4 Neural Stem Cells
When Ramón y Cajal wrote that “everything may die, nothing may be regenerated,”1 he may well have meant it literally, but such an interpretation would not be credible now. Since Ramón y Cajal’s time, we’ve learned that the brain has considerable plasticity, and many cellular elements—dendrites, spines, synapses—are constantly being generated and regenerated. But what about cells? There are stem cells in the brain that produce neurons. How do they behave in response to injury? In chapter 2, we asked: why doesn’t brain repair work? Perhaps we first need to ask: just how poor is the brain at repair?
In fact, the stem cell niches of the brain do respond to injury and disease. Studies in experimental animals show that the production and survival of new neurons is increased in response to damage following stroke, for example. One of the earliest such reports came in 2002 from Olle Lindvall and colleagues in Sweden.2 Using the technology we discussed in chapter 3, they identified newly generated neurons in adult rats that had suffered a stroke. This was induced by blocking a major cerebral artery, mimicking human stroke pathology. Four weeks after the lesion, they observed a 31-fold increase in the number of newly formed neurons detectable in the striatum—the brain area most damaged by the stroke. Moreover, markers confirmed that many of the newly formed cells were of the striatum’s major projection neuronal type—medium spiny neurons—not normally produced in the adult striatum.
That doesn’t sound so poor: a 31-fold increase sounds like a pretty substantial response. But are we looking at the right statistic? Remember, there is little neurogenesis in the undamaged striatum—just the few interneurons that we discussed in the last chapter. So, a 31-fold increase of not very much is still not very much. In fact, Lindvall and colleagues estimated that “the fraction of dead striatal neurons that has been replaced by the new neurons at 6 weeks after insult is small—only about 0.2 percent.”3 Not quite so impressive.
The researchers also noticed another problem: few of the newly formed neurons survived for very long. The stem cells in the subependymal niche were responding to the damage, but the new neurons they were generating were quickly dying off. Perhaps they were losing their way as they maneuvered out of the niche; perhaps the striatal tissue was just too damaged to accept them; or perhaps they were the wrong kind of neurons.
The Lindvall experiment had the virtue of reproducing a human disorder—stroke—as accurately as possible in an experimental model, but it had the disadvantage that stroke pathology is pretty messy. In both experimental animals and human patients, stroke damage is variable in extent and location, and, as we’ve already seen in chapter 2, what follows a stroke is pretty close to chaos. What if researchers presented the brain’s stem cells with a more precise challenge? What if they took out only a single cell type and left the remaining tissue intact?
This is just what Jeffrey Macklis, Sanjay Magavi, and Blair Leavitt at Harvard engineered.4 Through a clever manipulation, they managed to kill just pyramidal neurons of the mouse cerebral cortex. They injected fluorescent nanoparticles into the thalamus, a part of the brain with which the pyramidal neurons connect. The nanoparticles were taken up by the terminals of the pyramidal cells and transported back to the cell bodies, so that just pyramidal cells were labeled. Shining a laser onto the surface of the cortex, activated the fluorophores on the nanoparticles, producing a high-energy form of oxygen, which in turn killed the labeled cells. Hence the researchers were able to destroy just the targeted pyramidal cells, leaving the rest of the cortex intact.
As in the 2002 Lindvall experiment, more new neurons were generated than were seen in undamaged cortex. Moreover, Macklis, Magavi, and Leavitt were able to show that these new neurons didn’t just look like new pyramidal projection neurons, they acted like them, too, with some actually connecting to the thalamus. But again, the number of these new neurons was tiny. Moreover, other researchers have had trouble replicating the result of the Macklis, Magavi, and Leavitt experiment,5 so even this reported minimal cell replacement is in doubt.
Of a dog walking on its hind legs, Samuel Johnson famously observed that it might not be well done, but the surprise is to find it done at all. The same might be said of the neurogenic response to brain injury. It’s pretty inadequate, but, given the pessimism around cell replacement in the brain, the surprise is to find any replacement at all.
