Crispr Wins Key Approval to Fight Cancer in Human Trials | Hacker News Reader

Crispr Wins Key Approval to Fight Cancer in Human Trials

bloomberg.com

349 pointsadventured85 comments

If you think stuff like this is cool, and you'd like to be a part of the software engineering effort at Parker Institute for Cancer Immunotherapy, msg me. ccary@parkerici.org

We are building out our platform for bioinformatics. Some engineering topics:

* Large scale data processing (Cloud & on-prem HPC based)

* Bioinformatics / ML algorithms

* Data storage & retrieval (File systems & databases)

I think this is cool! Are there any particular software engineering skills or experiences that would be really valuable or relevant to this work?

Yes, please elaborate on what the software engineering effort involves! Is this a bioinformatics thing? What kind of work does it involve? Does it require a strong theoretical background? etc.

(Edit: Thanks! Just noticed you added it.)

maybe WRT to https://www.google.com/search?q=genome+sequencing+algorithms which would be needed to find unique patterns among billions of base pairs for CRISPR to target. it's microscopic big data :D

[1] http://benchling.engineering/dna-regex-search/?hn

[2] http://benchling.engineering/optimizing-crispr/

Seems like this is something you might use the Church lab's spinoff Curoverse for?

I've got a background in machine learning (NLP, and a little bit of vision) and electrical engineering, but I know very little about biology. Essentially, all I know is a working knowledge based on what I remember from high school. However, I'm incredibly fascinated by some of these recent developments in making biology more computational! I've started working through some Coursera courses on bioinformatics systems biology, and have been totally engrossed so far!

I noticed that the Parker Institute is a non-profit? I was curious what kind of ramp-up path you offer people like me, with a background in ML, and are excited to learn some more biology if they join the team? Since you're a non-profit, I was wondering if there'd be any interest in helping to curate a bunch of links/resources/etc. so people like me can learn and ramp-up in our spare time? I guess I'm wondering if it's difficult to hire for this role. So maybe other non-profits, e.g. J. Craig Venter Institute may also be interested in helping to curate these resources to help increase supply of possible hires?

Hey, could you msg me to talk about this. Thanks!

Cool, I've sent you a follow up email.

If this is a non-profit organization, does this mean that the resulting software will become open-source?

Having lost a sister and a son to cancer I hope for this to work. The issues are tremendous for this technique and I don't have high hopes for its use on cancer. I am still having high hopes on immune system techniques that helps the body to mark cancer cells and attack the cells for most cancers.

CRISPR big hope is that it can change the cell to stop reproducing or stop invading other cells. Simply put cancer cells are cells that grow that don't stop AND can spread to other cells. If you get one of the two you win.

HIGHLY RECCOMENDED : list to Radio Lab Famous Tumors http://www.radiolab.org/story/91713-famous-tumors/

Isn't the idea with CRISPR that you patch the patient's t-cells in such a way that they are able to recognize cancer cells as diseased cells, allowing the immune system to clear them away?

The holy grail would be fixing commonly occurring mutations that cause the cancer, but to my knowledge the edit/correction rate is too low for this to work currently, plus the possibility of causing more cancer by bad edits.

Thus, the T-cell approaches are more promising currently as it allows for direct modification of cells to fight cancer with known anti-cancer antigens, using a modified immune system.

Could this theoretically be used to fight all pathogen-based illnesses as well?

Theoretically, it could be used to get a person to grow a new set of arms. But technology that doesn't exist yet is all speculation.

The SENS approach to cancer is pretty much to turn off its ability to grow in unlimited fashion. If you could compensate for the side-effects, it would be possible to delete telomerase and ALT genes from humans, which would produce cancer free humans - if cancers can't lengthen their telomeres, they die.

Unfortunately, it would also produce humans would be dependent on some form of periodic telomerase therapy targeted to stem cells, or stem cell replacement, or similar. This might be a net gain in a suitably high-tech environment.

A better approach would be something other than permanent removal of telomerase and ALT genes, but perhaps suppressing them globally for a while, or suppressing them in cancerous tissue only. RNAi or other methodologies are pretty reliable for this sort of thing at the present time in the general sense.

As a parent and a brother, I'm so sorry you had to go trough that.

