X-Message-Number: 9762
From: Ralph Merkle <>
Subject: Repost from January 1993
Date: Sat, 23 May 1998 10:55:48 PDT
The following repost is from January of 1993:
--------------------------------------------------------------------
Comments On "The Effects of Cryopreservation on the Cat"
Mike Darwin recently posted "The Effects of Cryopreservation on the
Cat" by Michael Darwin, Jerry Leaf, and Hugh Hixon. This is "...a
research paper which is now in the (hopefully) final stages of
preparation for publication."
Before discussing the paper by Darwin et. al., we need to discuss
the general objectives of research in cryonics.
Fundamentally, cryonics aims to prevent the terminally ill patient
from dying when today's medical technology can't. We stabilize the
patient's condition as best we can using today's technology, and
then maintain the patient in a stable condition until future
medical technology can revive them. The method used is to store the
patient in liquid nitrogen.
It is generally agreed that the condition of a person stored in
liquid nitrogen is stable. However, the process of freezing
inflicts damage. The question of interest, therefore, is whether
future medical technology can revive the patient despite the damage
inflicted by the freezing process, and despite any other disease or
injury from which the patient might have suffered.
The first question is whether any future technology could revive
the frozen patient, even in principle. If revival is infeasible in
principle, then the resources allocated for cryonic suspension
might be better used for other activities (Bacchanalian orgies come
to mind). This line of thought leads to the information theoretic
criterion of death (discussed more extensively in "The Technical
Feasibility of Cryonics," by Ralph C. Merkle, Medical Hypotheses,
September 1992, 39, pages 6-16; and elsewhere). If information
theoretic death has occurred then cryonics won't work. The hopes,
dreams, memories, and personality of the person have been
obliterated in the information theoretic sense and no future
technology, regardless of how advanced, can recover them. They are
gone.
If, on the other hand, information theoretic death has not occurred
then revival is feasible in principle. In this case, we must ask
whether revival will prove feasible in practice given that it is
feasible in principle. Until relatively recently most people would
have said there's a big difference between "in principle" and "in
practice." Research in molecular manufacturing, however, tells us
that repair "in principle" and repair "in practice" will, at some
point in the future, likely become almost indistinguishable.
This point deserves further emphasis. The condition of a person
frozen at the temperature of liquid nitrogen is quite stable. The
recovery of the X, Y and Z coordinates of every atom in the frozen
structure should in principle be feasible. With the advent of the
STM and other scanning probe microscope techniques, it has become
clear that the recovery of such detailed structural information
(e.g., the coordinates of every atom) is possible.
Molecular manufacturing should let us economically build large
numbers of very small STM-like devices able to analyze frozen
tissue in complete detail. Thus, by combining low-cost
manufacturing with high-resolution imaging systems, we can recover
essentially all the information about the structure that can in
principle be recovered. By combining such analytical tools with
systems able to change molecular structure in a general way, we
will then have (metaphorically speaking) eyes with which to see and
hands with which to heal. Even damage at the molecular level will
then be susceptible to repair. As Feynman put it, "The problems of
chemistry and biology can be greatly helped if our ability to see
what we are doing, and to do things on an atomic level, is
ultimately developed -- a development which I think cannot be
avoided."
Thus, the fundamental objective of today's cryonic suspensions is
to maximize the likelihood of information theoretic survival. One
of the major purposes of research in cryonics is to determine if
current suspension methods prevent information theoretic death.
Before considering the other purposes of research in cryonics, it
is useful to ask what conclusions (if any) can be reached about
information theoretic survival based on the paper by Darwin et. al.
At the outset, I should say that I have not yet had an opportunity
to examine the pictures that the paper is based on, and so must base
my comments on the text and conversations with Mike. I have,
however, examined some of the pictures produced by Fahy of rabbit
brain (which are also relevant in the current discussion). I
should also say quite clearly that I greatly appreciate the work
and effort that went into this study, and feel that it makes a
valuable contribution to our understanding of the effects of
cryopreservation. The comparison that follows, illustrating the
great power of future technologies in comparison with current
technologies, should in no way be construed to mean that we should
not vigorously pursue research on the effects of cryopreservation.
It is intended as a caution against overinterpreting the
experimental data that we obtain with today's admittedly imperfect
methods.
