Update, March 10: Bluebird has reported data that makes it much less likely that this even was due to their lentivirus vector. Good news, and I hope it holds up.
There’s news today that Bluebird has suspended its gene therapy work on sickle cell disease because of two cases of cancer in its treatment population. Another had been reported in 2018, so that takes us to two cases of myelodysplastic syndrome and one case of myeloid leukemia (which can be a sequel of MDS in some cases). This isn’t good. You’ll note that all of these are diseases of the bone marrow, and the marrow is where a good deal of the action in this sort of gene therapy takes place.
There are several companies working in this space, and it’s no coincidence. Sickle cell anemia is the absolute prototype of a genetically linked disorder, famously first identified in 1949 by Linus Pauling and co-workers. That paper termed it “a molecular disease”, and Pauling certainly deserves the credit he gets as a founder of molecular biology. Both sickle cell and the conceptually related beta-thalassemias are defects in the production of hemoglobin, and it has been obvious for decades that if you could somehow yank the defective gene out of the patients and replace it with a normal sequence that they simply wouldn’t have these conditions any more.
There are by now plenty of other genetic disorders that fall into the same category, but these blood-cell based ones have a unique feature that has put them into the forefront of actual attempts at gene therapy. In these cases, all the relevant cells come from the same tissue, the bone marrow. And we actually have ways to kill that off and to swap in new tissue of our choosing: a bone marrow transplant. It is a tough procedure to go through, for sure, but not as tough as living a life of acute sickle cell attacks (or being killed off early by rampaging leukemia, to pick another application).
Contrast that with so many other gene-linked disorders – take Huntington’s, for example. We know the gene for that one, and the protein it codes for, and it is equally obvious that if you could magically yank out that gene from a patient and insert the normal one for the Htt protein that they would no longer have the disease. But there is no analogous procedure for killing off the basal ganglia of the brain and replacing it with new neuronal tissue. Not quite. No, bone marrow based disorders are a unique opportunity, and that’s why so much effort has gone into this area.
It’s a similar situation to the way that therapeutic RNAs have been aimed at liver disorders. In that case, you’re not wiping out the old cell population but rather trying to overwhelm it in situ, and the liver is chosen because we don’t really know how to make i.v. dosed RNA species accumulate anywhere else. So we make do with what we have and turn the Liver Problem into the Liver Advantage. If we ever get to the point of treating Huntington’s at a genetic level, it’s surely going to be via a similar rework-things-in-situ method as well, but figuring out to do that in only the desired regions of the brain without causing trouble elsewhere is quite a challenge – you’ve lost the Liver Advantage.
Now, Bluebird. They have been using a lentivirus vector to rewrite the bone marrow transplant tissue, and there’s a solid reason for that. Lentiviruses (of which HIV is the most famous/infamous example) insert their genetic payloads into the host cell’s DNA. It’s their key step, and they can do it even on non-dividing cells. Now, when a person hears “viral vector” these days, the thought is immediately of vaccines, and that takes us to the worry that the vaccines aimed at the COVID-19 pandemic will do things to our DNA. But we’re not using lentiviruses for the viral-vector vaccines – we’re using adenoviruses, because those explicitly do not work by inserting genes into host DNA. That’s also a feature of the mRNA vaccines: messenger RNA is not incorporated into our DNA. Those two species are constantly working in close proximity in living cells and there’s a huge pile of optimized protein machinery to keep them from getting crossed in that fashion. Nor does a messenger RNA sequence get turned back into DNA and inserted that way. Every cell has hundreds of thousands of mRNA molecules in it at any given time, and things would come to a catastrophic halt if these started getting reversed back into DNA sequences. (Our cells do have some RNA-to-DNA machinery in them, but it doesn’t work like that).
But for gene therapy, the opposite considerations apply – you most certainly want to insert new genes into human DNA, and you want it done quickly, efficiently, and right where you tell it to go. That last part is always the worry with any gene-insertion technique, be it some variety of CRISPR, zinc-finger nucleases, lentivirus vectors or what have you. This is one of the main reasons the human-editing experiment in China was so amazingly irresponsible, because our control over such things in a human embryo is just not acceptable yet. Not even close.
In fact, it’s tricky enough just in the stem cells pulled out of bone marrow. That’s one possibility for what Bluebird is seeing – that when they treated the patient’s extracted cells with their lentivirus vector, that some of the hemoglobin genetic data got mishandled and plopped into the wrong stretch of DNA, demolishing some other important gene’s function in the process. You can be sure that they’re sequencing the abnormal blood cells from these patients now to see if this shows up. The MDS patient from 2018 turned out not to have this problem, so it’s possible that these two just reported don’t, either. So what’s the problem, if not that?
Well, as mentioned, bone marrow transplantation is a grueling process no matter what. The process of (either mostly or completely) wiping out a person’s bone marrow stem cells involves severe treatments mixing chemotherapy with radiation, and one of the compounds used (and used by Bluebird) is called busulfan. The organic chemists in the crowd will find that one interesting: it’s the bis-mesylate of 1,4-butanediol, nothing more and nothing less, and if the thought of taking a reactive small molecule like that intravenously gives you the shivers, well, welcome to chemotherapy and get ready for some stuff that’s even worse. The thing is, busulfan itself is a Class I carcinogen (as one would expect from its structure). Many older chemotherapy agents are. They are destructive to cells, and the only way you would take any of them is if you have a population of cells that you actually want to see destroyed, and you are willing to take your chances that you can bear up under the collateral damage of doing that. So it’s certainly possible that the leukemia seen in Bluebird’s patients is at least partly driven by the bone marrow transplantation procedure rather than the gene alteration part. In case you’re wondering, this could well be happening with some bone marrow transplant patients who undergo this whole procedure to treat leukemia itself, in which case it lands silently in the “relapsed” category. No, you only do bone marrow transplants when there’s no alternative.
As that first link in today’s post (Adam Feuerstein at STAT) mentions, though, there’s ongoing research to make that part of the process less risky. Survival rates for bone marrow transplants in general have steadily improved over the years, and everyone knows that one of the rough parts is the pre-treatment. But that problem might or might not get solved in time to help out Bluebird (or to quell the worries that other gene-therapy outfits might have who are also targeting that hematopoietic tissue). If indeed it’s the problem in the first place. . .