SoylentNews is people
SoylentNews
Today, the majority of cancers are detected on the macroscopic level, when the tumor is already composed of millions of cancer cells and the disease is starting to advance into a more mature phase. But what if we could diagnose cancer before it took hold- while it was still only affecting a few localized cells? It would be like putting a fire out while it was still just a few sparks versus after having already caught on and spread to many areas of the house. An international team of researchers, led by ICFO- Institute of Photonic Sciences in Castelldefels, announce the successful development of a "lab-on-a-chip" platform capable of detecting protein cancer markers in the blood using the very latest advances in plasmonics, nano-fabrication, microfluids and surface chemistry. The device is able to detect very low concentrations of protein cancer markers in blood, enabling diagnoses of the disease in its earliest stages. The detection of cancer in its very early stages is seen as key to the successful diagnosis and treatment of this disease.
(Score: 1) by Immerman on Friday May 23 2014, @12:45AM
You're asking the wrong question. The correct question is: how do you inform the dog about the molecule you're looking for. Give him a chemical diagram? No. You give them a bunch of samples of "things to react to" mixed with samples of "things to not react to", and then train them to react to the appropriate samples.
Obviously to do that with cancer you need known cancer patients healing with the training of each dog, or at least biopsies giving off the proper odor, but once you have that then the training doesn't depend on us knowing what exactly the dog is smelling, only on the dog being able to figure out the common element in the samples it's being asked to identify.
(Score: 2) by opinionated_science on Friday May 23 2014, @02:13PM
That is my point. Read the study. There was no control for "not cancer" types. The cancer patients were all 20 years older.
Also in ANY clinical trial it is important to prevent cross contamination. The fact they even had a person in the room, suggest very poorly controlled,especially environment when one is primarily concerned about smell.
Please understand I am sympathetic to the idea that there are new methods to advance clinical medicine, and the phenomena of dog/cat/rat/mouse smelling a disease is not *entirely* whacky. But what's the molecule? What is the discriminator? Without that this study means nothing, because it is not possible to ask the dog "so rover, how much on a scale of one to ten did you smell cancer?"
We train rats for explosives and dogs for drugs, because we have isolated quantities of these substances and they give off an odour, which can be tested independently. More importantly it is possible to gauge a threshold of detection, since we can isolate the substance, and then present it to rover.
The 2012 article claims that cancer cells give off an odour (via aspiration), detectable by a dog.
I challenge this on the basis their articles data does not support such a conclusion. Cancer cells may WELL give off an odour, but this article does not show, or explain this.
It is very concerning, because I saw this trending and it gives the impression this is solved, and there are always people out there to exploit that by blurring the lines of fact/fiction.
(Score: 1) by Immerman on Friday May 23 2014, @10:58PM
I won't speak for the current article, but there have been several studies over the last century showing that dogs can detect cancer and some other diseases. As for a qualitative assessment, why would you need that for a preliminary screening? Like preganancy you can't have a "little bit" of cancer, tuberculosis, etc. If it's there, you want to get on top of it ASAP.
For animals detecting disease let me offer up those same rats detecting tuberculosiso -detect-land-mines-can-now-also-diagnose-tuberculo sis/ [thedoctorschannel.com]
http://www.thedoctorschannel.com/view/rats-used-t
And even if using an "electronic nose" such as this:
http://www.popularmechanics.com/science/4220196 [popularmechanics.com]
you're not going to (initially) program it with the target molecules - you'll do screenings of lots of people known to not/have the disease in question and look for commonalities. Just like a dog/rat would do. All the electronic nose gets you is the ability to copy that knowledge to additional "noses". Admittedly that's a huge advantage compared to training rats individually, but doesn't fundamentally change the use cases, or the method of identifying the compounds. It's just putting the identification of compounds on us and our computers instead of in the brains of an animal optimized for the job.
(Score: 2) by opinionated_science on Friday May 23 2014, @11:22PM
Indeed, the training it the hard part. But more importantly there are false readings too, and they need to be very close to 100% to be of use. Blood glucose and EtOH detectors are very highly rated because they are solid state. Read up on them, it is interesting just how technical it is.
I am not saying an animal is not useful, but as a final diagnostic it has many problems. Anecdotal support being only one of them.