We are faced with a fundamental dilemma in trying to determine with
today's technology whether future technologies will decide that
information theoretic death has (or has not) taken place. In the
future, we will have complete information about every molecule in
the frozen structure. In the present study, we have information
provided by light and transmission electron microscopy. Further,
the information is about the structure after it has been thawed,
fixed, and sectioned.
Even under the best of conditions, the information available using
transmission electron microscopy is grossly poorer than the
information that will be available in the future. We learn only the
electron density of the imaged section. If the section is 1 micron
thick (plausible, although the actual section thickness used in
the study was not specified) then resolution of detail much smaller
than 0.1 microns (0.1 of the section thickness) is difficult.
We will assume, however, that a resolution of one hundredth of the
section thickness, 0.01 microns or 10 nanometers, was in fact
achieved. Under these conditions a single pixel of our EM
photograph will be available for every rectangular block that is
one micron (or 1,000 nanometers) long by 10 nanometers by 10
nanometers. Assuming we are getting 10 bit grey scale data (again
optimistic), this means we have 10 bits of information for a volume
of 100,000 cubic nanometers. There are (very roughly) 100 atoms
per cubic nanometer, so we have 10 bits for 10,000,000 atoms, or 1
bit for 1,000,000 atoms. While we might reasonably debate the
number of bits of information that future systems will generate, it
is reasonable to suppose that they will give us at least 1 bit per
atom of raw information (again conservative), in the same way that
EM photography gives us 10 bits per pixel of raw data. We thus have
at least a factor of 1,000,000 less information about the frozen
structure when we are looking at an EM photograph than will be
available in the future. To put this a bit more graphically, if you
are looking at a computer screen with 1,000 x 1,000 resolution
(perhaps a 21" monitor) then our factor of 1,000,000 less
information is the same as taking a complex scene portrayed on this
21" screen and crushing it into a single dot with an equivalent
"screen size" of a 50th of an inch. The ability to discern
biologically significant structure despite such rudimentary imaging
methods is quite remarkable and speaks volumes about the great
redundancy in such structures.
The difference is actually more dramatic when we consider that,
qualitatively, an EM section gives us no depth information (unless
we use serial sections, a method not employed in the current
study). Thus, we have only a fuzzy view of the projection of a
single slab taken with a random orientation through a complex three
dimensional structure after it has been subjected to warming,
fixing, deglycerolization, and the mechanical insults of
sectioning.
While the paper did a good job of pointing out the inherent
weaknesses in the methods used, it should also be pointed out that
the method of fixation used for the control group (vascular
perfusion in situ) was different from the method used to fix the
experimental group (removal en bloc followed by immersion in a
fixative-containing solution). Vascular perfusion in situ is
superior and offers fewer opportunities to introduce artifacts. To
quote Palay et. al. from "Fixation of Neural Tissues for Electron
Microscopy by Perfusion with Solutions of Osmium Tetroxide:" "The
difference between the devastated tissue resulting from immersion
fixation and the coherent, intact tissue obtained by perfusion
fixation is at once evident on even superficial examination."
Thus, the introduction of artifacts because of this difference in
protocol between the experimental and the control group cannot be
excluded.
It should also be remembered that the 3 molar glycerol used is less
than the "Smith's Criterion," and is substantially less than the 5
to 6 molar glycerol used in today's suspensions. The ischemic cat
("FIGP" in the terminology of the paper) was subjected to a 24 hour
ischemic interval. Under good conditions today (certainly not
always achieved, but achieved sometimes) the ischemic interval can
be held to a few minutes. In the case of the terminally ill patient
who has elected to forego artificial methods of prolonging the
dieing process, suspension can quite literally be started shortly
after cessation of heartbeat at a point in time when the patient
could in fact be revived by current methods.
It is also clear that freezing to liquid nitrogen temperatures
introduces macroscopic fractures as the temperature is reduced
below the glass transition temperature (at about 130 Kelvins) to
the 77 Kelvins of liquid nitrogen. The paper also suggests that
smaller fractures exist. Fractures created at or below the glass
transistion temperature result in little or no loss of significant
structural information. From an information theoretic point of
view, provided that the tissue remains frozen both prior to and
during analysis, the presence of fractures is not a major concern
and is unlikely to cause information theoretic death. Upon
rewarming, however, such fractures will clearly contribute to
artifacts and result in loss of cellular contents and structure.
