Hacker News Reader

Comment by static_noise | Hacker News Reader

Two things come to mind immediately:

1. Cancer cells are human cells and behave very similar to human cells compared to foreign bacteria or viruses which have a vastly different metabolism.

2. Cancer is not a single disease but a gazillion different mutations which may have vastly different characteristics.

Point 1 most of the time prevents cures such as "kill the human cells" from working effectively without killing the patient, too. Successufull cancer cells look so "human" that even the immune system doesn't see the difference. Point 2 means that the "cure for cancer" may be found for some kinds of cancer but there are thousands more. If we cure all cancers known today we will find new ones when the patients are just a few years older. Remember that the death rate increases exponentially with age and so will diseases like cancer.

View on HN

3. Cancer cells can be hard to get to with drug therapy. Even if we have a drug that kills the cell in question we can't always get therapeutic concentrations into the core of solid tumors, and we can't get many drugs past the blood brain barrier.

4. Cancer cells can be highly unstable, genetically, depending on what part of the DNA got mangled. Unstable cells that mutate rapidly follow the same Darwinian logic you see in microbes. A single tumor can contain thousands of different mutations of the original cell type, and if your therapy kills 99% of them the 1% that come roaring back will be highly resistant to whatever strategy you used the first time around.

Recent articles about tumor growth heterogeneity :

http://www.nature.com/nature/journal/v525/n7568/full/nature1...

Videos below are pretty telling.

Can't we just kill half the cells periodically until there is no more cancerous cells ?

We can, but that kills the patient.

One of the approaches used in chemotherapy is drugs which disrupt cell division or associated processes. Which means the cells dividing most frequently (i.e., cancerous cells) are the ones hardest hit by the drug.

Though this does affect other cell types which divide frequently, which is why chemotherapy patients lose their hair (hair follicles need rapid cell division) have digestive problems (digestive cells divide rapidly) and can suffer immune-related issues (bone marrow divides rapidly).

This is not far off from what much traditional chemotherapy actually does. The approach is to literally poison yourself, and hope that as a whole you can take more of it than the cancerous cells can.

Do you study biochemistry or medicine? Because it would seem that your first point is completely true but technically an oversimplification when it comes to treatment of common cancers.

https://en.wikipedia.org/wiki/Asparaginase#As_a_drug

https://en.wikipedia.org/wiki/Bcr-Abl_tyrosine-kinase_inhibi...

Those are classic success stories mainly because they were the low-hanging fruit for targeted cancer treatment. For blood-cell cancers like leukemia and lymphoma, treatment has come a long way due to advances like the ones you mentioned.

For "solid" tumor cancers like breast, pancreatic, prostate, and skin cancer, we're still struggling to make a dent in the death rate.

Lets distinguish between diagnosis and treatment. Many cancers can be identified by the over-expression of genes or other genetic material. This knowledge is immediately useful to clinical staff. Although its hard to use this information to target specific cells, it hints at mechanisms that are of interest to academic research.

> For "solid" tumor cancers like breast, pancreatic, prostate, and skin cancer, we're still struggling to make a dent in the death rate.

Actually, we are not "struggling" on breast cancer. The death rate comes down a reasonable percentage each year.

It's just that a small percentage compounds slower and slower once you take out the low hanging fruit.

As you point out, the problem is that cancer isn't just one disease with one mechanism. Even "breast" cancer isn't just one disease with one mechanism.

So, it's not terribly profitable to go after say "cancer X with mutations P, D, and Q". This really needs government funding.

We'd have been far better off with a governmental "War on Cancer" than on any of the other "War on X" that they've done.

>So, it's not terribly profitable to go after say "cancer X with mutations P, D, and Q". This really needs government funding.

This is one of the key problems. The effective size of most true cancers types is so small that pharmaceutical companies can’t make a profit developing a drug that targets these cancers. Imagine you identify a new drug that 100% cures people with a cancer that has three specific mutations. Now lets assume that out of the millions of people who get cancer each year only 500 have this particular combinations of mutations. This drug will cost the same as any other drug to develop (lets assume $1 billion), but you can only sell it to 500 people per year. Even if you set the price at $100,000 per person you don’t come close to earning enough money to pay back the development costs. Add in the fact that very few drugs make it through the whole development process you can see why we have a problem.

This is not as big a problem as:

1. Enrolling enough people in a trial to prove you can cure with a combination therapy

2. Evolved resistance with an entirely new set of mutations

Well these two problems are technical problems that can only occur if the key economic problem can be overcome. You can only have issues with enrolling people in a trial if the trial is at least attempted.

The current thinking is that cancers are not evolving for resistance during treatment, but that there are a sub-population of cells present in the tumour mass that are being selected for by treatment (i.e. we kill 99% of the tumour that is sensitive and the 1% that is resistant grows back).