The Right Kind of Neurons
Neuroscientists face a conundrum when doing such experiments. They know that normally, the adult brain produces few neurons of any particular type. They damage the brain, then they see more. But what exactly has happened? They’d like to believe that the stem cell niche has reacted to the damage, acting to restore the lost neurons. This would be the targeted cell replacement that we see in the hematopoietic system. But there’s another possibility. Perhaps the niche stem cells are producing the same neurons they usually make— olfactory neurons in the case of the subependymal zone—but these neurons are straying off course because of the damage. Neurons are notoriously promiscuous; they’ll try to wire up with the right targets, but if they aren’t available, they’ll wire up with any that are. (As the old Stephen Stills song goes: “If you can’t be with the one you love, honey, love the one you’re with.”) The result might be that some of the newly formed neurons end up in the space that used to be occupied by the lost cells. Superficially, this might look like cell replacement—the ectopic cells are in the right slots, but like redundant factory workers now stacking supermarket shelves, their hearts just aren’t in it.
Thus the first question we need to answer here is: are the niche stem cells really making the right kind of neurons, or are these the wrong kind of neurons turning up in the right place? They say that if it looks like a duck, walks like a duck, and quacks like a duck, then it probably is a duck. In principle, it’s the same with neurons. If the new neurons can be shown to have the right properties for their replacement role, then it’s probably safe to assume they’re really doing the job. The problem, however, is that the data are often inadequate: researchers, so to speak, have to decide whether it’s a duck, even though they’ve neither seen it walk nor heard it quack
In the Macklis, Magavi, and Leavitt 2000 experiment, the new neurons really did seem to have gotten it right. Not only did they look like pyramidal neurons, but they also projected to the thalamus as pyramidal neurons should. Yet for all that, there was no evidence of correct physiological activity in this experiment. Which is to say, the new neurons might have achieved everything “duck-like” except “the quack.” Appropriate physiological activity would have been difficult to demonstrate. Again, in the 2002 Lindvall experiment, the new neurons looked the part, but it would have been equally difficult to show that they were actually contributing anything useful to neural circuits. In neither experiment would it have been reasonable to have claimed that the new neurons had contributed to brain repair, although I note that, in their 2002 study, Lindvall and colleagues do claim to “provide the first evidence that the adult brain can use neuronal replacement from endogenous precursors to repair itself after stroke.”6
Researchers mostly rely on markers to help them work out what cells have become. But as we’ve already seen, these markers can be misleading. Sometimes markers are expressed by more than one cell type. And sometimes they’re not cell type markers but maturation markers. Rather than being expressed by just a single cell type, they’re expressed on multiple cell types at a particular stage of development. They represent the phase the cells are going through rather than the cells themselves. This phenomenon has been described for some of the markers typically used in the striatum. The 2009 study by Fang Liu and colleagues,7 for example, suggested that some cells in the damaged striatum might have been misidentified because the markers were not as reliable as was presumed. Moreover, expression of markers might change when tissue gets damaged. Cells might begin to express a marker in response to injury that they never express normally. Finally, of course, markers can’t usually be interpreted to mean that the cells are doing anything functional. The marker says that the cells looks like ducks, but it doesn’t say they can quack.
All this has put pressure on researchers to devise better means to assess neuronal roles and functions. We’ll encounter optogenetics, one such exciting development, in chapter 10 when we consider its application to stem cell therapies, but first let’s conclude this discussion by asking whether we can distinguish between any of the alternative explanations for the increased production of new neurons discussed above: is there any direct evidence to be had that neurogenesis can switch on as a consequence of brain damage?
