Two paths toward improved structural preservation in brain banking
You can try to make the highest-quality, more resource-intensive method better, or you can try to make the more scalable method better
It is common in biobanking to have different procedures depending on the available resources.
Vonsattel 1995 describes how having a basic protocol for hemibrain cryopreservation (en bloc freezing, #1 below) allowed for much more collaboration with remote pathologists. On the other hand, a more complicated protocol (dissection followed by freezing of sections using liquid nitrogen vapor, #2 below) was used when the brain was donated locally and staff were available to perform it:
In this post, I’m going to explore the idea of having a similar dichotomy for fixation procedures. Just to be clear, there are two major technical aspects of brain banking: (a) the initial preservation procedure and (b) storage. For this post, I’m focusing on fixation methods for the initial preservation procedure.
But first, what methods for the initial fixation in brain banking are available today?
Let’s first list out the main methods that one could use for the initial preservation procedure, as I see it today. Of course, you could subdivide these endlessly further, and the boundaries between them are far from distinct, but here’s one way of lumping categories of methods together.
1: Perfusion fixation in ideal animal laboratory conditions.
Perfusing fixatives through the brain’s vascular system is the gold standard in animal models and has a very long track record of efficacy. Notably, even in animals, there is often a period of post-fixation following perfusion fixation, where the brain sits in fixative and diffuses from inside and out.
While this method obviously isn’t directly relevant to human brain banking, I’m listing it here to make explicit that this literature is distinct from that of brain banking, where there are many more barriers to adequate perfusion. So the protocols and results here, while certainly helpful, can’t always be directly applied to brain banking.
In a rapid autopsy or medical aid in dying (MAID) brain donation case, it might be possible to have more similar circumstances as in laboratory animals. But even then, the conditions are likely to be much worse than is possible in laboratory animals, due to age, cerebrovascular morbidity, agonal changes, the minutes-long delay prior to the start of the procedure, etc.
2: Perfusion fixation in realistic postmortem human brain donation cases.
In theory, there are still plenty of good reasons to expect that perfusion fixation should improve preservation quality in many cases, as long as the perfusion is successful. In the race to halt postmortem decomposition via fixation, using the vascular system to distribute chemicals seems to be the fastest route available. And in practice, a review I was a part of found that there was moderate quality evidence that perfusion fixation improves preservation quality over a standard immersion fixation procedure for inner brain areas.
However, these upsides need to be balanced against the challenges of perfusion fixation. Among these are technical difficulties, the presence of pre-existing alterations in the cerebrovascular system inhibiting perfusion quality, and the potential for introducing artifacts.
There is much more discussion on the upsides and downsides of using perfusion fixation in brain banking available elsewhere. For the purposes of our discussion here, I think it’s fair to say that using perfusion fixation in realistic brain banking cases is promising but it is still far from a solved problem.
3: Ventricular perfusion or ventricular injection alongside fixative diffusion.
This is a bit of a niche technique but I think it is promising. The idea makes a lot of intuitive sense: if perfusion fixation is challenging, perhaps we can at least allow fixative solution to access the ventricles to speed up a diffusion-based fixation procedure. This would allow us to substantially increase the surface area through which fixatives diffuse into the parenchyma.
Let’s review a few of the articles that have used this.
Toga 1994 is perhaps the bull case for the ventricular delivery method. They compared two methods:
We tested 2 head fixation protocols. In the first protocol, the head was perfused through the common carotid arteries by a constant pressure system (120 mm Hg). The vasculature was flushed with normal saline (4 l) followed by 8% neutral buffered formalin (4-8 l) and the specimen postfixed in 8% formalin for several days to ensure thorough fixation. In the second protocol, the cerebral intraventricular space was directly accessed by metal cranial shunts into one occipital horn and both lateral ventricles anteriorly. Shunts were fixed in place with dental cement and the ventricular system perfused for 72 h with a slow flow of formalin fixative during immersion fixation of the entire specimen.