Sorry to hear that. Hang in there. I lost my mom to cancer about a year ago.

> Hang in there

I have learned through this that A) I depend on God B) I love life and cherish more because of the journey. SO they have made enjoy life more because they were in my life.

> I lost my mom to cancer

Sorry to hear that. Cancer SUCKS

I'm glad to hear you found some good from such a terrible situation.

So sorry, no father should see his son die.

I hope that by the time I'm succumbing to the effects of aging, we'll have techniques with tools like CRISPR to make it easier.

I'll see you at my 1000th birthday party! high fives

Ow! Not so hard!

It seems likely, assuming a couple of decades before you're "succumbing".

Who knows... Ain't that the mystery?

And are you doing something about it (donating/investing/working/etc) ?

Chances are anyone in a developed-world democracy is investing/donating via their tax dollars.

Not very effectively or with the focus you might want, though.

I tend to believe the NIH is better at administrating grants than I am.

I'd absolutely agree they're better than I am personally, since I'm neither a medical researcher nor an expert at evaluating medical research. But I don't believe they're better than an organization focused on solving a specific problem.

As a random example of inefficiency: there's a lot more money going into research on specific degenerative diseases than into the causes and solutions for degeneration. Solve the root cause, not the symptoms.

While the idea of pursuing a 'root cause' is attractive, in many cases, including the one you point out, there is not a clear cause to be attacked.

Neurodegeneration is a pattern of response to multiple disease processes - we can only understand the response by studying the cases in which it arises.

For the tech-minded crowd, this is like wanting to create responsive designs without actually making websites to demonstrate them.

They're actually pretty terrible. It isn't a high bar to do better than the NIH when it comes to supporting new technologies. Almost all important early fundamental research, the high-risk part leading up to new knowledge and potential new technologies, is supported by philanthropy.

The NIH really only funds follow-on work for things that are already known. In order to win a grant you have to have already proven your thesis. It is somewhat ridiculous.

So if you want to delegate your vote, delegate it to a non-profit. Or invest in a startup in the field you care about. Anything other than giving to the NIH.

Ironically, you skipped "Give it to a lab whose research you want to support."

True. In fairness, that's much harder to do at lower levels of money than either of the options I suggested, despite efforts like labcures.com. You need to know a lot more about the field than you'd need to know in order to give to a non-profit in the space.

True. Unfortunately, there's also a lot of spaces in science where there isn't a clear and direct non-profit necessarily working in the space and funding research.

But the proportion of your tax dollar they're getting makes that a bit of a moot point if you were thinking about donating directly and making a tax deduction.

I'm sad to see this downvoted. It's reasonable to try to turn expressions of support on a forum into action.

For anyone who wants to support research into ending aging, age-related degeneration, and related diseases, I'd recommend supporting SENS (http://www.sens.org/).

Off-topic: is your account named after the diatonic scale? I love the dorian scale.

First time reading about it. I just iterated until I found a unique and easily rememberable username a long time ago (d-day 1943).

Funny coincidence. The most common dorian scale is called "D dorian".

You don't have to be old to get diseases. The way I see it a few aches and pains are a small price to pay for not being dead.

As someone about to start PD-1 inhibitors after failing Ipilimumab (now in lungs (1))... well it'd be just plain stupid to die now with all these treatments on the horizon.

---

(1) This just happened this month. The Doctor called me with the results literally the week after I handed in my notice at a secure job to join an early-stage startup.

Life really is short! (http://www.paulgraham.com/vb.html)

No I don't regret my decision :)

(Also thank Jeebus I live in Canada where my current lack of benefits does not preclude me from getting the best treatment.)

If you want to better understand the CRISPR patent dispute, this article lays out the arguments very clearly. https://www.statnews.com/2016/03/18/crispr-patent-dispute/

Ha, are you sure you didn't mean http://www.cell.com/cell/fulltext/S0092-8674(15)01705-5 ;D

(For those of you who aren't in on the joke, "The Heroes of CRISPR" was written by the director of the Broad Institute, which incidentally has Feng Zheng on its payroll. For that reason Eric Lander has been roundly criticized for attempting to weigh in on the history early. My friends who work in and around Kendall Square say that some of the criticism is overblown, the article points out a lot of people who didn't get much recognition prior, but still... it's a bit skeevy so you might want to take the latter half of the article with a grain of salt as it's not necessarily factually wrong, but the emphasis might be distorted and there is a risk of omissions.)

