August 18, 2016 at 03:47AM

Today I Learned: 1) Industrial science is serious. Today I learned about GMP, or general manufacturing practices. In scientific and medical industry labs, there's a more specific version called GLP -- general lab practices. The gist of GMP (and GLP) is that everything is SUPER STANDARDIZED and SUPER GOOD. I don't really know most of the details of GMP (or GLP), but it includes things like online, company-accessible, government auditable notebooks. For everyone. Specifically for medical labs, it also includes the use of *closed systems* wherever possible, where here, "closed systems" are systems physically isolated from people as much as possible -- everything's piped straight from one reactor to the next, until the last possible second. I presume this is to cut down on contamination and to improve reproducibility of manufacture. You know, I never fully appreciated how complex it is to produce a drug, much less a biologic like a stem cell or an antibody. Pretty much every step of the synthesis has to be done in-house, in a totally controlled and perfectly reproducible way, in a manner that scales with demand, with *no* contamination allowed. It's not just manufacture, either -- a drug company also has to market, sell, and distribute the drug. Autologous* stem cell therapies are possibly the most difficult case on the market right now. Here are some of the processes that an autologous stem cell therapy company has to provide: -- Sales to point-of-care locations (including payment processing) -- Registration of patient cells -- Transportation of patient cells (not trivial! Especially at scale!) -- Tracking and receipt of cell transportation to facility -- Internal sorting and processing of patient cells -- WHATEVER THE ACTUAL SCIENCE IS TO MAKE MODIFIED CELLS -- Internal sorting and processing of modified patient cells -- Shipping of modified cells -- Tracking and receipt of modified cells to point-of-care locations Note how many places there are where a line of cells could potentially be mixed up with another vial of identical-looking cells from another patient. That can't happen. Ever. And that "WHATEVER THE ACTUAL SCIENCE IS TO MAKE MODIFIED CELLS" bit? Yeah, that's done under GLP -- it has to be closed-loop, reliably timed, and certified free of mycoplasma and other common contaminants. Oh, but wait, it gets worse -- this entire process has to be designed and finalized during phase III trials (or phase II for accellerated trials -- one of the many reasons accellerated trials might not actually be good for a company). If you discover that there's an inefficiency in your manufacturing strategy during the trial, or you discover that you can't meet demand (or that you're manufacturing way more than is demanded), tough luck unless you want to convince your investors to start the phase III over again (not happening). It also means you can't start off making everything small-scale and then plan on figuring out scaling late in the process -- you have to consider this stuff pretty early in development to have a chance of making it all work. Investors (and the FDA) know this, so one of the things they look for froma company in the early stages of testing is some cognition of the difficulties of moving to GLP and scale manufacturing. * Derived from the patient's own cells. Contrast with "allosteric", which means the stem cells are standardized cells from a central stock. 2) Basically everyone I know who's worked with stem cells has told me that it's really up in the air whether those cells are actually what scientists claim them to be. Stem cells, you see, have a nasty habit of differentiating, since that's *what they do* in the body. Worse, some of these differentiation steps seem to be phenotypically silent -- you can't tell that they've changed by looking at them, they just won't differentiate the way you expect. This problem is particularly important for induced pluripotent stem cells (iPSCs, or iPSs), which are stem cells derived from differentiated adult tissue (commonly skin cells). The early claim was that iPSCs were identical to embryonic stem cells (ESCs), and therefore could be used in place of them without the need for expensive, difficult-to-work with embryos (or the ethical controversies that come with them). But are they really the same? Today I learned that yeah, they are. Mostly. Almost. A professor (and former industrial scientist!) named Jeanne Loring* got really obsessed with the question of iPSC identity**, and her lab has subjected iPSCs and eSCs to genomic, transcriptomic, and epigenetic characterization at multiple stages of culture and production. Here are some results: -- iPSCs are genetically identical to normal cells... -- ...but they don't stay that way. When induced, the cells are fine; when expanded in culture, they start to accumulate cancer-like mutations, especially gene duplications and *especially* deletions in p53 (a critical cell cycle checkpoint and one of the first things to go wrong in most cancers). Therefore, expansion is the most critical stage of production for iPSCs, and will require the most stringent controls. -- iPSC gene expression clusters very nicely with eSC gene expression. The Loring lab put together a phone app for predicting stem-cell-ness from RNA-seq data. I'm not sure how you're supposed to pipe RNA-seq data through a phone, but that's another story.... -- The epigentics of iPSCs are *almost* normal, as validated by a screen of about 50,000 known epigenetic markers. The major differences are that, consistently, 1) Barr bodies are unravelled, so both X chromosomes are active in female cells, and 2) genetic imprinting (permanent silencing of either the maternal or paternal allele of a gene) is completely wiped. These changes do make some sense -- they're exactly what you would expect from a cell that just underwent reversion to an embryonic state. Interestingly, Loring is one of the few scientists I've encountered in medical areas who advocates for *more* stringent regulations by the FDA in her field. As of right now, the only QC the FDA requires in terms of validating the identity of stem cells is that they're karyotyped (their chromosomes are counted). Which is nice, I guess... but yeah, it seems like some additional validation would be nice. * very friendly woman, btw, if cheerfully blunt. ** The reason *why* she was interested in this question is really worthy of its own TIL entry. Loring was involved in some early (1990s-2000s) stem cell therapy trials for treating Parkinson's Disease. They harvested dopaminergic neurons from a tiny sliver of an aborted fetus's brain, and implanted those neurons in PD patients. The trial had wildly mixed results -- on the one hand, some of the patients were pretty much completely cured, and lived out the rest of their lives drug-free, continuously improving in behavioral measures for at least 14 years after the surgery. On the other hand, a bunch of them developed additional dyskinesia (involuntary tremors), which made their quality of life significantly worse. Overall promising, but a dud. Loring is now quite convinced that the original failures were due to non-dopaminergic neuron contamination from when the neurons were extracted from the fetus. The issue of quality control turned into a quite-publishable 10-year project. 3) Many kinds of common cancers commonly metastasize to the brain (~30$ of breast cancer metastases, for example). I'm not sure if that's because brain is a particularly fertile ground for cancers (less immune system access?) or if cancers commonly metastasize *everywhere*.