Breast cancer screening is one of those things where there is a serious risk of overdiagnosis and iatrogenic injury and one has to seriously think about how many such injuries are acceptable per life saved.
I personally think that the targeted use of MR -- which, with contrast, has its own (minute, marginal) risks (broadly comparable to mammography) is appropriate, as the image quality really is an order of magnitude better than plain film alone.
Nevertheless, this rate of iatrogenic harm has to be considered -- heck, just from basic Bayesian probability, you'll know that any imperfect test for a rare disease results in a lot of false positives.
Here in the UK, it's generally accepted that the ultimate rate of 'harm', that is, over-treatment, is about 3:1; "that is one breast cancer death prevented for about every three overdiagnosed cases identified and treated". Both the NHS and most women feel that, knowing this, "accepting the offer of breast screening is worthwhile" -- and I personally agree with them. [both 1]
The argument for increased iatrogenic injury is always brought up with new technologies (and sometimes used to dismiss their use). In this case, presumably, you mean increased amounts of invasive biopsies w/ potential complications, bleeding, pain, etc. Esp with false positives.
The converse thought is, with the increased resolution, image quality and addition of a 3rd dimension over plain film mammographies, how much reduction in false positives iatrogenic can one reasonably expect?
The situation is a bit more nuanced than you suggest. MRI gives you 3D information and better tissue contrast, but much worse resolution. This makes things difficult with micro-calcifications, but really helps with some lesions in dense breasts.
The best specific information comes with a contrast agent, but that's not appropriate for screening for the iatrogenic reasons you mention, but is suited to diagnostic followup.
Worth noting on a big enough screening population, biopsy complications include death by infection, etc. You also have to consider the opportunity cost of the machinery and tech time, etc. So while sometimes approaches are dismissed for this (e.g. contrast agents for screening) it is not done without though.
As an overall health system, too, you really have to evaluate the systemic costs and opportunity costs, particularly with a screening program. Two big variables here are the infrastructure & workflow costs for introducing MRI mammo into a screening program (those machine and tech hours come from some other worthy use) and radiologist attention. Breast screening already struggles with allowing enough time for careful review of 2D data sets, adding more information of a 3D set (potentially additionally!) means you may have to trade off increased cost/time (or reduced coverage) against FN count.
So that's the tip of that particular iceberg. It's not easy, and a lot of smart people have been thinking about this since the 90s.
An interval cancer is one that is diagnosed outside of the regularly-scheduled screening exams.
Example: A woman feels a lump, goes to see her doctor immediately rather than wait until her next scheduled screening, and this results in a cancer diagnosis.
One of the main reasons this happens is because the previous regular screening exam failed to detect the cancer.
Mammography misses cancers more often in dense breasts than in typical breasts. MRI has long been known to be much more sensitive than mammography at finding cancer in dense breasts.
This study looked at whether adding an MRI to a screening mammogram for extremely dense breasts would result in fewer interval cancers. It did, which is not surprising, but this was a relatively small study that resulted in only 4 cancers diagnosed in women who actually underwent the supplemental MRI.
PET isn’t really suited for a screening population due to relatively low availability and high cost (esp. compared to a mammography). Resolution is also bad and it tends to also light up inflammations. So it probably wouldn’t help too much for reducing false positives.
In the case of patients with dense breasts, which in the article undergo MRI, an argument can be made for PET in both availability and cost now that these patients are in MRI territory. Also, organ-specific PET is working on the resolution problem but I'd agree that if you were to put a patient into a run of the mill PET scanner there wouldn't be much clinical usefulness.
It’s unclear to me how screening reduces cancer. That said, this additional screening with MRI seems like it may reliably detect cancers earlier in the population of women with dense breast tissue, who are more likely to develop cancer in the first place. Earlier detection generally results in better outcomes.
Based on a quick look around the web, "interval cancer" means specifically a cancer that develops after having previously received a cancer screen where no cancer or pre-cancer was found.
So basically it reduces "interval cancer", not "cancer", by more effectively finding warning signs.
Yes, the study showed that cancers were less likely to be diagnosed outside of the regular screening exams. The implication is that the added MRI detected some cancers that would have otherwise been missed during screening and then later found before the next scheduled screening.
The benefit of earlier detection really depends on the bottom-line number (and nature) of improved outcomes vs the harm caused by false positives and overdiagnosis[1][2].
Fear of iatrogenic harms is well founded and important. It doesn't look like you even tried to argue otherwise - you just implied that aggressive screening is worth the risk. This is silly and harmful, especially considering cancer screening has never been shown to save lives: https://www.bmj.com/content/352/bmj.h6080
I personally think that the targeted use of MR -- which, with contrast, has its own (minute, marginal) risks (broadly comparable to mammography) is appropriate, as the image quality really is an order of magnitude better than plain film alone.
Nevertheless, this rate of iatrogenic harm has to be considered -- heck, just from basic Bayesian probability, you'll know that any imperfect test for a rare disease results in a lot of false positives.
Here in the UK, it's generally accepted that the ultimate rate of 'harm', that is, over-treatment, is about 3:1; "that is one breast cancer death prevented for about every three overdiagnosed cases identified and treated". Both the NHS and most women feel that, knowing this, "accepting the offer of breast screening is worthwhile" -- and I personally agree with them. [both 1]
---
[1] https://www.thelancet.com/journals/lancet/article/PIIS0140-6...