Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Blood test that finds 50 types of cancer is accurate enough to be rolled out

theguardian.com

Post reply on HN
Print view View original

191–200 of 389 posts

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#191

Post by giantg2 »

elif wrote: post #177
giantg2 wrote: post #171

If it's detecting 50 types of cancer and the patient doesn't have symptoms, do they just do a full body MRI to find the source? If so, why not just do full body scans which can find other issues, like aneurysms or the other almost 50% of positives that it missed? I get that cost is a big issue, but it seems like the test is missing a lot and you might get more bang for the buck with a periodic MRI from the perspectiv…

I mostly agree. I had a CT scan of 1/3 of my body for a specific issue, but as a consequence I got very high confidence that I was cancer-free in that portion of my body. It is such a good feeling that part of me wants to pay out-of-pocket to get the rest checked. I think though that this test has massive value in earlier detection, cost, and remote lab work.

Dude, CTs can be risky. They use ionizing radiation, which increases your risk of cancer. It's really only an issue with repeated use, but I think there's a study out there claiming 50k early deaths each year can be attributed to repeated use of CTs. I don't have any medical background, but personally I would prefer imaging that does not use ionizing radiation when possible, just to save up some "credits" for when I may need to use the other type (like CT for possible stroke).

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#192

Post by jacquesm »

Silhouette wrote: post #183
jacquesm wrote: post #123

Earlier quoted context omitted.

No, that is not the assumption. The assumption is that the only way in which such a test will be useful is by applying it absent symptoms to the population en-masse aka screening. And that - no matter who good the test, and no matter how early - leads to a decrease in positive patient outcomes. This is established medical science, and it pains me to have to continue to point out the same thing over and over again, bu…

And that - no matter who good the test, and no matter how early - leads to a decrease in positive patient outcomes. You keep writing comments in absolute terms and talking about evidence, but how do you reconcile your position with the results of successful screening programmes like cervical smear testing? Detecting and treating high risk HPV before it causes changes that can turn into cervical cancer has dramaticall…

Yes, but that's one specific cancer. There are a few others for which this is the case and absent genetic data a few that are borderline cases (notably: breast cancer, where the presence or absence of a mutation is a very relevant bit of data).

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#194

Post by starkd »

"It correctly identified when cancer was present in 51.5% of cases, across all stages of the disease, and wrongly detected cancer in only 0.5% of cases."

Doesn't this seem kind of low? Just a bit better than a coin flip. Of course, it rises to 65% and 87% for certain circumstances and the false positive rate is low, but it seems like a lot of cancers could fail to be detected and give a false assurance patients are cancer free. When symptoms emerge later they may be less concerned with getting it checked out. Is this in line with standard performance of tests?

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#195

Post by jacquesm »

dekhn wrote: post #185
jacquesm wrote: post #80

Earlier quoted context omitted.

The problem is not my mindset, but your understanding of the subject material. Patient outcomes have very little to do with the quality of the tests, even a 100% accurate test would not necessarily lead to improved patient outcomes because these are defined independent of the tests. You are treating this like a software problem, but it isn't, it's a medical problem, and medical problems tend to be complex because the…

I think you should stop trying to derate people who you think know less than you. It's very unattractive. Be aware there are folks with extensive public health experience (that would be me) watching you insist you're right. While I totally appreciate what you're trying to do (help software engineers understand why medicine is complex), please do understand that GRAIL is tightly integrated with the PH community, did t…

That's perfectly accurate: the problem is that people tend to run with things like this as though they are the miracle that everybody has been waiting for and that is not the case, it is a very important development but in the right hands and not as a tool that will lead to - yet another - round of disappointment in the 'war on cancer'. The whole thread has devolved into people arguing from hope rather than from facts. This set of tests is a very useful thing to have. But the 50% true positive rate, 18% true positive rate for stage I cancers and the false positive rate combined make it at present - as far as I understand this - a tool that could when used as a mass screening tool easily do more harm than good.

I'm more than willing to be convinced otherwise but with relevant data. FWIW I've been following this particular development closely because it has direct implications for a start-up that I have had a lot of contact with that is also in the early detection space and they were adamant that the combination of factors is such - and so complex - that test accuracy is trumped significantly by absence of symptoms in otherwise healthy patients.

Your input on this would be greatly appreciated.

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#196

Post by cortesoft »

melling wrote: post #126
jacquesm wrote: post #56

Earlier quoted context omitted.

Your belief has no representation in medical science, which is mostly evidence based. A lot of data has been collected on this, studies (many) have been conducted and the general consensus is that more testing absent symptoms does not lead to improved patient outcomes. That you want to have some kind of theoretical argument in the face of this evidence might be interesting to you but it isn't to me.