Again, the paper correctly points out this mechanism for the
introduction of artifacts, and proposes further studies which
freeze to the glass transition temperature and not below to
eliminate this source of error.
It would also be most advisable for a future revision of the paper
to state clearly that future analysis on the frozen structure is
anticipated. Attempting to infer the correct structure after
thawing significantly overstates the problem.
The paper clearly establishes that gross macroscopic fractures are
present. As mentioned earlier, fractures that occur at or below
the glass transition temperature are unlikely to cause significant
loss of structural information (though they are likely to preclude
revival by any technology short of a medical technology that fully
utilizes a mature nanotechnology). Unless one is eager to be
revived using primitive or intermediate technologies (which might
be able to cure most, though not all, injuries) the presence of
fractures is of little direct concern. Two secondary concerns are
(a) the greater deterioration that would occur if a (hopefully
unlikely) thaw-refreeze event were to occur and (b) the negative
appeal of fracture damage to many people. Should storage at the
glass transition temperature become available it would be
marginally preferable, but only if it were certain that it was as
reliable as the current highly reliable method (e.g., pouring
liquid nitrogen into a large dewar once every few weeks).
Blockage of the circulatory system in the ischemic group is of
serious concern. Long ischemic intervals are sometimes
unavoidable in current suspensions, particularly given the poor
social and legal environment. While blockages in the central
nervous system created "grossly visible infarcted areas..."
"...these were relatively few" and generally "no larger than 2 mm
to 3 mm in diameter..." This essentially means that the areas which
did not receive cryoprotection were subjected to a straight freeze.
While I would suspect that survival following a straight freeze is
likely, there is insufficient data at the present time to support
such a conclusion with confidence.
In the non-ischemic group, dehydration increased the difficulty of
identifying structures. As discussed previously, future methods
should have no difficulty in identifying structures that are
obscure today. Irregularly shaped cavities were present,
presumably formed during freezing by the growth of blocks of ice.
The slow growth of ice during freezing is likely to cause
compression of tissue. Compression is of little concern from an
information theoretic point of view. More significant damage
(e.g., tears, rips, or microfractures) were also present. Given
the evidence in other systems that substantial ice formation is
compatible with functional recovery, it is likely that either (a)
the observed damage is compatible with functional recovery, or (b)
it occurred after most of the water had frozen (and presumably
after most of the damage caused by freezing had occurred). In
either case, information theoretic loss should not be great.
The presence of tears or rips are also seen in the freeze
substituted preparations of rabbit brain prepared by Fahy. In that
case, the tears appeared to be "clean" with matching surfaces,
making hypothesis (b) more likely.
The presence of unidentified "organized debris" in the spaces
presumably created by ice during the freezing process might have
occurred either during freezing or thawing. The hypothesis that
the debris was moved to the space during freezing is complicated by
the observation that the space was, at that time, occupied by ice.
After thawing, the volume occupied by ice would become a small pool
of water. Anything which broke free from the wall or lining of such
a pool would then drift freely in the pool, thus creating debris.
An attractive and simple hypthesis for the formation of the debris
during thawing is available, while hypotheses for the formation
of debris during freezing face significant difficulties.
The increasingly dehydrated and confined regions between the blocks
of ice formed during freezing should make movement of any structure
of significant size quite difficult. While there might be some
concern that the currents created during freezing will result in
turbulent flow, this appears quite unlikely. [The following
discussion has been revised and corrected from the original post] The
approximate criterion for the onset of turbulence in a liquid volume
with characteristic size r is that the Reynolds number ~rdv/n exceed
a few thousand, where d is the density of the liquid, v the velocity of
the flow, and n the viscosity. The characteristic dimensions in a
cell are about a micron, the density is roughly a gram per cubic
centimeter (or 1,000 kilograms/cubic meter), the velocity is
probably much less than a meter per second (and probably much less
than a micron per second), and the viscosity of water at room
temperature is about 0.01 poise (or 0.001 newton-second/meter^2)
(viscosity increases both with decreasing temperature and with an
increasing concentration of glycerol, so 0.01 is conservative). This
produces a Reynolds number less than 1.
