always will be. You merely vote for your favorite flavor of bullshit every few years. Nothing will change, and we have politicians selling that to gullible voters.
Re: (Score:1)
Dementia Joe will have the mind of a child soon.
What's the problem, Ronnie 'Alz' Reagan did almost a full term without his brains, and noone noticed.
Re: (Score:2)
Dementia Joe will have the mind of a child soon.
What's the problem, Ronnie 'Alz' Reagan did almost a full term without his brains, and noone noticed.
Reagan once infamously claimed that "Trees cause more pollution that automobiles do."
Believe me we noticed. You just couldn't find enough people who actually gave a shit about that problem.
Just like today.
Re: (Score:2)
Maybe in the USA, where hospitals and prisons are run for profits.
In Canada, health care is a cost to the government, which has to be minimized. Not to say that they won't treat you to keep costs down, but they also won't prescribe drugs and treatments simply to increase your bill because there isn't one
Re: (Score:1)
Maybe in the USA, where hospitals and prisons are run for profits.
In Canada, health care is a cost to the government, which has to be minimized. Not to say that they won't treat you to keep costs down, but they also won't prescribe drugs and treatments simply to increase your bill because there isn't one. They get their pay check for the hours they work, one way or the other.
The only thing more delusional than the "great" United States for-profit healthcare system, is the idea that governments can do that any better. I have family in both Canada and England. I know damn well how a socialist medical system "works". Sure you won't have to worry about a huge medical bill, if you can live long enough to actually get help.
Re: (Score:1)
>that special melting pot gene that magically appears in every human born within America's borders that makes us somehow the world leader in cases and deaths. By a long shot.
It's called anti-intellectualism, combined and a near-total lack of a sense of social responsibility. We don't have a huge number of cases because we're especially physically susceptible to the disease, but rather because we're especially culturally inclined to ignore the advice of experts and refuse to take precautions to slow the
Re: (Score:2)
Many doctors are self-employed, or co-owners of a small medical group.
Medical diagnoses can be hellishly difficult, and many doctors aren't markedly better at it than auto mechanics making car diagnoses. No-one has more at stake in health care than the person being examined. The patient should take the doctor's words as advice, and then research the subject thoroughly.
Re: (Score:2)
Another wonderful effect of capitalism. [slow hand clap]
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
And your evidence that it doesn't result in transmission or illness?
Asymptomatic carriers are a well understood vector for disease - perhaps you've heard the name "Typhoid Mary" before? And from what I've heard, it's estimated that roughly 1/3rd of new COVID cases were infected by asymptomatic carriers.
Such carriers don't get sick, but they also aren't immune - the disease doesn't make them sick, but their bodies don't kill off the infection either, so they can still spread it. And they can potentially rem
Hopefully your doc reads the summary (Score:5, Insightful)
Hopefully your doctor reads at least the Slashdot summary, where it says the false positive rate is 5%. Meaning if a person tests positive, 95% chance they're going to have full-blown cancer if something isn't done - BUT that treatment can be relatively mild at that stage, or a year later when the cancer is definitely there, but small.
Presumably, given the cancer doc has read at least one paragraph about the test on which she is basing treatment, this would be followed by further testing and potentially appropriate treatment at the right time based on risk. In any event, 95% is pretty strong. If I got a positive test, I'd darn sure take a treatment that makes me feel really crappy for a week and knocks out the stage 0 cancer.
Re: (Score:2)
With a sample size of 191, which is smaller than the number of possible cancers, it's really hard to be confident in a 5% false positive rate.
Re: Hopefully your doc reads the summary (Score:2)
Well, it's not gonna be 50%, if that's what you mean
It is too small, yes, and 1,000-10,000 people would be more up my alley, but enough for me to choose doing it if I'd ever need it.
And I figure it's not exactly easy to get both 10k potential cancer patients and the financing for it too.
Re: (Score:2)
You're right, it's not going to be 50% overall. But the percentage is not likely to be evenly distributed. Some cancers may have a much higher false positive rate than others. In other words, it's the kind of thing that would lend itself to edge cases.
It's also still being promoted by the original researcher / company. Others have yet to duplicate this kind of success.
Re: (Score:2)
With a sample size of 191, which is smaller than the number of possible cancers, it's really hard to be confident in a 5% false positive rate.
Summaries, how do they work?
Re:Hopefully your doc reads the summary (Score:5, Insightful)
Consider if 10,000 people are tested and 100 (1%) actually have one of the detected cancers.
An "accuracy rate" of 90% is a "sensitivity" of 90% and means that 90 (90%) of the 100 with cancer would get a positive result and the remaining 10 would incorrectly get a negative result.