Less information is better? That would be a bit of a paradox in science. I suppose we might take 2 steps forward and 1 step backward in the short term. As medical science advances, hopefully we can address any shortcomings from the additional early knowledge. I noticed that pancreatic is on the list. This cancer is almost always fatal because we can't detect it early.

> Less information is better?

This really depends. Information that does not help make a good decision is just noise. It might seem like diagnostic information should always help make a good decision, but that isn’t always the case. If the false positive rate is higher than the base rate, a positive test would be more likely to be wrong than right, even with a very high accuracy.

https://en.wikipedia.org/wiki/Base_rate_fallacy

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#198

Post by Silhouette »

jacquesm wrote: post #192
Silhouette wrote: post #183

Earlier quoted context omitted.

And that - no matter who good the test, and no matter how early - leads to a decrease in positive patient outcomes. You keep writing comments in absolute terms and talking about evidence, but how do you reconcile your position with the results of successful screening programmes like cervical smear testing? Detecting and treating high risk HPV before it causes changes that can turn into cervical cancer has dramaticall…

Yes, but that's one specific cancer. There are a few others for which this is the case and absent genetic data a few that are borderline cases (notably: breast cancer, where the presence or absence of a mutation is a very relevant bit of data).

So you do agree that evidence-based screening programmes can be effective then? In that case, I'm sorry but I don't quite understand the point you're trying to make here.

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#199

Post by handrous »

carlmr wrote: post #27
xwdv wrote: post #6

Earlier quoted context omitted.

My experience is people don’t really care much about their bodies beyond the aesthetics. Most treat it like a dumpster and ingest whatever crap feels good and hope for the best. Caring too much makes you some sort of health freak in their eyes, with way too much time on your hands.

I'm not sure this is true. At least in my friend circle most people put a lot of care into their nutrition and fitness. Maybe talk to your friends about it and get them to be healthier?

How do you think we get the rates of diabetes and obesity we have in the US, without that being true for most people in most places and in most social circles? Rich social networks tend to live much healthier lifestyles than average. Normal is fried starch and tens to hundreds of ounces of soda more days than not, and exercise is that maybe you golf sometimes or play in a seasonal company softball league (drinking way more calories than you're burning, almost certainly).

Re: Blood test that finds 50 types of cancer is accurate enough to be rolled out

#200

Post by iandanforth »

jacquesm wrote: post #80
iandanforth wrote: post #76

Earlier quoted context omitted.

I detest this mindset. It is incredibly counter productive. "Our tests are bad, so lets not test" is not a thought worthy of respect. This states that you know about a problem but want to continue to ignore that problem. Reprehensible. No, we should take the exact opposite approach. Test everyone constantly until such methods become both cheap and powerful. So yeah we all have some cancer load but if the diagnostic o…

The problem is not my mindset, but your understanding of the subject material. Patient outcomes have very little to do with the quality of the tests, even a 100% accurate test would not necessarily lead to improved patient outcomes because these are defined independent of the tests. You are treating this like a software problem, but it isn't, it's a medical problem, and medical problems tend to be complex because the…

You don't know what you're talking about.

Before my software career I spent a few years working in a diagnostic medical field. Specifically osteoporosis testing. I worked both in a research capacity at Stanford looking at osteoporosis in older men (not pretending to be the PI here) and in a day to day testing clinic. So I've seen exactly what happens when you test a cross-sectional asymptomatic sample of the populous and what happens during the normal course of referred testing.

Low bone mass at the spine, hip, heel, and forearm as measured by DXA are correlated with increased risk of fracture, but it's only a correlation. Some people have resilient architecture which looks porous on an x-ray but only leads to serious fracture much later in life.

Because the current diagnostic tests are set up with levels like 'osteopenia' and 'osteoporosis' the reaction to clinical referrals was most often treatment. Some of those treatments have serious side effects like necrotizing impacts on the jaw. However the reaction to testing in a large asymptomatic population was much more likely to be an increase in preventative behavior (exercise) or no treatment except in extreme cases. While our study was exploratory and didn't have a treatment cohort (we cared about the impact of sleep quality on bone mass) we got to see a lot of older men who if they were referred to a clinic might have received treatment because that's what clinics do. Instead we had sufficient data to discuss what's normal and what isn't. For a time we were the leading experts in the world on what "normal" meant.

Because I've conducted these tests myself and seen data from hundreds of experimental and clinical patients I feel comfortable contrasting the two. The problem is the clinical medical field reacting to a lack of data with over-prescription of treatments.

Post reply on HN