GMO virus vs B-Cell Acute Lymphoblastic Leukemia: Why is this not a standard therapy yet??


Efficacy and Toxicity Management of 19-28z CAR T Cell Therapy in B Cell Acute Lymphoblastic Leukemia

I wrote about this last year:

‘Dismal prognosis’ with leukemia? Nothing a GMO virus cant fix.

In the previous study, scientists took cytotoxic T-cells from five B-ALL patients, and infected them with a genetically modified virus. This GMO virus had a genome that essentially contained a 'cheat sheet' for teaching those CTLs how to kill B-ALL cancer cells. Four out of the five B-ALL patients recovered enough that they could get a bone marrow transplant.

In this new paper, they treated another 11 patients.

As per the current standard of care for adults with relapsed or refractory B-ALL, the initial primary aim of therapy is to reinduce a CR (8–10). This, in turn, renders the patient eligible for an allo-SCT, which is, at present, the only therapeutic modality with curative potential.

This means, they need to get the patient healthy enough for a bone marrow transplant. *That* is what 'cures' the cancer, not chemo, not radiation, not GMO viruses.

With standard 'salvage chemotherapy', 30% respond well, and 5% respond well enough to get a bone marrow transplant.

With 'salvage chemotherapy' and this GMO virus therapy, 88% responded well, and 44% were able to get a bone marrow transplant.

To date, 7 of the 16 (44%) treated patients have successfully undergone an allo-SCT post-CAR T cell therapy with no relapses.

So...... This makes me wonder. What would happen if you just gave people the GMO virus? No more chemo?

Chemo suuuuuuuuuuuuuuucks, while the side-effects associated with this GMO virus are limited and controllable (your immune system gets PISSED OFF when the T-cells are 'trained' by the virus, but steroids help control this). Scientists described in this paper that they found a number of hallmarks in patients that had a 'bad' inflammation response. This means they could predict whether a patient would have a 'bad' response, and perhaps one day figure out how to abrogate these troublesome side-effects (steroids calm down the immune system, but also calm down the genetically modified T-cells that should be killing the cancer).

But I mean, that is the side-effect of taking CTLs out of cancer patients, treating them with a GMO virus, and putting them back into patients. The ability to tolerate a bone marrow transplant goes up from 5% to 44%, but you *might* get a controllable fever.


What happens if you just skip the 'salvage chemotherapy' step?

Considering the apparent benefits of this GMO virus, its relatively humane side effects when compared to chemo, when does *it* become the standard therapy, and chemo gets put on the shelf for 'just in case the GMO virus doesnt work'?

It will be a while longer.

These studies were just Phase I trials. Phase I arent even asking whether the treatment works. Theyre just asking if the treatment hurts/kills someone straight up. I cant emphasize this enough-- sometimes unexpectedly deadly things happen. So, there are 5, 11 people in a trial. If the treatment happens to help some people, thats great, but thats not 'the point'.

Its the next phase, Phase II, that is looking to enroll enough people to really assess whether the treatment is beneficial or not. And then the patient numbers get scaled up again for Phase III. And then the treatment gets put up for approval by the FDA.

There is still a long road to basic approval, certainly we are nowhere near replacing old therapies yet.



More like this

So the answer to your rather provocatively phrased title question is: We don't know how many people this will hurt, yet!

The statistics you quote are 7+/-4 of 16, or 44+/-22 % success rate. Certainly a 66% success rate argues for changing the standard of care. But what about 22%? What if this low statistics result turns out to be an upward fluctuation?

And I've only used 1-sigma (sqrt(N)) above! What if I used the typical life-sciences "P = 0.05" value, and quoted two-sigma ranges? Now that 44% goes down to potentially zero (44 +/- 44%).

As you point out in your conclusion, this is not the standard of therapy yet because we DON'T KNOW IF IT WORKS. Arguing otherwise, based on what is little better than anecdote (five patients), is what the quack-shills do in pushing their useless "alternative" therapies on uninformed and desperate patients. The explainers of science-based medicine need to be better than their opposition, or we run the great risk of being sucked into the "false balance" fallacy.

By Michael Kelsey (not verified) on 24 Feb 2014 #permalink

The "1 sigma = sqrt(N)" is a standard method to approximate the binomial distribution by using a normal distribution. This works well (and simplifies number crunching) when N is "large". It isn't a terribly good method when N is "small". But if N is small, it isn't hard to fully enumerate the different possibilities in the binomial distribution, so there is no need to approximate anyway.

A sample of 16 is a pretty small number. The normal approximation kinda sucks here, but it is relatively easy to run the numbers for the binomial. Also, the idea that a probability of zero might be in the 2-sigma interval of something that yielded 7/16 in trials should have rung some alarm bells.

Using the inverse binomial CDF for a range of possible population probabilities bracketing a [2.5% 97.5%] interval (i.e. 95% confidence) that includes 7/16, I get bounds of between 20% and 71% from the limited trials already undertaken - zero is most certainly not there, and the lower bound remains bigger than the 5% from chemo.

It is a difficult situation and the ethics of short circuiting trials is a tricky one. But you have to remember: this isn't a medication for treating something curable where undetected, low probability devastating impacts on the patient would mean the risk is unacceptably high. This is something for patients that have no other option available to them and a really, really horrible prognosis. But the trials through which the technology must pass through are presumably structured very similarly in both cases (someone feel free to correct me here; I am not speaking from a position of knowledge). I think Abbie asks a good question on whether this is the right thing to do in this case.