A "false-positive rate" of 5% is a "specificity" of 95% and means that 495 of those 9,900 without cancer would incorrectly get a positive result.
Therefore out of 504 positive tests, only 1.8% (9 of 504) of those with positive tests actually have one of these cancers. Presumably, each of those tests would need to be followed up on -- possibly with a procedure that is invasive (such as a biopsy), damaging to the patient (such as PET/CT), and/or expensive. If no follow up is done, there is no reason to do the test on that person in the first place. Since these would generally be detecting very early stage cancer, the followup may not be very accurate either.
So, it would make little medical or economic sense to do this test every year on all adults (where perhaps 1% of that population may have one of these cancers in a given year).
For high risk groups (age, family history, lifestyle) it might make sense as the percent of that population would have a much higher rate of one of those cancers in a given year.
Re: (Score:2)
Consider if 10,000 people are tested and 100 (1%) actually have one of the detected cancers.
Given that both lung and colorectal are included here the actual cancers should be higher than 1%, significantly so.
An "accuracy rate" of 90% is a "sensitivity" of 90% and means that 90 (90%) of the 100 with cancer would get a positive result and the remaining 10 would incorrectly get a negative result.
A "false-positive rate" of 5% is a "specificity" of 95% and means that 495 of those 9,900 without cancer would incorrectly get a positive result.
Therefore out of 504 positive tests, only 1.8% (9 of 504) of those with positive tests actually have one of these cancers. Presumably, each of those tests would need to be followed up on -- possibly with a procedure that is invasive (such as a biopsy), damaging to the patient (such as PET/CT), and/or expensive.
In the civilized world where healthcare is not paid by the individual, cancers are quite expensive. That's why we have things like screening; the cost of screening an entire population is lower than treating the cancer cases that otherwise could be prevented. Picking up 9 cancer cases (the actual number would be higher than your example when screening at-risk sub-populations) out of 504 is a d
Re: (Score:2)
There's a difference between "not paid for by the individual" and "paid for by government, complete with price controls."
The latter slows development, leading to net more deaths as the decades crawl by and invention of cures is slowed.
Proof: Let's do price controls on new iPhones and see how development rates stay the same instead of falling off a cliff!
Re: (Score:2)
In any given year, across the entire population, 1% of the population first contracting (i.e., being detectable) one of these cancers is probably in the ballpark.
MR scans are not (ignoring occasional issues such as someone who has a metal fragment in their eye from an injury long ago and are unaware of it and get an MRI anyway) which is why I didn't inc
Re:Hopefully your doc reads the summary (Score:4, Insightful)
Depends why the false positives happen. If it's because of some temporary condition then you can just repeat the test again in a month or two and see if it goes away. You might even be able to test for whatever is known to cause false positives, say some hormone or something, and correct it.
If this can detect cancer years in advance of where it normally gets detected then time is on our side.
Re: (Score:2)
True. The origins of the false positives would have to be better understood and perhaps they could be mediated. If they could be mediated, the false positive rate would effectively be much lower.
For example if a retest of positives two months later results in 99% of the false positives now coming back negative, the test protocol perhaps could be defined as "One test negative => negative. Two positive tests separated by two months => positive". This would effectively increase the specificity of the tes
Good point - though it's more like 10-20%, not 1% (Score:2)
You make a good point.
When testing for something that is extremely rare a priori, the false positives can be higher than the true positives, even though the false positive *rate* is low.
On the other hand, cancer is not extremely rare. 38% of Americans get cancer at some point. If you didn't pay attention to high-risk groups and tested people at random every five years, something like 10% would be on their way to developing cancer. If you test people over 40 years old, the rate is higher.
You also used the r
Re: (Score:2)
Of course I didn't RTFA and am relying in the summary and title. The title indicates that this test can detect cancer up to four years before symptoms appear.
Assume, for the sake of argument, that on the average it will detect cancer two years before symptoms appear. In this case, screening every four years would result in (roughly) half the people testing negative (ignoring false positives and negatives) and showing symptoms before (or at) the time of their next test. In that scenario, the test would have
Questionable result: Re-test in a year (Score:2)
I note that after looking for methylation at 500 DNA locations, the program classified them into two groups a) expect cancer in the next few years or b) methylation is low, probably no cancer soon.
Obviously there will be patients that are borderline, who have moderate methylation so they are close to the line between "cancer is very likely" and "cancer is noy likely". If I were the doctor, I would Re-test the borderline people 6 or 12 months later.
Re: (Score:2)
Re: (Score:2)
Good catch. I started the analysis with a smaller sample size and ended up with a fractional person with cancer (or a person with a fractional cancer?). I therefore increased the sample size by a factor of ten to eliminate the fraction but failed to propagate that change everywhere :(
This
should read
Re: (Score:3)
"Hopefully your doctor reads at least the Slashdot summary,"
You must be new here, nobody does, not even doctors.