Rememeber, the sense receptors are G-protein coupled receptors (GPCR) and therefore they need to have a specific ligand(s) to trigger them. Dogs have a better sense of smell than humans because their evolution has required them to continue to hunt for food. Mice, in contrast have even more receptors than dogs, as their body mass is small and poisons would kill them quickly. Also , probably for predator evasion....take your pick.
Since there needs to be a receptor to detect a molecule, the connection is not guaranteed. Some humans cannot smell asparagus in urine, say. The molecules that were isolated are S-methyl thiacrylate and S-methyl 3-(methylthio)thispropionate.
The point is the molecules are known. We can now craft an objective test for these compounds.
A mutation in the olfactory gene knocks out the detection. Without the molecule, how do you find the receptor and know your dog smells them?
Find the ones for cancers in aspiration and you are onto a winner. Having a dog make happy noises 90% of the time might help, but only if it leads to the molecular basis being found.
A final point is that cancers are unique living systems. They are genetically heterogeneous, making it difficult to assume they would give off the same molecules.
I would not have written so much but the 2012 article had gaping holes in it. That's the good thing about peer-review publications, at least there is something we can both look at, even if we don't agree...
(Score: 1) by Immerman on Saturday May 24 2014, @02:45PM
As a final diagnostic, certainly. But as a fast and cheap screening procedure to decide whether more invasive screening is called for they would be extremely useful, provided the false-negative rate was sufficiently low. Multi-stage screening is hardly a new concept - plenty of screening tests are used that have embarrassingly high false-positive rates simply because they are far cheaper and less invasive than the more reliable tests. Coming up positive may still mean you probably don't have the disease, but if it reduces the number of people who need to be subjected to the more expensive screening by 80% then it's a net win (or alternately, if it means you can cheaply screen people who would otherwise not be screened at all, thus catching many more victims before the disease becomes difficult to treat.)
Take breast cancer for example (one of the cancers reliably detectable by the new "breathalyzers" entering field trials, and thus presumably a safe bet for being dog-detectable). Mammograms are time-consuming, painful, and actually increase the risk of cancer developing by bombarding the tissue with x-rays on a semi-regular basis. If a woman could instead breathe on a dog to get a preliminary evaluation that could rule her out as "safe" then it's a net win - you only need to give mammograms to those who the dog identifies as at-risk. And even if the false negative rate is moderately high, so long as it's comparable to mammograms (which have an embarrassingly high false-negative rate) it's still a net win for the patients who are spared the discomfort and x-ray bombardment.
As for the mutation, so what? A dog with such a mutation couldn't be trained for screening, so it wouldn't be used. Knowing exactly which molecule it's detecting won't make any difference.
(Score: 2) by opinionated_science on Sunday May 25 2014, @02:14AM
well that is one of the major research problems. What effects to the mutations make? Go look at the 1000 genome project, for example, plenty of non-synonymous mutations (e.g. NTSR1 the neutrotensin 1 receptor has 83 non-synonymous mutations from population of 1000). Some mutation will cause constitutive activy (the receptors gets turned "on" permanently), or will cause a receptor to become less receptive with different salt concentrations.
My point is with the exact molecule, we can produce both positive and negative control tests.
So let's leave it as "it's a nice idea but needs more work". The particular study was deeply flawed. And we always want to know the active molecule...
(Score: 1) by Immerman on Sunday May 25 2014, @04:12AM
I won't argue that with the exact molecule and a cheap mass spectrometer we can do far better than a dogs nose - that's obvious. My point is that we've known for almost a hundred years how to do so *without* knowing any specific molecules, just as we've used dogs for search and rescue without knowing exactly what (combination of) molecules allow them to track a specific individual.
And again I say - mutations don't matter. Obviously a dog carrying a particular "debilitating" mutation may be unable to identify cancer - but we don't care. Such a dog won't be able to pass the identification training course and thus won't be an issue - it doesn't matter if it's because they have a nose-compromising mutation or are just particularly dumb or stubborn - they get weeded out by the training regime. It's not like a mutation is going to suddenly disable an already-certified dog, and even if it somehow did regular re-certification would solve that easily enough (and would be a good idea for several other reasons).