It might be a good legal overview (I wouldn't know), but the scientific merit is overwhelmingly in Doudna's favor. The actual novelty and value of the method is the ability to edit nucleotides, not whether it can be done in prokaryotes vs. eukaryotes.

In the case where this was a known technique that had significant difficulties in translating to eukaryotic work, then the Broad Institute's work would be significant. That was not, however, the case. Very little needed to change, and the technique has been widely applicable.

Just a quick look though Doudna's publications shows that this is an issue with a great deal of background work behind it, years of effort that fortuitously resulted in an incredible invention. Then, someone comes along, copies your work in some different bugs, and claims ownership. It's really no more complicated than that.

Why doesn't the article mention Emmanuelle Charpentier?

No doubt, CRISPR is the next big thing that will make sense for everyone.

I work for Synthego and we make synthetic RNA for CRISPR/Cas9 Genome Engineering. Our process is fully automated with robots and an architected LAB synthesizer.

We are actively hiring:

* Full Stack Engineers with strong DB and Python skillS * Robotics/Mechanical Engineers with automation experience AND who think outside the box! *Extraordinary Human Beings are welcome too

If you are interested --> email me: victoria.danahy@synthego.com :-)

>rewrite the blueprints of life itself

Something unsettling about how that is worded, but this awesome. There hasn't really been any major breakthrough like this in a very long time, as the era of antibiotics ends, new treatments like this might open the door for a new wave of medical treatments.

I'm calling it here, this trial will fail hard.

The basic idea is this: the cancer has a number of mutations that produce novel antigens that don't exist in normal cells. If we can train T-cells to recognize these, they can attack the tumor with great specificity and kill those cells.

The problem is, all the tumor has to do to evade this targeting is shed the problematic mutation, which is easily done - tumors are great at acquiring novel mutations to become resistance to therapy, losing a mutation is even easier. All you have to do is ditch one arm of a chromosome, revert the mutation, or otherwise eliminate the novel mutation, and your tumor cell is now resistant to this therapy.

Add to this the fact that we are basically shit at predicting which mutant proteins will produce neoantigens and you have a recipe for total failure, even in concept.

How long will this process take? Is it enough to create a new therapy to target the mutated cancer?

People develop resistance to TKIs (tyrosine kinase inhibitors, the previous mode of therapy) within months; you could keep chasing the dragon, but the tumor is just going to keep doing the same thing, shedding whatever mutation you target. In this case you're not even targeting mutations that the tumor relies on (as you are with a TKI), so there's absolutely no fitness cost to the tumor to evade.

But what if the antigens are vital for the tumor ? What mutation are you referring to? Presumably the immune system will kill the cancerous cells ( shedding the mutation as well).

Could you simultaneously target multiple mutations to reduce the risk of this? Similarly to using a cocktail of antibiotics to prevent the development of resistance.

i.e. the probability of shedding a single mutation is 1/x, but the probability of shedding two in the same organism is (1/x)^2, so targeting more mutations make it exponentially harder for the organism to adapt.

Yes, this reasoning is correct. Same principle applies to combination therapies for HIV, malaria, etc., where you're at risk of developing a resistant strain within the patient. Not sure why it isn't more common among cancer therapies. Maybe not enough unique targets?

Not enough unique drugs, difficulty determining mutations/susceptibility of tumour to drug, high side effect profile of drugs used in isolation and higher still in combination

Can you turn off a mutation's ability to adapt and mutate?

It would be called killing the tumour. The only way to destroy mutation is to destroy replication. We don't know how to increase the fidelity of DNA replication, and cancer cells have their error correct mechanisms all destroyed

hey - just want clarification. I thought TKIs like gleevec give an asymptomatic period of years - do you mean by "develop resistance" the initial seed population of cells that are resistant emerge from a (presumably clonal or near-clonal) pool within months?

