Whole body MRI

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

sloh

Full Member
15+ Year Member
Joined
Mar 31, 2008
Messages
1,746
Reaction score
2,514
There’s a thread over in the EM forum discussing the merits and risks of whole body MRI’s. You guys here are a smart bunch and way more knowledgeable than us when it comes to cancer medicine. Could you guys weigh in? Link to thread below:

3 years out - officially a millionaire and one step closer to saying goodbye to EM

Members don't see this ad.
 
So, we are talking about early detection of cancer with whole body MRI?

This is certainly something to think about.
However, I do not think there is any evidence pointing at enhanced OS and the number-of-MRIs-needed-to-be-done will be high.

I also think that cancer may very well not be the sole issue here.
Catching an abdominal aorta aneurysm early on for instance, may also be beneficial.
 
So, we are talking about early detection of cancer with whole body MRI?

This is certainly something to think about.
However, I do not think there is any evidence pointing at enhanced OS and the number-of-MRIs-needed-to-be-done will be high.

I also think that cancer may very well not be the sole issue here.
Catching an abdominal aorta aneurysm early on for instance, may also be beneficial.
Whole body MRI would have limited potential for finding early cancers. Lots of technical issues. Cancers in different organs often require different sequences to visualize. Resolution of extended field of view imagers is questionable...especially for finding small subtle findings. Many cancers are invisible without contrast and the timing of contrast for tumors in different organs is different. Beyond that, there is the issue of the herculean effort it would take to carefully evaluate a whole body MRI looking for evidence of early cancers in every possible organ. AI would probably be essential here.
 
Members don't see this ad :)
Whole body MRI would have limited potential for finding early cancers. Lots of technical issues. Cancers in different organs often require different sequences to visualize. Resolution of extended field of view imagers is questionable...especially for finding small subtle findings. Many cancers are invisible without contrast and the timing of contrast for tumors in different organs is different. Beyond that, there is the issue of the herculean effort it would take to carefully evaluate a whole body MRI looking for evidence of early cancers in every possible organ. AI would probably be essential here.
Indeed. One would likely need to focus on certain organs.

Perhaps it could be hepful in order to find
- small HCCs
- IPMNs that may evolve to pacreatic cancer
- early bladder cancer (although hematuria happens early here too)
- small sarcomas
- low grade gliomas

The overall incidence of these tumors is however not large, thus you won't find many, anyhow.

You would certainly have a ton of incidental findings, which would trigger additional follow-up, think alone meningeomas, for instance.
 
Whole body MRI would have limited potential for finding early cancers. Lots of technical issues. Cancers in different organs often require different sequences to visualize. Resolution of extended field of view imagers is questionable...especially for finding small subtle findings. Many cancers are invisible without contrast and the timing of contrast for tumors in different organs is different. Beyond that, there is the issue of the herculean effort it would take to carefully evaluate a whole body MRI looking for evidence of early cancers in every possible organ. AI would probably be essential here.

Sounds like something an ambitious concierge radiologist with the capital to buy an MRI and market to affluent hypochondriacs could do...
 
I’ve thought a lot about this since reading Outlive by Peter Attia. This is already out there with companies like Prenuvo and is envogue with ultra high net individuals. Would be nice to know if I had a sarcoma, pancreatic cancer or brain cancer brewing. I know it’s not evidence based, but I think we can all agree that your only chance with some of these cancers is super early detection.
 

Should be mandatory reading for doctors. Not a popular idea, but screening for most cancers actually has very little data that shows it improves outcomes. Most data showing screening benefits suffers from lead time bias (Lead time bias - Wikipedia).
1737052576395.png


In addition, our immune system actually has the remarkable ability to take care of many subclinical cancers that we are not even aware of (ie see what happens to skin cancers in transplant pts). Not sure chasing down asymptomatic incidentalomas will have the intended effect. The bar for introducing interventions with potential side effects in asymptomatic healthy people needs to be very very high
 
Why not draw a TSH at every OTV.. and a D-Dimer while you are at it? Don't want to miss a PE, do you?

Screening tests are helpful to the extent that the test for specific problems in an at-risk population where early detection can yield better outcomes. As far as I know, all of our screening tests became SOC because of trials showing benefits in survival (or a reasonable surrogate)

...but to even consider running a clinical trial, you need to be able to answer the following questions

1) who should get this test? (i.e. eligibility criteria)

2) what specific abnormalities are we looking for? (i.e. what would be a "positive" result -would need to be defined on an organ-specific basis)

3) what is the required workup for a "positive" result? (should all liver lesions be biopsied? what about a dural based brain lesion? will insurance cover the workup?)

4) what specific endpoint(s) would a trial need to have to demonstrate a benefit? (OS? may need to wait a REALLY long time if we are screening 20 year olds)

5) What are the negative consequences to testing (morbidity from workup? financial toxicity of necessary procedures)

6) and if there IS an net benefit... what is the number-needed-to-test to realize it?
 
One other major issue that I have seen in most of the screening trials is frequency of the tests.

Some of the lung cancer detection trials ran CTs for 5 years, for instance the Nelson trial ran a CT at baseline, after 1 year, after 3 years and then after 5.5 years from baseline. Then, they stopped. 4 scans over 5.5 years

So, you had 60 year old that went into that trial, had his CTs for 5.5 years and then no more. Great, now he is 65 years old. Is his risk of developing lung cancer so low, that it's not worth running another CT scan a couple of years later? Bear in mind, practically all of the trial participants were smokers.

About 1% of all scans in that trial (0.7-1.1) on each round of scans found a lung cancer (scan 1: 0.9%, scan 2: 0.7%, scan 3: 1.1%, scan 4: 0.8%).