Re: (Score:1)
You need to know more than the false positive rate alone. Rolling a D20 would give you a false positive rate of 5% too.
Re: (Score:2)
Hopefully your doctor reads at least the Slashdot summary, where it says the false positive rate is 5%. Meaning if a person tests positive, 95% chance they're going to have full-blown cancer if something isn't done
NO! It does not mean that at all! This is a very common misinterpretation of false positive and false negative rates. What it means is that if a group of 100 people were all cancer free then about 5 of them would test positive under this test. In order to determine the probability of actually having cancer if you test positive, you need to fold in the a priori probability of having cancer and apply Bayes' Theorem [wikipedia.org].
Re: (Score:2)
Yep, we covered that yesterday here:
https://slashdot.org/comments.... [slashdot.org]
I'm glad that was pointed out.
Re: Too sensative? (Score:2)
I was surprised too ... SJWs with mod points, on MY Slashdot?? :)
Re: (Score:2)
Sounds like a good business model.
Re: (Score:3)
This article clarifies that screening is very good for business by leading to overdiagnosis and overtreatment
https://blogs.scientificameric... [scientificamerican.com]
Two Words (Score:2)
Elizabeth Holmes
Re: (Score:2)
Sell stocks of US Hospitals (Score:3, Insightful)
The US medical system is about profit not health. Hospitals dont make much money if preventive measures are used to prevent cancer. They do make a lot of money from cancer surgery , chemo , radiation etc
If this test works and cancer rates actually fall most hospital corporations will no longer be able to afford to pay multi million dollar CEO salaries.
Re: (Score:2, Insightful)
Insurance companies can pay those salaries, why can't HMOs? They just have to sell the fear of cancer and insurance, not the cancer itself.
Re:Sell stocks of US Hospitals (Score:4, Insightful)
Out of 123K individuals, 1K developed cancer.
Which one is more profitable? Yearly testing of 123K people + early treatment of 1K people, or only later stage cancer treatment of 1K people?
With many business hoping to move to subscription model, it seems to me that yearly testing would be much more profitable, easier to streamline/increase efficiency and generally make more business sense.
Not to mention the potential business opportunity of "up-selling" to 123x more regular customer.
The problem with US business is the extreme short-sightedness of only looking at the next quarter, instead of looking at the next decade.
Re: (Score:3)
The problem with US business is the extreme short-sightedness of only looking at the next quarter, instead of looking at the next decade.
That "US" problem, drives every damn stock market in the entire world.
The problem with analyzing Greed, is we often try and color it, which is stupid. Humans suffer from the Disease of Greed, and that's been raging for thousands of years.
Re: (Score:2)
The problem with US business is the extreme short-sightedness of only looking at the next quarter, instead of looking at the next decade.
That "US" problem, drives every damn stock market in the entire world.
Which is why most of the rest of the sane world do NOT let private for-profit companies takeover the health care of its citizens, and have some form of nationalized health care.
Letting profit dictate health care makes this a US specific problem.
Re: (Score:2)
Its not only the US. India has the same problem too. However in India once in a while the mob lynches a few doctors and hospital administrators which keeps the greed and fear in balance
Re: (Score:2)
Which one is more profitable? Yearly testing of 123K people + early treatment of 1K people, or only later stage cancer treatment of 1K people?
Or continuous treatment of cancer as a chronic disease. If they do it right, they can come up with a treatment that just keeps it at bay, that you have to take once a month for the rest of your life....
(Yeah, that's better than dying, but at some point, profit seeking leads to clearly worse outcomes.)
Re: (Score:2)
I'm really glad we have universal healthcare that is motivated to keep you healthy and detect problems early.
Re: (Score:2)
The US medical system is about profit not health.
A cancer test sounds like a great way to deny people insurance.
Not sure (Score:5, Interesting)
Always wait for a second lab to independently confirm. These journals don't verify what you're saying is true or false, they only validate that what you REPORT is scientifically sound. As in, you have to report that you did all the proper verifications and controls properly. It could be 100% fabricated/made up data. In fact that's happened to me before numerous times .. we often waste months trying to replicate stuff idiots publish. The worst example so far (there are many) was when I wasted a bunch of hours one summer because some fool reported that an Argonaute protein (NgAgo) can be used for DNA editing. Believe me I'm still fucking bitter about that. Not only because it was false but because stupidly it made intuitive sense to me at the time so I wanted to believe it was true and kept trying to repeat the experiment without being more stringent.