Switching Fate
Probably the cleanest experiment to answer this question was performed by Fiona Doetsch and Constance Scharff in 2001.8 When they looked at the bird motor control system we considered in chapter 3, they saw, as did we, that there are neurons in the higher vocal center (HVc) nucleus that project to the nucleus robustus archiststriatalis (RA)—which Doetsch and Sharff referred to as the “HVc→RA projection”—and that these neurons undergo normal continuous replacement, just like the granule neurons of the mammalian dentate gyrus. It turns out that there’s also another population of HVc neurons that project to a different nucleus called “area X.” But the neurons of the HVc→X projection are not normally replaced during adulthood. So the adult stem cell niche of the bird provides the higher vocal center nucleus with new HVc→RA neurons that project to the nucleus robustus archiststriatalis, but not new HVc→X neurons that project to the area X nucleus.
But what happens if either of these two projections gets damaged? Making use of the same clever stratagem that Macklis, Magavi, and Leavitt used to kill off specific populations of neurons in their 2000 experiment, Doetsch and Scharff injected fluorescent nanoparticles into first one then the other of the two target populations, so that either the HVc→RA and the HVc→X populations would each be labeled with the nanoparticles. And, just like the Macklis team, Doetsch and Scharff used a laser to kill either one or the other of these target populations. How did the stem cell niche respond?
The two researchers found that when the HVc→RA neurons were killed, the stem cell niche responded with a considerable increase in its production of new neurons. Though initially the HVc→RA projection was severely compromised by the damage, three months later, it had been restored to normal. By contrast, when the HVc→X neurons were killed, the HVc→X projection never recovered. In other words, responding to brain damage, the stem cell niche could produce more neurons of the type it was already committed to producing, but it couldn’t switch to produce neurons of another type.
It’s as if the stem cell niche were a ball-bearing factory. If demand for ball bearings goes up, the factory just makes more. But if there’s suddenly an increased demand for paperclips, then too bad. Maybe they don’t have the equipment to make paperclips, or the expertise. Maybe the marketing department doesn’t even recognize that there is an increased demand for paperclips: all they watch is ball-bearing sales. This seems to be how the forebrain stem cell niche operates.
We have to be careful not to extrapolate too far from this single result achieved by Doetsch and Scharff. Nonetheless, retooling appears to be one of the major challenges in the field of neural stem cell therapy. How can neural stem cells be coaxed out of their comfort zone? How can they be made to generate neurons they don’t otherwise produce?
The Predicament
So we have a frustrating situation. Demonstrably, there are neural stem cells in the adult brain. We’ve dispelled the theory that adult neurogenesis is incompatible with complex brain circuits. So there is at least the possibility that if neural stem cells could respond to cell loss—like the bone marrow stem cells do—then some replacement could take place. But like the ball-bearing factory, the niche is too stuck in its ways.
Neuroscientists see two ways forward. One is to work out how to drive the brain stem cells to make the cells required: in other words, to convince the ball-bearing factory to make paperclips. Considerable progress has been made in this direction in recent years, albeit only in experimental animals. It turns out that there are several stem cell types in the brain that might be coaxed into taking on this job, and we’ll meet them in chapter 12, when we consider how a patient’s own neural stem cells could be repurposed directly to generate new neurons.
The other approach, with the longer heritage, is to bypass the brain’s ineffectual stem cells and put the right cells in the right place directly. We’ll discuss this shortly, but first let’s look a bit more closely at how difficult it can be to define “the right cells in the right place.”
Parkinson’s Disease and “Dopaminergic Cells”
Some diseases have a pathology that is so iniquitous that one is inclined to imagine that it was diabolically conceived. Parkinson’s disease is such a disorder. It is characterized by a slow loss of motor control. In the early stages, the signs might be slight—a tremor, a failure to swing the arms evenly while walking—but this loss increases insidiously. The face freezes into a mask; balance is lost; and walking halts abruptly. Eventually, many sufferers simply get stuck in the “off” state. Many sufferers may be walking across a room when they lose the ability to maintain the movement and are left rooted to the spot.