Meaning, if you scan a smoker every couple of years, there's a 1% chance that your CT will find cancer.

So, why don't we simply keep on scanning them until they are (I don't know...) 80?
 
One other major issue that I have seen in most of the screening trials is frequency of the tests.
Seems to me that this is just an artifact of how we collect evidence in medicine. For a given condition (e.g., lung cancer) and intervention (e.g., CT), the people running the trial really want a positive result (p<0.05) so that they are "successful" and so that the world has learned something. The endpoint should be something like cause-specific mortality or OS, and these are hard, so you want to stack the deck by (1) including the high-risk individuals only (e.g. >30 pack-years) and (2) screening pretty frequently.

OK, so even when you do this, most trials still fail. This is why you stack the deck. But let's say you do succeed. How do you go back and justify expanding the eligibility pool or screening less frequently? Both of these knobs make sense from a "scarce resources" standpoint but it becomes hard to power these studies, so we have to reach into non-randomized observational data. You might want to conclude from your trial to screen annually until 80, but that was never explicitly tested, nor is it practical to test this in a trial setting.

Same reasoning affects chemo dosing. Maybe we could get a better benefit/adverse event ratio by lowering dose, but what pharma company would want to risk a "Type II error" with a lower dose in the trial? Success in the primary endpoint trumps all considerations. But once approved it becomes harder to back out what the "right" dose is because the standard of care is already established.
 
The endpoint should be something like cause-specific mortality or OS, and these are hard, so you want to stack the deck by (1) including the high-risk individuals only (e.g. >30 pack-years) and (2) screening pretty frequently.

Indeed, if you assume that it's crucial to catch a lung cancer at 0.4 cm and not at 1.4 cm. But what if one designed a trial where CT screening for lung cancer is done in 5 year intervals instead, for instance at 60,65,70,75,80 years of age. Would that trial also perhaps be positive?
 
Indeed, if you assume that it's crucial to catch a lung cancer at 0.4 cm and not at 1.4 cm. But what if one designed a trial where CT screening for lung cancer is done in 5 year intervals instead, for instance at 60,65,70,75,80 years of age. Would that trial also perhaps be positive?
Many years ago I spoke with some of the investigators on NLST. One of them said something like "we always were expecting to find that 5 mm lesion that would grow large and dangerous in the next round but we never found one." Someone at the table pressed on this question -- wouldn't every 30 mm lesion be 5 mm at one point? and he softened a bit saying, "ok there might have been one, but the point is that this didn't really happen." This experience is reflected in LUNG-RADS now where, in a screening population, sub-6mm nodules are just ignored. But going behind the scenes, my impression is that *when the trial was designed* they thought that annual screening would allow them to catch some of these tumors when they were very small and still progressing, but this didn't materialize as expected.

Possible that there's some telephone game happening here, lung screening is not my main thing, but that's what I took from these conversations.

OK, putting that history aside. You expect that if you screen every 5 years, you will catch at least as much per screen as if you screen annually, because you are "accumulating" more of those slower growing tumors. So it stands to reason that a different version of NLST that screened every 5 years instead of every year would have also been positive, probably more positive and with more signal. The problem now is that you have to wait 20 years for results, and nobody wants that. Even if you were going to wait 20 years for results, nobody wants to find p=0.07 at the end and you all look around and say, "if only we listened to Dr. X and screened every year, it would definitely have been positive now."
 
Many years ago I spoke with some of the investigators on NLST. One of them said something like "we always were expecting to find that 5 mm lesion that would grow large and dangerous in the next round but we never found one." Someone at the table pressed on this question -- wouldn't every 30 mm lesion be 5 mm at one point? and he softened a bit saying, "ok there might have been one, but the point is that this didn't really happen." This experience is reflected in LUNG-RADS now where, in a screening population, sub-6mm nodules are just ignored. But going behind the scenes, my impression is that *when the trial was designed* they thought that annual screening would allow them to catch some of these tumors when they were very small and still progressing, but this didn't materialize as expected.

Possible that there's some telephone game happening here, lung screening is not my main thing, but that's what I took from these conversations.

OK, putting that history aside. You expect that if you screen every 5 years, you will catch at least as much per screen as if you screen annually, because you are "accumulating" more of those slower growing tumors. So it stands to reason that a different version of NLST that screened every 5 years instead of every year would have also been positive, probably more positive and with more signal. The problem now is that you have to wait 20 years for results, and nobody wants that. Even if you were going to wait 20 years for results, nobody wants to find p=0.07 at the end and you all look around and say, "if only we listened to Dr. X and screened every year, it would definitely have been positive now."
RE: “<6 mm is ignored”. That is not how lung rads works.

6 mm is the threshold size for going from LR 2 to LR 3 at baseline scan. Less than 6 mm means “continue screening” aka see you next year. It does not mean that nodule is ignored on subsequent time points.

whatever the NLST guys were saying, I can tell you as a reader of LCS, there’s so much randomness with lung cancer because there are multiple histologies lurking in the scan. You can have the small nodule which grows over time a la adenoma-carcinoma sequence of the colon, or you can have the rapidly growing SCC which just appears out of nowhere within 1 year. I used to be a hater of the “oh man another patient trapped in nodule follow up hell” but I have seen nodules that seemingly sit unchanged for years then suddenly blossom into cancer. Maybe the immune system held it in check and then they got immunosuppressant therapy for something else and bam, cancer.


As far as the MR screening thing, I say what I said in that thread: find a rad owned imaging center and pay cash for diagnostic exams of whatever you are concerned about. Don’t spend $2500+ on prenuvo crap image quality. I promise you can get a whole body diagnostic MR for cheaper than what prenuvo and friends are charging. I’ll send you links to an imaging center taking reservations today.
 
Top