Second: Do you know if clinically, agents like gleevec are primarily used as standalone or in conjunction with more aggressive (and, unfortunately, side-effect laden) therapies? I was always a bit aware of the pernicious "maintenance drug" status of gleevec, but presumably if you combine it with an orthogonal therapy you could get higher coverage and better statistical odds of sending the patient into remission.

Imatinib and other TKIs provide a range of effectiveness, from weeks/months to years. Nothing is forever (although the CD1 Ligands like ipilimumab etc show promise).

Mostly, the TKIs spread the Kaplan-Meyer curve out/lengthen it, but because it is still relying on the immune system to kill the cells (mostly) essentially TKIs are providing cellular senescence rather than cellular death. The bcr/abl mutation that imatinib targets is a cellular replication pathway, not s cell death pathway

ah... so standard chemo/rad isn't used because the immune response to clear out the senecent bastards will be compromised.

Agreed that for most tumor types targeting single antigens is a fool's errand, but the claim that "here is absolutely no fitness cost" for mutational evasion seem like is an overstatement. This is clearly false when using bacterial resistance to antibiotics as a more tractable proxy.

I assume that you haven't heard that immunotherapy is the new big thing in cancer research?

It's how they cured former President Carter:

http://www.forbes.com/sites/arleneweintraub/2015/08/21/jimmy...

I do cancer research at UCSF on an immunotherapy project...

I'm not one to usually complain about downvotes, but this is ridiculous. The HN hivemind at its most absurd.

This person is literally working in the research sector involved, giving their opinion (it's not a nice opinion, but it is presumably informed), and getting downvoted because the HN crowd doesn't like it.

EDIT: At the time I wrote this comment the parent comment and the root comment were both far into the negatives with downvotes.

Agreed, people seem to be downvoting because they hope he/she's wrong rather than to post a reply explaining their reasoning. I see nothing wrong with the root comment and just upvoted it to give it more attention.

Give it an hour or so. the hivemind is usually strong in the beginning.

Well, we all know that now but not for the first comment.

Should be

Down votes = troll or argumentative or factually wrong

Actually

Down votes = I disagree

I have quit HN many times over that but there isn't anything that can replace HN right now.

I agree with you that the down voting is silly, as s/he makes some good points, but suspect the down voting was in reaction to the tone of the comment. If one was holding a lab meeting and a comment arose in that tone, one could image other lab member getting defensive. tl;dr The comment would have been more effective without the passion of fail'hard' and other language.

I'd certainly be interested in hearing what's effective and what is not. How does your project work, for example? Are you discounting most immunotherapy drugs? Is there anything we should pay particular attention to when reading the mainstream media?

Actually, immunotherapy is the new old big thing in cancer research. As I recall it was first hot in the late 70s/early 80s after the invention of hybridomas, but failed to produce dramatic early successes.

I share the parent's skepticism. Fighting cancer is fighting basic biology population dynamics: the population that survives the assault is by definition harder to kill. It's made even harder by the very nature of cancer cells that freed from following the rules: they can relocate at will while simultaneously ignoring the body's zoning board (metastasis), steal nutrients (angiogenesis), dump waste in their backyard (poor lymphatic recruitment) and create an area that is inhospitable to other cells while simultaneously making for difficult drug delivery, ignore court orders to stop what they're doing (evade apoptosis), and replicate faster than other cells in the body (both by producing their own growth signals and ignoring anti-growth signals, including senescence). Also, the cancer cells have a head start. The deck is stacked against "curing" cancer by chemotherapeutics or immunotherapy.

"Add to this the fact that we are basically shit at predicting which mutant proteins will produce neoantigens and you have a recipe for total failure, even in concept."

Do you think if we had more data surrounding those mutant proteins we could start predicting better?

What about the CAR-T therapy that has work pretty well against blood cancers (ALL and CLL if I remember correctly)? Seems there were a handful of remissions in patients who had failed everything else.

What makes this approach so destine to fail? It's an honest question as I'm not as familiar with the details.

CAR-T is a step in the right direction for the cancer research establishment, not because it solves the problem noted by the parent comment, but because it offers the prospect of greatly reducing the cost of targeting many different types of cancer.

It is very much the luck of the draw going cancer by cancer as to whether you can pick a target that is more or less likely to be evolved around by the cancer. Leukemias were a lucky hit, but that may or may not hold up for other targets and other cancers as they are put forward.

