Why Early Detection Is the Best Way to Beat Cancer | Hacker News Reader

Why Early Detection Is the Best Way to Beat Cancer

wired.com

35 pointsoutdooricon20 comments

I read through the article (first published in 2008, it appears) to see what issues were considered, and what the background of the article author is. "Deputy editor Thomas Goetz (thomas@wired.com) wrote about the Personal Genome Project in issue 16.08. He has a new blog about health and medicine at www.thedecisiontree.com" is what I see at the very bottom of the article. His blog appears to be updated every once in a while, to promote the article author's 2010 book.

The article submitted here is a popular article by a layman, and popular to a fault in that it discusses trade-offs in different screening methodologies, but doesn't even introduce the reader to important concepts like sensitivity and specificity of medical tests.

http://en.wikipedia.org/wiki/Sensitivity_and_specificity

It's very important that HN participants digest the article kindly submitted here with a compare-and-contrast reading of "The early detection of cancer and improved survival: More complicated than most people think"

http://www.sciencebasedmedicine.org/index.php/the-early-dete...

by David Gorski, M.D., an experienced cancer researcher and experienced blogger about science-based medical decision-making for popular readers. His conclusion, "The bottom line is that the ever-earlier detection of many diseases, particularly cancer, is not necessarily an unalloyed good. As the detection threshold moves ever earlier in the course of a disease or abnormality (in the case of cancer, to ever smaller tumors all the way down to the level of clusters of cells), the apparent prevalence of the disease being screened for increases, and abnormalities that may never turn into the disease start to be detected at an increasing frequency. In other words, the signal-to-noise ratio falls precipitously," is a necessary caution in predicting the benefit of early detection of what appears to be cancer.

The American Cancer Society has a good 2012 publication, "Cancer Treatment & Survivorship: Facts & Figures 2012-2013"

http://www.cancer.org/acs/groups/content/@epidemiologysurvei...

that provides more details on what it means to survive a case of cancer.

David was the one who taught me about the Will Rogers phenomenon. Merely by reclassifying something, you can make it look like all categories improve, while overall no improvement happened.

[1]http://en.wikipedia.org/wiki/Will_Rogers_phenomenon

There are huge drawbacks to early detection, though. Surgery, radiation therapy and chemo-therapy are very invasive techniques, doing huge amounts of damage to a person's body as well as to the cancerous cells. It very often is the case that the cure is worse than the disease. Early detection also submits people to the huge mental stress of thinking that they're going to die soon.

And that's the case even if the early detection techniques were perfect. They aren't, far from it. They have a significant false positive rate. Many people who don't have cancer are subjected to the mental and physical stresses of cancer treatment. They then think that this treatment saved their lives, so they say to everybody they know: go get tested!

The other problem is the prevalence of cancer. 80% of men in their 70s have prostate cancer. Some of them will die from it, but most will die from heart disease or another cause. Treating them for prostate cancer will cause a huge decrease in their quality of life for no benefit.

Yes, early detection of cancer is a very potent tool for saving people's lives. But it's not a panacea.

[EDIT: the article actually does discuss most of these issues, I just wanted to try and balance things. Read the article.]

Except that early stage intervention tends to mean that less invasive techniques can be used. Removing a small cancerous lump might only require keyhole surgery, while in later stages it might require significantly more invasive surgery. Sure, all surgery carries some risk (much of which, I believe, comes from having to use anesthetics rather than the surgery per-se) but the less invasive it is, the faster the patient can recover from it. Similarly, early stage cancers might call for far less agressive chemotherapy drugs with reduced side-effects and requiring fewer doses.

Still, the question of early detection is an interesting one, as is the question of when intervention is necessary. Your comments about the stress of knowing that you might have cancer is interesting in light of a paper which my father published a few years ago about whether to restart treatment with people who had relapsed ovarian cancer when their tumour marker levels started rising again, or when they started to exhibit symptoms. As it turned out, while treating people once their tumour markers crossed a 'red line' meant that intervention started earlier, waiting for symptoms to be exhibited resulted in nearly identical life expectancy and also a higher quality of life as they'd spent less time on chemo. Consequently, it also raised the question of whether or not it's worth regularly checking tumour marker levels after their first cancer had gone into remission, as having to have blood tests every three months and waiting for results each time is stressful in itself.

> Surgery, radiation therapy and chemo-therapy are very invasive techniques

A not so well-known fact to drive that home: cytostatics (the bulk of traditional chemotherapy) are derivatives of some of the nastiest chemical weapons developed in WWI, specifically mustard gas.

More than that, Mustard Gas was fairly important in the development of chemotherapy after US Army Doctors noticed it's effect on white blood cells following an air raid in WW2 when a ship carrying Mustard Gas was bombed. From there, in part, was sparked the idea of using similar drugs to attack cancers.

http://www.the-hospitalist.org/details/article/243771/From_W... http://en.wikipedia.org/wiki/Sulfur_mustard#Development_of_t...

Indeed, it's worth reading some of the stuff people like Otis Webb Brawley have written on the subject. Early screening for prostate cancer has indeed reduced the deaths attributed to that disease — because the resulting treatments are killing patients before the cancer could. It's not always as simple as it seems.

I have lived with canner in two loved ones very close to me. My wife and my Dad. My wifes TNBC was caught early and and the treatment was horrific but she has survived. I wasn't so lucky with my Dad as he came out of the gate with prostate cancer, stage IV, earlier detection in his case would have given me the time for fishing and the other things we'd put off for his retirement.

There are a number of reasons why early detection is often oversold as a way to improve survival rates (which is not to say that early detection is bad, it just might not always help). The article doesn't seem to reference all of them directly, though I'll admit to skimming a bit. Here's a brief list, paraphrased from here: http://www.ncbi.nlm.nih.gov/books/NBK20938/

Lead-Time Bias: The earlier you detect, the earlier the clock on your survival starts. 5/10 year survival rates go up simply because the cancer is at an earlier stage when we started tracking how long people survive, not because early detection necessarily lead to more effective treatment.