These symptom are the result of our now familiar adversary, neuronal loss, and the cells in question have the grand title of the dopaminergic neurons of the substantia nigra pars compacta. These nerve cells die, for reasons that scientists are still struggling to clarify, and since they play a pivotal role in maintaining the brain circuits that initiate voluntary motion, a movement disorder is the clinical consequence. But the truly diabolical touch is this: by the time the patient gets a diagnosis, close to half of this neuronal population has already been lost.9 Even if we could devise a strategy to save afflicted neurons, by the time the disease presents, the patient has probably lost over two hundred thousand dopaminergic neurons. Ironically, as the disease progresses, the rate of neuron loss seems to slow, but, by then of course, it’s too late. The damage is done.
If ever there was a need for cell replacement it is here,10 and this is in fact where cell transplantation into the brain began, at least in recent times.11 But it began with a mistaken oversimplification that we need to grasp if we’re to understand the challenge of cell therapy.
Normally, dopaminergic neurons form a pathway from the substantia nigra to the striatum, a brain region we met in the last chapter (figure 4.1), where they connect with striatal neurons and release dopamine, their neurotransmitter. In one of the most expansive pathways in the brain, between a quarter and half a million dopaminergic neurons make this projection. This is not a particularly large number in the context of the human brain, but each of these cells forms between one and two million connections with its target striatal cells. The distance from the substantia nigra to the striatum is probably less than ten centimeters, yet each dopaminergic neuron elaborates some four and a half meters of axon; indicating just how much branching and rebranching their processes undergo.12 This makes these neurons among the most ramifying cells in the whole brain, a neural fountain that showers the striatum in dopamine.
Dopamine and Parkinson’s disease. Dopaminergic neurons of the substantia nigra normally project widely through the striatum and forebrain (right side of figure). In Parkinson’s disease, however, these dopaminergic neurons die, and this projection is lost (left side of figure).
Neural connectivity is normally thought to be very specific. Neurons make precise synapses on their target cells, and deliver their neurotransmitter within a tenth of a micrometer or so of these targets. Moreover, neurons have mechanisms to isolate synapses, so that the transmitter signals don’t spread too far from their intended sites of action. But just the density of dopaminergic synapses in the striatum raises the possibility that the dopamine pathway may be different. If cells could be introduced into the striatum that would inundate the tissue with dopamine, even without specific synaptic connections, perhaps this would counter the effect of dopaminergic cell loss.
Existing drug therapy suggested this might work. Patients can’t be administered with dopamine directly because its charged structure cannot penetrate the brain. They can, however, receive it indirectly by mouth, as L-DOPA, a precursor the brain then converts into dopamine. This treatment with L-DOPA been so successful since its introduction in the 1960s, that it has become the frontline treatment for Parkinson’s disease.
If just boosting the dopaminergic signal is so effective, what if cells secreting dopamine could be engrafted into the striatum? This was attempted in patients in the 1980s, first in Sweden and later in a number of other countries. Had these researchers used proper dopaminergic neurons for their initial trials, the history of cell therapy might have been different, but, instead, the first cells they used were adrenal medulla cells. Now the adrenal medulla is widely known as the source of adrenaline (epinephrine), the hormone behind the famous “fight or flight” response, whereby the body prepares for action in response to a potential threat. In addition to adrenaline, however, the adrenal medulla also makes dopamine. In fact, these two catecholamines are very closely related. So although transplanting adrenal medulla cells into the brain at first sounds alarming, there’s a clear logic to it. If just secreting dopamine into the striatum would be efficacious in treating Parkinson’s patients, what does it matter if it comes from dopaminergic neurons or from adrenal medulla cells, or indeed from any old dopamine-producing cells?
The problem arose when researchers failed to realize that while this was a reasonable (if somewhat unconventional) hypothesis, there were lots of reasons why it might not be true. First, the adrenal medullary cells were not brain cells, and might not behave appropriately in that strange environment. Second, even though adrenal medulla cells can indeed become neuron-like in culture, the neurons they resemble most are sympathetic neurons, not striatal dopaminergic neurons, and while they make some dopamine, they make mostly noradrenaline. You might imagine that some serious research would ensue before such a therapy found its way into the clinic.