False positives (Score:2)
They claim very high specificity; but looking at the graphs it sure seemed like there were more false positives than one would like. I'd hope this tool - if it pans out in further testing - is mainly used for "let's monitor you more frequently" rather than "yup, you should get your affairs in order now".
Still wildly innacurate (Score:4, Interesting)
So a 5% false positive rate, and a 0.8% actual occurrence rate? So a positive test tells you you have a 1 in 7 chance of actually getting cancer? I see value in maybe identifying folks needing more frequent follow up checking, but 6 out of 7 positive results will not pan out but leave those folks freaking the hell out assuming their is a dagger over their heads constantly.
Re: (Score:2)
That is actually not bad numbers. Often it is far worse than one out 7.
The main issue is how the doctors present it to the patient. They should not be saying "you tested positive for cancer". Instead they need to say something like "You have an indicator of cancer, we should do a test to be absolutely sure."
stomach, esophageal, colorectal, lung and liver .. (Score:4, Interesting)
Why does that sound like exactly the ones caused directly by pollution and crap in our food and air?
Most deadly types of cancer (Score:3)
I was wondering why they chose these types of cancer (stomach, esophageal, colorectal, lung and liver) but did not include breast and prostate cancer, which are very common. It seems that they targeted the most deadly forms of cancer rather than the most common. Some numbers from WHO [who.int]:
The most common cancers are:
Lung (2.09 million cases)
Breast (2.09 million cases)
Colorectal (1.80 million cases)
Prostate (1.28 million cases)
Skin cancer (non-melanoma) (1.04 million cases)
Stomach (1.03 million cases)
The most common causes of cancer death are cancers of:
Lung (1.76 million deaths)
Colorectal (862 000 deaths)
Stomach (783 000 deaths)
Liver (782 000 deaths)
Breast (627 000 deaths)
Or breast and prostate cancer are just harder to spot ?
Re:Most deadly types of cancer (Score:5, Interesting)
Male
Female
Re: (Score:3)
This is really interesting information, with some obvious hypothesis from the data (though they may be wrong.
Lung China> USA for men but not women: Air pollution and working in a factory vs home, or is it the far greater rate of smoking - do men in china smoke > women?
Prostate may simply be food, and breast cancer may simply be a combination of Ashkenazi jews (genetic predisposition) plus the older pollution that America filled itself with before the 60's regulated it.
Skin cancer may simply be due to
Re: (Score:2)
Maybe the higher use of squat toilets? We know there's a connection to sitting toilets and hemorrhoids. Perhaps there are other perks to using them we haven't really realized. Or perhaps it's related to a slight difference in life expectancy - data appears to be in the US' favor by a couple years.
Would be interesting to see how other places in the work line up as it might make for interesting research.
Re: (Score:2)
Interesting data. I`ve read promising studies [nih.gov] of frequent green tea drinking slowing any potential prostate progression. As chinese generally take this with almost every meal, this data may support these theories.
Bigger question: then what? (Score:1)
Let's say the test is accurate enough. An individual gets diagnosed 4 years before symptoms appear. Then what happens? Immediate preventative chemo? Change of lifestyle?
Or more likely: "We'll keep an eye on this for the next few years and then we'll be able to start addressing it. Maybe you'll be fortunate and nothing will come of it."
So we have this cancer detection breakthrough (Score:2)
Aaaaand 95% of the posts: "Big Pharma is just going to use this to overcharge us!"
5% of the posts: "This is so good we'll never hear of it again!"
Whenever one of our universities or foundations makes a significant discovery with practical applications, perhaps we would all be making use of it sooner if we just gave it away to China and let them run with it.
Or validity (Score:2)
The researchers tested blood samples from 191 participants who eventually developed cancer, paired with the same number of matching healthy individuals
I see no reason to use the same number of people in the non-cancer group. Ideally you's use the whole 132,000 to build a statistically solid sample. Using 191 healthy people to compare to buys you nothing over a larger amount. There is no magic in that number, or their equality.
Cost, maybe, but not statistical accuracy.
Lets do the math. (Score:2)
So they have 1.6M tests, but 1000 people developed cancer. Of those, about 20% developed cancer that was relevant to the test, so we'll assume the relevant healthy population was 320000 people.
Of those 5% will get a "false positive" test: 16000 people. Of the 200 that will develop one of the tested-for cancers 180 will get a positive test like the 16000 people who won't. And 20 will be negative like the 304000 people who won't get cancer in the relevant period.
Anyway, so now we're stuck with 16180 people wi
Then what? (Score:1)
Now what? Which cancer? How do I (pre-?)treat?
Now I get to spend the next 4 years anxious over my impending disease, with no idea how it will manifest!
I can make predictions, too. I predict that you will die sometime in the future (presuming you are live now).