Back to homepage
In autoimmune conditions, such as rheumatoid
arthritis, the body produces antibodies to its own tissues -
"autoantibodies". Until now it has not been clear why people make
autoantibodies or whether autoantibodies cause the illness. This webpage
gives an new explanation for autoimmunity and suggests a new treatment that
may produce long term relief for conditions such as rheumatoid arthritis
and lupus. Promising results obtained in early trials
of this treatment are summarised on the linked page "Update on B cell
depletion therapy.
Our conclusion is that
autoantbodies are first made by chance but can set up a vicious cycle so
that they go on being produced in large amounts, causing damage to body
tissues. Removing the cells that make antibodies should break this cycle.
It is now possible to remove most of these cells safely and effectively.
Although we almost certainly need to increase the power of the treatment to
remove all the unwanted cells, preliminary results are encouraging.
To explain how autoimmunity
may work it is necessary to go into some detail about the immune system.
However, we have tried to make this reasonably easy to follow.
The immune system includes two types of white blood
cell, B and T cells, which recognise the difference between "self"
molecules belonging to the body and "non-self" molecules from bacteria and
viruses. B cells make antibodies which should only attach to non-self molecules.
Each B cell makes antibodies to one type of
non-self molecule. However, it can only do so if given "help" by T cells
recognising parts of the same non-self molecule. T cells are, therefore,
seen as directing antibody production.
Because T cells direct normal immunity, it has been thought that
autoimmunity starts with T cells mistaking self for non-self. These T cells
would then direct the production of autoantibodies by helping B cells which
recognise self molecules. This sort of autoimmunity can be created in
animals if the animal is immunised with self molecules, but usually dies
down after a few weeks. This is quite different from autoimmunity in people
and we think that human autoimmunity works in a different way.
Most forms of autoimmunity in humans start up for no obvious reason,
apparently by chance. Once autoimmunity has developed it often goes on for
ever. There is nothing to suggest that anything from outside the body is
fooling the T cells into confusing self and non-self. Hereditary factors
make some people more at risk than others, but even within "autoimmune
families" there is no way of knowing who will develop disease or when. This
suggests that human autoimmunity starts with a chance event within the
immune system, which develops into a perpetual vicious cycle. We believe
that the chance event is the production of an autoantibody which has the very unusual ability to
stimulate its own production.
The first reason for thinking that the chance event which
starts autoimmunity is the production of a new type of antibody is that all
new antibodies are made by chance. B cells cannot deliberately make an
antibody to a molecule. All they can do is make random changes in the
antibody they are already making. If the changed antibody binds better to a
non-self molecule the B cell is encouraged to go on making antibody. If the
new antibody binds less well or binds to a self molecule the B cell is told
to die. However, under rare circumstances, instead of dying, B cells making
autoantibodies may be able to keep themselves alive.
This idea does not mean that T cells are not involved, because B cells can
only make antibody with help from T cells. It means that T cells must be
fooled into helping B cells to make autoantibodies, not by something from
outside the body, but by autoantibodies themselves. The autoantibodies also
need to fool the B cells, because B cells have their own means of checking
they are making the right antibodies. Very few autoantibodies are likely to
be able to fool both systems. This explains why autoimmunity does not
develop all the time and why autoantibodies are only made to a few self
molecules.
Even if autoantibodies are made they may cause no harm. Many people make enough autoantibodies to be measurable in blood samples. (This can cause confusion because some people have a positive rheumatoid arthritis autoantibody blood test but never get arthritis.) As we get older we accumulate autoantibodies. These are probably autoantibodies which can stimulate their own production but are of no importance because they are harmless to the tissues. In most cases, if an antibody attaches to a self molecule, it is probably cleared away without any problem. However, certain autoantibodies are dangerous. Increasingly, it is possible to see how different autoantibodies can cause particular illnesses. The mechanisms are pretty clear for rheumatoid arthritis and lupus, but for other conditions there is still some way to go.
In rheumatoid arthritis there are antibodies against
antibodies - which is perhaps the most obvious way of setting up a vicious
cycle. Antibodies come in five types: IgM, IgG, IgA, IgD or IgE. The
autoantibodies in rheumatoid arthritis, known as rheumatoid factors, are
antibodies against IgG.
The usual blood test for rheumatoid arthritis measures IgM antibodies
against IgG (IgM-anti-IgG). These antibodies are probably harmless because
they do not get out of the bloodstream and can be cleared away by a set of
molecules called the complement system. However, some antibodies in
rheumatoid arthritis are IgG anti-IgG i.e. they bind to themselves. These
are more difficult to measure but are likely to be more important.