In fact, the first convincing report of adrenal medulla implants working in rats with experimentally induced parkinsonism appeared in 1981. The first patient treatment with the same technique emerged the following year. Then, through the 1980s, a series of Parkinson’s patients—first in Sweden, then in Mexico—were given transplants of their own adrenal medulla tissue engrafted into the striatum. Eventually, several hundred patients were treated with this approach. Though initial reports were positive, even enthusiastic,13 it subsequently transpired that there was little or no efficacy. There was even a suggestion of unacceptable side effects and increased morbidity and mortality.14 The most telling observation was that the engrafted tissue didn’t even survive for long, so any dopamine replacement could only have been transient. Clearly, the transplant was not doing what was intended, and the therapy was quite rapidly withdrawn..
What did we learn from this episode? First quite simply, we learnt that if we were going to pursue this approach, we needed a better source of dopaminergic cells. The field quickly moved on to proper dopaminergic neurons taken from the substantia nigra itself. This has proven to be a more robust strategy, though also controversial at times. We’ll look more closely at this avenue of research in the next chapter.
The second point would seem a fairly obvious one. The logistical underpinnings for the therapy were weak. Little had been done to optimize graft size, location, or viability. Grafts weren’t standardized. The surgery had not been optimized. Patient selection had not been thought through very carefully. These points may seem obvious, but cell therapy still struggles to get these basic parameters right. As we’ll see when we return to Parkinson’s disease in chapter 5, when substantial controlled trials were finally undertaken in the 1990s, they were still subject to this criticism.
For many scientists and their critics, there were more profound lessons that needed to be learned; you shouldn’t go to clinical trial until you are absolutely sure you know what you are doing. The preclinical data supporting the medullary approach did not provide a clear neurobiological understanding of how the treatment was working. The supposed mode of action of the cells had not been clearly elucidated. The US Food and Drug Administration (FDA) defines “mode of action” as “the means by which a product achieves an intended therapeutic effect of action.”15 As such, it is the pivotal concept around which any therapy is constructed: what is this treatment doing to the body to bring about a positive change? But the simplicity of the concept hides a quite contentious issue: how much do researchers need to understand about how a potential therapy works before they try it on patients?
A pretty good idea, you might imagine, but that is not necessarily the case. Most treatments, be they conventional drugs or more novel therapeutics, emerge from some sort of screening, usually an animal model, or a cellular or chemical assay. This screening would identify a property of the therapy thought to be of value in treatment. In the Parkinson’s example we have been considering, this was an animal model: the dopaminergic neurons on one side of a rat’s brain are killed experimentally to mimic the cell loss seen in Parkinson’s patients. As a consequence, the rat loses motor coordination on one side of its body, while retaining control on the other. This asymmetry is exacerbated by injecting the rat with a dopaminergic drug such as amphetamine. In response to this stimulus, the rat starts to chase its tail, round and round. If a therapy significantly reduces the frequency of this rotation, it is deemed to have efficacy. You might think this is some way from treating elderly Parkinson’s patients, but despite its gross artificiality, it has been enormously useful in developing potential therapies. Significantly, regulators such as the Food and Drug Administration (FDA) in the United States or the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom consider this an “approved model”; that is, they are inclined to allow potential therapies to enter clinical trials if they have proved robustly effective using this model. But note, the assay tells you little about the mode of action. All you know from this assay is that the turning behavior has been reversed. You don’t know how the drug—or the implanted cells—brought about this effect.