Length Bias: More screening means a higher probability of finding cancers which would have stayed relatively benign for a very long time (as opposed to cancers presenting very obvious symptoms). Survival rates appear to go up simply because you're detecting less threatening cancers. I believe the article references this idea.

Overdiagnosis: Heavy screening means finding cancers that may never have become threatening at all (as the referenced link states, an extreme example of length bias). Survival rates go up, but not because screening helped. It's been suggested that countries which screen more often have better survival rates simply because they find cancers which would have benignly resolved themselves and gone unnoticed in countries which screen less often.

Patient Self-Selection: Patients who seek out screening may be more likely to live healthy lifestyles.

I've long thought that machine learning would be a great tool for detecting cancer and other diseases. There is a small but growing body of research to support this idea, but very few practical implementations. [1] [2] [3]

Ultimately detection is an issue of pattern recognition, and computers are really good at that. The two biggest barriers are the amount of data that needs to be collected and annotated, and the sensitivity of medical data. If someone reputable were to start a large scale data collection effort, we could make huge strides toward identifying some of the key cancer indicators. It would even benefit people who can't afford top quality medical care (servers are cheaper to deploy than doctors).

Cancer is the ultimate medical opponent. It comes in so many forms and can change into so many more. Many cancers spend 10-20 years as harmless lumps of cells before doing any serious damage. It seems like the best bet is to every tool available to find them long before they become a problem.

[1] http://cs229.stanford.edu/proj2011/Planey-Machine%20Learning...

[2] http://www.mendeley.com/research/data-analytic-strategy-prot...

[3] http://www.popsci.com/science/article/2011-11/new-computer-m...

The concept of early detection or screening is very attractive, but for many cancers the results have been disappointing:

Prostate cancer: (not effective) http://summaries.cochrane.org/CD004720/screening-for-prostat...

Breast cancer: (evidence conflicting, minimal improvement) http://summaries.cochrane.org/CD001877/screening-for-breast-...

And as other commenters have noted, early detection does not mean the ultimate outcome will be improved -- some cancers have a spectrum of presentations, with some persons having more slowly growing / indolent tumors that can be detected and treated (but may not have actually been likely to harm the person harboring the lesion), while other persons develop rapidly growing and very aggressive tumors that will be fatal (but are hard to detect with screening because they develop and spread rapidly). Only carefully constructed and controlled studies can determine if screening for a particular cancer with a particular screening test actually improves survival or reduces morbidity -- short term or uncontrolled studies can be very misleading.

Also the whole subject is subject to anger and political correctness issues -- some good hearted persons want to believe that some screening metholodogy must help because they want it to help, the evidence be damned! So there can be more heat than light ... anybody who says a specific screening test not helpful can be accused of running a "death panel".

In the case of ovarian cancer, as described in the article, it's likely that "early" ovarian cancer is a separate disease from disseminated ovarian cancer.

Assuming that one leads to the other has lead to a large research effort that has failed to help. Screening is ineffective at best, and probably harmful as it:

picks up the earlies that would have been detected and cured in an ad hoc fashion; picks up a host of benign pathology, removal of which leads to iatrogenic harm; fails to pick up aggressive cancer which spreads almost immediately and may even arise in multiple locations at the same time.

The current best methods for prevention are using the contraceptive pill, removing the Fallopian tubes (50% of "ovarian cancers" are probably tubal), and removing the tubes and ovaries of people in high-risk families after childbearing.

Screening does not work.

Sources to back up the idea that early stage ovarian cancer is separate from late state ovarian cancer?

The article mentions the hypothesis that the particularly malignant ovarian cancers are actually fallopian cancers that have spread.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2928939/

My unfounded $0.02 is that the best way to beat cancer is the same as the best way to beat HIV: multidrug targeted molecular therapy to prevent resistance. This is currently mostly a pipe dream as our targeted molecular therapies are few.

Early detection is useful if you can treat appropriately. Early detection of HIV (where 'early' means no AIDS-defining illness) is an example here. Early detection is a tool, but it's not the point. It's the debates, but not the election.

Again, I am not a cancer biologist but the parallels have long struck me. The goal IMHO is to make cancer a typical chronic disease that you live with but can tolerate for decades. Perhaps I am not dreaming big enough, but I think this would be a huge stepping stone. If you can cure a cancer already, of course, by all means do so.

The problem with cancer is that, much like with the HIV virus, the immune system is fooled into thinking the cancerous cells are part of the human body.

From what I hear researchers are optimistic that any such disease can soon (i.e. within the next 20 years) be cured by injecting blood-cell sized nanobots that can reliably target the foreign cells or viruses in your body and kill them (and, unlike chemotherapy, don't destroy the good parts, too). I think the chronic disease is just a crutch that'll soon be obsolete.

Cancerous cells are part of your body — they've just mutated such that they don't respond to signals that tell them to die or slow their rate of replication. The reason your immune system doesn't recognize them is because they are literally part of you. This is why specifically targeted cancer cures are so hard: there's almost nothing that distinguishes cancer cells from normal cells.

Single Page Version: http://www.wired.com/medtech/health/magazine/17-01/ff_cancer...

The article concludes with the hope that Contrast-Enhanced Ultrasound (CEUS) could be extremely useful as another step towards reducing false-postives. This article was written in 2008 and even though it's 4 years later, according to the Canary Foundation's website ( http://www.canaryfoundation.org/research/ovarian-cancer-clin... ), they are STILL waiting on FDA approval for the CEUS microbubbles...

Reply on news.ycombinator.com