We know how to solve this problem from HIV. HIV evolves resistance to drugs faster than any cancer yet we can treat patients for years by using HAART (i.e. give lots of drugs at once that target different key components). We need to do this with cancer, but this will require developing new drugs that are less lethal since you can't just give a patient a dozen standard anti-cancer drugs and expect them to live.

>The current thinking is that cancers are not evolving for resistance during treatment, but that there are a sub-population of cells present in the tumour mass that are being selected for by treatment (i.e. we kill 99% of the tumour that is sensitive and the 1% that is resistant grows back).

No, the cancer evolves resistance during treatment. Evolution is change in allele frequency in a population over time, not the emergence of new mutation. It does not matter whether the mutation was a rare, preexisting mutation or it occurred after treatment; both likely happen. The "evolution" part is the outgrowth as a result of a fitness differential produced by the new environment (therapy).

>We know how to solve this problem from HIV. HIV evolves resistance to drugs faster than any cancer yet we can treat patients for years by using HAART (i.e. give lots of drugs at once that target different key components). We need to do this with cancer, but this will require developing new drugs that are less lethal since you can't just give a patient a dozen standard anti-cancer drugs and expect them to live.

We will never be able to do this with cancer. HIV has only nine proteins in its genome. This means any resistance mutation it develops must be on-target and will tend to operate through specific bottleneck points. This is not true with cancer - there is on-target resistance, of course, but this is easily overcome through better drug design. We can develop drugs that get around EGFR T790M, but in a cancer there are a dozen different new mutations in other genes that can emerge to produce resistance. Drugging all of these targets in advance is impossible - we can barely drug one gene with patients tolerating it. We'll lose this game of whack-a-mole.

This is a losing game in general. The tumor has too much space to adapt - it's not limited to a tiny genomic space like HIV, it has the whole human genome to work with.

This is why I think successful strategies are going to result from not inhibiting particular genes but treating the tumor as a tissue and modulating the host/tumor interactions. Immunotherapy acts on this axis, and it's already looking more promising in this regard (responders don't tend to become resistant).

I think we are in agreement, but you did say "2. Evolved resistance with an entirely new set of mutations”. I was pointing out that evolution is not coming about from the selection of new mutations, but the selection of pre-existing mutations present with the tumour before treatment.

>Drugging all of these targets in advance is impossible - we can barely drug one gene with patients tolerating it. We'll lose this game of whack-a-mole.

It is not impossible, just difficult. It will certainly be impossible if we keep trying to develop new treatments the way we have for the last 50 years. We need to change the entire way we screen and test new cancer treatments if we are going to be able to use the HAART approach in human cancer.

I agree with you that immunotherapy is looking the most promising approach right now, but done right chemotherapy has a lot of potential too. I would not rule one or the other out.

I'm skeptical it's as bad as that. We classify cancers by the type of cell that went bad, i.e. lung cancer if it occurs in the lung, liver cancer if it occurs in the liver, brain cancer if it occurs in the brain, etc. But you can have, say, lung and liver cancers that result from the same mutation, and there's a chance, at least, they'll respond to the same therapy.

And sometimes there are other diseases caused by problems in whatever pathway you're messing with. Tamoxifen, for example, can be used to treat two relatively rare non-cancer diseases.

Classify cancers by the tissue origin is only of partial help in treating. A cancer may arise in the liver, but be totally refractory to one treatment, but respond to another.

One way to workout how many different cancers there are when classified by treatment response is to look at the true cure rate to when a cancer is treated using a treatment. Since the treatments we have are biased in targeting the most common sub-types this can only give us a lower bound, but even this number is huge since the cancer cure rate is not that high for the most common cancers (I am not talking about 5 year survival, but real cure). My rough guess is there are 10,000s of different cancers when classified by treatment response.

>Classify cancers by the tissue origin is only of partial help in treating. A cancer may arise in the liver, but be totally refractory to one treatment, but respond to another.

Well, sure. That's my point, though. Cancers that arise indifferent parts of the body may be treatable with the same drug.

>One way to workout how many different cancers there are when classified by treatment response is to look at the true cure rate to when a cancer is treated using a treatment.

That's the current situation as it stands today. It's not necessarily what we'll see in five or ten years.

The problem is that we don’t have the range of treatment types to target all the the different mutation types. Most of the drugs we have target the same basic pathways like DNA damage repair - we have dozens of these drugs, but very few targeting specific mutations. We need a lot more diversity of action in our drugs if we have a chance to cure cancer.

Having said this the most hopeful area is the immunotherapies. Scaled out to their full potential we could have a very powerful set of drugs to go after the cancer diversity problem.

If it costs $1 billion for private companies to develop a drug, why wouldn't it cost $1 billion for government to develop it? Spending $1 billion to save 500 people will bankrupt the country one way or another, if you keep going after cures for 500 people groups, whether government funds it or not.

A more effective approach would be to find ways to reduce the cost of new drug development.

It certainly would, but that is another discussion :)

Reply on news.ycombinator.com