You might imagine that regulators would require researchers to have a pretty definitive understanding of the mode of action of a novel therapeutic before allowing it into the clinic. Indeed, you might think that the clinical researchers themselves would want such an understanding. It might come as a surprise therefore to learn that this is not actually the case. Certainly, a regulator will expect some rational explanation for the therapeutic approach, but not usually a confirmed mode of action. Why? Because doing so simply sets the bar too high. Many drugs work through poorly understood mechanisms. The time-honored example, of course, is aspirin, an enormously popular drug through the first half of the twentieth century, long before the discovery in the 1970s of prostaglandins, the signalling molecules that we now think mediate aspirin’s effects.
Other examples relevant to our consideration of brain disorders would be the class of antidepressant drugs called “selective serotonin re-uptake inhibitors” (SSRIs). These popular drugs, which include the iconic Prozac, were designed to work in line with the “monoamine hypothesis” of depression, which proposed that depression was associated with a reduced availability of the neurotransmitter, serotonin. The SSRI’s were supposed to work by increasing the concentration of serotonin in the synapse, and were licenced as medicines with that as the proposed mode of action. But from the outset there were data that didn’t conform with this idea, not least the fact that such a drug should act quickly whereas many depressed patients take weeks to respond. We now know that SSRIs have multiple actions in the brain. Interestingly, one way they act is to stimulate hippocampal neurogenesis. There is evidence now that depression might involve a decrease in the production of new dentate gyrus granule neurons and that antidepressants work by reversing this decrease. So, in this and many other cases, drugs were approved with a mode of action that subsequent research showed to be inaccurate.
Rather than requiring a firm mode of action, researchers and regulators actually start from the position that the therapy should be demonstrably safe and efficacious. There should be sufficient data to suggest both that there is a minimal risk of harm to participants in the first clinical trials. Then there should be robust efficacy data from an approved model (such as the rat model just discussed) to suggest that there is at least a reasonable chance the therapy might work in patients. In practice, this entails a delicate risk-benefit analysis. Is the potential benefit of a new therapy worth the risk inherent in clinical trials? Regarding the mode of action, regulators will usually be satisfied with a credible narrative plus a clear program of research, running concurrently with the trial. But is this practice good enough for distinctly novel therapies such as stem cell therapies? This is a question that becomes particularly acute in regard to the therapies we will consider later, which went to clinical trial at a time when it was quite clear that the mode of action was not cell replacement, and was indeed quite unknown.
Controls
If the mode of action of cellular therapies has been a source of ongoing debate, the issue of controls has been even more troublesome. In conventional drug trials, in addition to the group of patients receiving the new medicine, there would usually be a placebo control group. These patients would be treated precisely the same as those in the drug-treated broup, except that the pill or the injection each received would not contain the active drug substance. The need for this control group is clear: the researchers need to be able to distinguish an actual effect of the drug from the “placebo” effect, the positive effect that sometimes follows simply from receiving clinical attention. Of course, having a placebo control group involves deception: patients don’t know whether they’re receiving the active medicine or nothing more than a sugar pill. For conventional drug trials, the deception is generally (though not universally) thought to be ethically acceptable. For one thing, all the patients would have given their informed consent to the structure of the trials, and for another, their taking a harmless placebo should put them at no significant risk.
In the case of surgery, particularly neurosurgery, the situation is different. Consider what is involved. The placebo arm of the trial should mimic the “active” arm as closely as possible, so patients would need to be anesthetized, taken into the operating room, have a stereotaxic apparatus attached to their heads, a hole drilled into their skulls, then be injected with fluid. Is this ethical?
In fact, the first cell therapy trials for Parkinson’s disease had no control arm. This remains the case today for the planned early-stage trials of stem cell therapies, as indeed for most Phase 1 drug trials. But, ultimately, without randomly controlled trials, new medicines highly unlikely to receive regulatory approval. More importantly, without this control, we can never be sure any efficacy is the consequence of the cells and not a placebo effect driven by the high expectations raised by stem cell therapy in desperate, severely afflicted patients
These ethical and practical concerns will be considered further in a later chapter. But first, let’s look at how the dopaminergic approach recovered from this shaky start.