They found that the intraventricular system was more effective at delivering fixative evenly throughout the brain, which they attributed to erratic blood clot formation preventing adequate cerebrovascular perfusion.
Bass 1993 reported the use of ventricular injection in neonatal brains. They accessed the brain tissue through the anterior fontanelle and injected fixatives into the lateral ventricles:
From a technical perspective, the open anterior fontanelle in neonates makes this much easier. In adults, the skull would need to be dissected first, for example as is commonly done in neurosurgery during the placement of an extraventricular drain.
Ventricular injections can also sometimes be combined with perfusion fixation. For example, Sharma 2006 reported an injection of formalin through the base of the ventricles, alongside vascular perfusion. In our recent publication, we reported the injection of formalin into the ventricles through multiple routes, following vascular perfusion fixation.
4: Static immersion fixation of the whole brain at room temperature.
So far all of the other methods discussed are uncommon. What seems to be the most common method used to preserve brain tissue is to simply immerse it in a container of formalin. Here, I call this “static immersion fixation” because the brain is not moved during the procedure.
This method has major advantages in its simplicity and low cost. The major disadvantage is that it seems to take a substantial amount of time for the fixative to reach the inner areas of the brain, during which time it will be decomposing to a significant degree.
For example, Scott 2012 found that after 24 hours of immersion fixation with 20% formalin at room temperature, human brains seem to be unfixed in the center areas, while the outer areas are well-fixed, at least based on visible color changes:
How does this affect microscopic tissue quality? I don’t know. As far as I can tell, this is not a very well-settled question.
One study that compared perfusion to immersion fixation of whole brains, Grinberg 2008, reported the following:
Immersion fixation had a clear-cut gradient effect on the brains even after prolonged fixation (3 months). The more superficial tissue was well fixed, while the results on deep brain structures were unpredictable. On visual inspection, both thalamus and basal ganglia exhibited a reddish discoloration that proved to be autolytic changes after microscopic investigation. For this reason, immersion fixation was discontinued in favor of perfusion fixation that provided a more rapid and uniform penetration of the fixing agent into all parts of the brain.
But what exactly the degree of “autolytic changes” are, and whether that would prevent important research questions from being addressed in this brain tissue, is less clear.
A counter-example where preservation quality seems to be sufficient with this basic method is Krause 2016. In this study, whole brains were immersion fixed in formalin, with postmortem intervals from 18 to 50 hours prior to fixation.
They studied the anterior cingulate cortex (ACC), which is a fairly inner area of the brain that should not be quickly preserved by immersion fixation, with electron microscopy.
Microscopically, they found that some structures such as myelin had predictable artifacts, but that microtubules, neurofilaments, synapses, vesicles, and mitochondria were well preserved, and that neurons could be clearly depicted:
5: Static immersion fixation of a hemibrain specimen.
An important variant of the previous category is the immersion fixation of a hemibrain specimen. This is very common in brain banking because often half of the brain is cryopreserved and half is fixed. For example, this is what is used for fixation by the Vonsattel 1995 article described at the beginning of this post.
This approach substantially increases the speed of immersion fixation, both by decreasing the volume by half and by increasing access to the ventricular system.
To my knowledge, Adickes 1997 is the study with the most data on the perfusion vs immersion fixation comparison in hemibrains. Overall, they found that subjective histology quality was higher in the perfusion-fixed group.
Yet, to my eye, the histology quality of representative images from the immersion-fixed brain tissue doesn’t look terrible (left = perfusion, right = immersion; brain regions: top = frontal cortex, middle = hippocampus; bottom = basal ganglia) in the H&E images they show:
As they describe it: “Note the photographs on the left (perfusion) reveal, overall less cellular retraction, less pericellular and perivascular shrinkage, and a more dense neuropil with greater cellular detail.” The worst region in terms of preservation quality seems to be the basal ganglia, which is consistent with it being the innermost region of these three.