As the approximate criterion for the onset of turbulent flow is a
Reynolds number of a few thousand or more, this implies that
turbulent flow is implausible. When we take into account the fact
that the viscosity of tissue is larger than the viscosity of water
(assuming that we can apply "viscosity," in an approximate way,
to a structure as heterogenous as tissue), that the tissue is
loaded with glycerol (which has a higher viscosity than water),
that the whole system has been chilled to some low temperature
(which also increases viscosity) and that the velocity of flow will
typically be substantially less than a meter per second, then we
conclude that the actual Reynolds number will be significantly
smaller than 1. This reinforces the basic conclusion.
The "Encyclopedia of Physics" says "Blood flow in capillaries is laminar,
but water flow in household pipes is turbulent unless the flow is about
that allowed by a leaky faucet or less."
Another concern is that two small pieces of tissue, originally
adjacent, might be subjected to differential forces that would
cause one to move past the other for some distance. The examples
that have on occasion been used by Mike Darwin are boulders moved by
glaciers or bits of debris moved by the formation of ice on window
panes.
In both these cases, water is converted almost completely into ice.
In the case of a cryonic suspension in the presence of even moderate
amounts of cryoprotectant, freezing is incomplete and the result is
more similar to slush than ice. In addition, in both the examples
cited there was a rigid surface which provided an attachment point
for the ice. The idea that slush might create a differential force
on two adjacent points of tissue otherwise suspended in solution is
more difficult to envision.
The damage to the axons of myelinated nerve cells, secondary to a
failure of the cryprotectant to penetrate the myelination (many
layers of cell membranes wrapped tightly around the axon) is very
plausible. The function of a myelinated axon, however, is to carry
information (much like a wire). Complete obliteration of the axon,
analagous to damage to a wire, will result in little or no
information loss if the myelin sheath (somewhat like the insulation
around a wire) is still present. Myelinated axons are relatively
large, so even substantial damage to the axon would not obscure or
obliterate the pathway.
Turning to the ischemic group, we find that axons in myelinated
tissue were sometimes sufficiently damaged that "nothing but debris
remained." Even if the axonal debris were to prove unidentifiable,
the course of the neuron would still not be obscured. As discussed
earlier, however, the application of future analytical tools will
almost certainly resolve the nature of "unidentified" EM images.
In summary: the available evidence, though clearly incomplete, tends
to support the idea that information theoretic survival is likely even
when today's rather primitive suspension methods are used. This should
not be taken as a reason for discontinuing or ignoring research in
this area. Even a moderate risk of dying is unacceptable and
should be reduced.
A more pressing motivation for research has little if anything to
do with information theoretic survival per se. The most serious
risks to survival stem from the more or less complete failure of the
medical community to either understand or support cryonics. This
failure leads directly to preventable delays in initiating
suspensions, inadequate support for suspensions if they are
tolerated, etc. The single most effective method of decreasing the
risk of death would be to gain even a moderate level of acceptance
from the mainstream medical community. To gain such acceptance
will require a body of research which supports the idea that
suspension protocols do in fact provide a good chance of survival.
For various reasons it seems likely that such research will have to
provide almost conclusive evidence that cryonics is likely to work,
despite the obvious disadvantages of requiring "proof" that
cryonics works before using it. The idea that freezing a person is
a "risky" course of action while cremation and burial alive are
"conservative" is quite absurd, but also deeply entrenched.
Thus, along with information theoretic survival, the second main
objective of research is the fuzzier one of gaining general
acceptance by the medical community. Therefore, besides
understanding current suspension protocols and improving future
protocols measured against the criterion of information theoretic
survival, we must also understand and improve suspension protocols
measured against the (somewhat fuzzy) criterion used by the
mainstream medical community.
While it is more difficult to specify exactly what will be needed to
satisfy this second criterion, the two obvious objectives are (1)
demonstrate reversible cryopreservation of a mammalian brain, e.g.,
freeze and thaw the brain of an animal and show functional recovery
for at least a short period of time following thawing; or
(2) demonstrate that suspension techniques, while they do not
preserve function, provide good preservation of the structures that
are crucial to the correct functioning of the human brain and memory,
e.g., get pictures (either from light or electron microscopy) that
show good preservation of structure and ultrastructure.
There is much work to be done to develop this body of research, and
all efforts in this direction should be encouraged.
Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=9762