Notes
1. Santiago Ramón y Cajal, as quoted in Rubin, “Changing Brains,” 410. http://doi.org/10.1017/S1745855209990330.
2. Arvidsson, A., Collin, T., Kirik, D., Kokaia, Z., and Lindvall, O., “Neuronal Replacement from Endogenous Precursors in the Adult Brain after Stroke,” Nature Medicine 8, no. 9 (2002): 963–970. http://doi.org/10.1038/nm747.
3. Arvidsson et al., “Neuronal Replacement from Endogenous Precursors,” 968.
4. Magavi, S. S., Leavitt, B. R., and Macklis, J. D., “Induction of Neurogenesis in the Neocortex of Adult Mice,” Nature 405, no. 6789 (2000): 951–955. http://doi.org/10.1038/35016083.
5. Diaz, F., McKeehan, N., Kang, W., and Hébert, J. M., “Apoptosis of Glutamatergic Neurons Fails to Trigger a Neurogenic Response in the Adult Neocortex,” Journal of Neuroscience 33, no. 15 (2013): 6278–6284. http://doi.org/10.1523/JNEUROSCI.5885-12.2013.
6. Arvidsson et al., “Neuronal Replacement from Endogenous Precursors,” 967.
7. Liu, F., You, Y., Li, X., Ma, T., Nie, Y., Wei, B., et al., “Brain Injury Does Not Alter the Intrinsic Differentiation Potential of Adult Neuroblasts,” Journal of Neuroscience 29, no. 16 (2009): 5075–5087. http://doi.org/10.1523/JNEUROSCI.0201-09.2009.
8. Doetsch, F., and Scharff, C., “Challenges for Brain Repair: Insights from Adult Neurogenesis in Birds and Mammals,” Brain Behavior and Evolution 58, no. 5 (2001): 306–322.
9. Kordower, J. H., Olanow, C. W., Dodiya, H. B., Chu, Y., Beach, T. G., Adler, C. H., et al., “Disease Duration and the Integrity of the Nigrostriatal System in Parkinson’s Disease,” Brain 136, no. 8 (2013): 2419–2431. http://doi.org/10.1093/brain/awt192.
10. Also valuable would be a diagnostic that could detect the neuron loss before it became so extensive. For a discussion of prospects in this regard, see Berg, D., Lang, A. E., Postuma, R. B., Maetzler, W., Deutsch, G., Gasser, T., et al., “Changing the Research Criteria for the Diagnosis of Parkinson’s Disease: Obstacles and Opportunities,” Lancet Neurology 12, no. 5 (2013): 514–524. http://doi.org/10.1016/S1474-4422(13)70047-4.
11. Although cell transplantation studies actually date back to the nineteenth century, they have had a significant therapeutic history only since the 1970s. See Barker, R. A., Parmar, M., Kirkeby, A., Björklund, A., Thompson, L., and Brundin, P., “Are Stem Cell-Based Therapies for Parkinson’s Disease Ready for the Clinic in 2016?,” Journal of Parkinson’s Disease 6, no. 1 (2016): 57–63.
12. Bolam, J. P., and Pissadaki, E. K., “Living on the Edge with Too Many Mouths to Feed: Why Dopamine Neurons Die,” Movement Disorders 27, no. 12 (2012): 1478–1483. http://doi.org/10.1002/mds.25135.
13. The initial report was the subject of an enthusiastic editorial. See Moore, R. Y., “Parkinson’s Disease—A New Therapy?,” New England Journal of Medicine 316, no.14(1987): 872–873.
14. See Dunnett, S. B., Björklund, A., and Lindvall, O., “Cell Therapy in Parkinson’s Disease—Stop or Go?,” Nature Reviews Neuroscience 2, no. 5 (2001): 365–369.
15. See US Food and Drug Administration, Code of Federal Regulations, Title 21, revised as of April 1, 2018. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=3.2.