Almost all trials fail, but it's interesting that CRISPR is getting the nod. It has so many uses that if it doesn't work in this specific cancer study it doesn't really matter.

Which is a good reason to back things like telomere lengthening interdiction. That would work on all cancers, and cancers can't evolve around that dramatic a hurdle, so far as we know. It is a very fundamental set of machinery, and all cancers have to lengthen telomeres in order to stay ahead of the limits to replication.

Some groups are working on blocking the action of telomerase, but to make a viable treatment ALT (alternative lengthening of telomeres) will have to be blocked too, since researchers have already seen telomerase-abusing cancers turn into ALT-abusing cancers.

Depends on what you mean by "fail hard," I suspect. Sure, cancer cells can gain or lose mutations, but if we target a mutation important for cancer development only a fraction of the target cancer will survive the treatment. We might achieve nice-to-have results like reduction in tumor size and increased patient life expectancy, or a "cure" in lucky patients with less agile cancers, which would put this radical new intervention in line with most standard cancer treatments. As a total cure for cancer it will "fail," but then, so have all the breakthrough cancer treatments that have collectively shattered the mortality rates of cancer patients.

(You mention you're involved in cancer research, so you presumably know all this already- I'm wondering if there's a more nuanced reason for this pessimism.)

The best theory I've heard so far as to how we'll "cure cancer" altogether is that we'll simply develop a big enough armory of anticancer treatments that no individual cancer case could evolve fast enough to beat them all. Unlike bacteria or viruses, cancers evolve on their own and when they're done their evolutionary innovations die with them. Our technology lives on, though, so we can hope to eventually just totally outclass their capabilities. If CRISPR gives us another source of these partial treatments, we're just that much closer to that level.

This method is not targeting mutations important for cancer, this is targeting "passenger" somatic mutations. You can't use this mechanism to target important mutations because only a minority of patients are going to be able to present an oncogenic mutation (e.g. BRAF V600E) as an antigen (because of differing HLA types). In those rare cases perhaps this therapy might succeed, although the tumor probably still has ways of evading (downregulating the offending HLA for example).

As for the "kitchen sink" approach, so far none of the methods we've developed are free from side effects. Cancer patients are often on the edge of death; each successive therapy will probably do more damage to them. As it is most people fail on current immunotherapies because of autoimmune reactions. So relentless application of therapies is not necessarily going to work, though of course having more weapons in our arsenal is great.

I was under the impression that the major reason for the failure of current immunotherapy treatments is due to non-response, not excessive autoimmunity.

Sorry, you're absolutely right. I should say 'many', not 'most', my comment was hasty. It varies with the therapy and only a minority of patients fail because of adverse immune events, and the drug matters - ipilimumab (anti-CLTA4) had 10-15% severe immune-related adverse events, whereas anti-PD1 is much better tolerated (probably because the immune checkpoint is later). Combination therapies (ipi + nivo) have much higher adverse event rates, unsurprisingly.

For the downvoters: the fact that the parent does not think it will work says nothing about whether or not he would hope that it would work, it's just that he's a bit more informed than most here.

I agree that this trial is almost certain to fail, but as I am sure you know the resistance mutatants are normally preexisting in the patient before treatment and the treatment just selects for the resistant mutants.

I think one of our bigger problems with fighting cancer is we only trial new treatments in people on deaths door who have failed everything else. Treatments that would help someone newly diagnosed (the treatments we really want) are just not possible to bring to market.

Well, if it only fails to cause autoimmune diseases as a collateral effect, it'll be already a huge win for CRISPR techniques in the future.

My understanding is that this trial is just

    1. adoptive CAR T cell transfer 
        (which has been shown to work 
         with varying degrees of success)
    2. PD-1 is CRISPR'd out (to combat exhaustion) 
    3. The TCR is CRISPR'd out (to increase specificity)

CAR Ts have certainly had some success; my concern with this trial would be more autoimmune problems, not that it would fail hard.

Is there something else about this trial that would make it unsuccessful?

Well, it's worth a shot.

http://i.imgur.com/BbgL7x3.gif?noredirect

Reply on news.ycombinator.com