Is this the quality of preservation in this immersion fixed tissue so bad that connectivity cannot be traced? I’m really not sure. If pressed, my guess is no. It seems more like artifacts that could still allow inference of the original state. But this is just a guess, and better imaging data and image analysis tools to analyze that image data would be really helpful — basically essential — to answer that question more definitively.
6: Immersion fixation with acceleration techniques, such as agitation, increasing the volume of fixative, or refrigeration.
If immersion fixation is not as good, then one option is to try to do things to make it better. There are a lot of “acceleration” techniques whose efficacy in improving preservation quality in brain banking is largely unexplored.
Refrigeration is included here because even though it is expected to decrease diffusion speed, it is expected to more than makeup for this by decreasing the rate of postmortem decomposition in brain tissue that has not yet been fixed.
7: Immersion fixation of brain sections.
Finally, another approach is to section the brain tissue fresh and then immersion fix the sections separately. This allows for a much faster rate of fixative diffusion. The main downsides of this are damage at the cut interfaces and the inability to easily map circuitry across the brain unless advanced registration techniques make this possible. Beach 1987 found that even compared with this method, perfusion fixation had advantages.
§
Now that we have a basic sense of the different methods for the initial preservation in brain banking, we can discuss how they could be improved.
My sense is that there are two major pathways toward improvement:
(a) making the best methods better, and
(b) assessing how good the simplest approaches are and attempting to improve them in ways that would be possible to implement at scale. For example, by remote pathologists or funeral directors without access to expensive equipment.
A: The first goal is: how can we design a procedure that achieves as good of preservation as possible, without worrying about resource constraints?
This might involve:
Optimizing perfusion fixation performance with neuroimaging and iterating the procedure based on real-time feedback. For example, switching to delivery through the ventricles and/or accelerated immersion techniques if the perfusion appears low quality.
How can damage associated with preservative chemicals be minimized? For example, using the shortest fixation time possible to preserve antigenicity, while still maintaining tissue morphology.
Here, we already basically know what kind of preservation quality we are aiming for — the kind seen in ideal laboratory animal conditions where widespread fixation is started within minutes.
So improving methods for measuring preservation quality is likely not as big of a factor here, because simple methods for assessment, such as tactile measurements for fixation, or evaluating blood vessels for the presence of red blood cells, might be sufficient.
This “highest quality” procedure would likely only be possible in a brain bank with high investment and resources. This is because the field is very small and the equipment and technical skills required to perform this type of procedure would likely be — relatively speaking — quite costly.
In my view, the major open questions here are how to make perfusion fixation and ventricular perfusion procedures technically successful on a consistent basis.
B: The other goal is: how can we achieve “good enough” preservation quality for brain mapping with a method that is practical, cheap, and scalable?
The addition of real-world constraints makes the problem much different. Most of the focus immediately becomes on what is “good enough”. This, in turn, will obviously depend upon the investigator's goals.
Personally, I’m mostly talking about “good enough” preservation for connectivity and circuit mapping studies, since this seems to be a top priority for research today into neurobiological disorders.
For achieving the “best possible” preservation, it makes sense to start at the top with perfusion fixation and work one’s way down if the procedure is not working. But for achieving “good enough” preservation with the lowest cost, it may make more sense to start at the bottom, with standard immersion fixation, and only add complexity or expense to the procedure if it is necessary.
Because placing the isolated brain into a container of formalin is as close to as basic as one can imagine, it’s hard to imagine that things could be cheaper or easier than this. (A possible exception is a procedure that could be done with the brain still inside of the skull, although that adds its own complexities.)
So here, the question is largely about the measurement of brain tissue preservation quality. We first need to know if the standard methods for immersion fixation are good enough.
In my view, the major open question for this goal is how to measure whether existing brain preservation quality is sufficient for connectivity mapping. Quite plausibly, the answer might even be in the literature, but I haven’t yet found it myself.