Pairs of IgG anti-IgG antibodies stuck to each other are dangerous because they are not cleared away by the complement system. They are small enough to pass out of the bloodstream into the tissues. Once in certain tissues they probably cause inflammation by stimulating macrophage cells. They do this by binding to a receptor, known as CD16a (or FcRIIIa), on the macrophage surface. Macrophages only carry this receptor in the tissues affected by rheumatoid arthritis, including joint lining, the lungs and the pericardial lining of the heart, so inflammation only occurs in these tissues. The situation is made worse in the joints by the fact that, during inflammation, cells which make anti-IgG antibodies come into the joint lining and make autoantibodies inside the joint.
Click on image to view:
It has been known for a long time that B cells that
make anti-IgG antibodies can "borrow" help from T cells which recognise
non-self molecules, breaking the normal rules.
A B cell normally gets help only from a T cell that recognises the same
non-self molecule. The B cell does this by carrying some antibody on its
surface. This surface antibody picks up the non-self molecule, which the
cell then digests. The broken down fragments of the non-self molecule are
then "presented", attached to a carrier molecule on the B cell surface, to
nearby T cells. If a T cell recognises the fragment as non-self it gives
the B cell a "help" signal to make more antibody. (The B cell divides and
daughter cells known as plasma cells make most of the antibody.)
B cells that make anti-IgG antibodies (rheumatoid factors) can cheat the
system. The anti-IgG antibodies on the B cell surface pick up IgG
molecules, which are themselves antibodies already attached to non-self
molecules. The B cell digests the non-self molecule, presents it, and gets
help from a T cell recognising a non-self molecule fragment. The carrier
molecule that the B cell uses to present fragments, known as MHC Class II,
is different in different people, with the result that molecule fragments
are presented differently. Perhaps not surprisingly, the type of MHC Class
II a person carries has an effect on how likely they are to develop
rheumatoid arthritis. Other genetic factors probably influence the
likelihood of the development of arthritis , but they are as yet
unknown.
Luckily, there is another checking system which
ensures that B cells that make anti-IgG antibodies only last for short
periods. If a B cell is given help by a T cell it then moves into an area
of a lymph node (lymph gland) called a follicle centre. If at this time it
picks up unattached self molecules with its surface antibody it will die.
This means certain death for most B cells that make anti-IgG antibodies
because there is plenty of unattached IgG around.
Unfortunately, if the B cell is making IgG anti-IgG antibodies the
situation is more complicated. When a B cell first makes antibody it makes
the IgM type. Before it can change to making IgG antibody the B cell has to
run the gauntlet of the follicle centre. This is probably why very few
people have B cells that make IgG anti-IgG, they will have died before they
have a chance to change to IgG. However, very occasionally it may be
possible for B cells already making useful IgG antibodies to a non-self
molecule from a virus or bacteria to change to making IgG anti-IgG
antibodies by making a random change in the structure of their antibody.
These cells can return to the follicle centre bathed not only in unattached
IgG but also in their own IgG anti-IgG antibodies clumped together.
Although picking up unattached molecules via cell surface antibody is
lethal to the B cell in the follicle centre, picking up molecules stuck to
antibody (and complement, which follows the antibody attachment) will keep
the B cell alive. This means that a B cell that makes IgG anti-IgG may
survive the death signals from unattached IgG. In other words, if an
anti-IgG antibody producing B cell has switched to making IgG, its own
antibodies may keep it alive both in the company of T cells and in the
follicle centre so that it can produce more antibody. The vicious cycle is
complete.
Click on image to view:
Lupus is a condition rather like rheumatoid arthritis
but in lupus several sorts of autoantibodies can occur. This and other complexities
of lupus suggest that whereas rheumatoid arthritis may be driven by a single
vicious cycle lupus may be driven by connected antibody cycles - like a Swiss
watch.
One type of antibody
that may be particularly important is antibody to a complement system molecule called
C1q. The complement system has three uses. It can kill bacteria. It clears away
antibodies bound to unwanted material. It is also involved in selecting
which B cells are allowed to survive in the follicle centre and make
antibodies. Binding of autoantibodies to C1q has the effect of using up
complement proteins. There are probably other ways in which complement is
used up in lupus as well. If complement has been used up, people have difficulty fighting
infections. They cannot clear away antibodies bound to unwanted material in
the blood, which end up depositing in the walls of blood vessels in tissues
such as the kidney. The production of other autoantibodies in lupus may be
due partly to poor clearance of unwanted material and partly to poor
selection of antibody-forming B cells.
People with lupus have repeated attacks of illness of different types,
depending on how much the complement system has been used up and what
autoantibodies are being made. For instance, antibodies to red blood cells
can cause anaemia, antibodies to DNA are often associated with kidney
problems, and antibodies to phospholipids in the blood can cause
thrombosis.
When an antibody binds to a molecule, complement
system molecules tend to attach as well. C1q is the first to attach. The
details are quite complicated, but this means that the ways in which
rheumatoid factor anti-IgG antibodies can fool T and B cells are also
available to anti-C1q antibodies. A B cell making anti-C1q antibodies can
pick up non-self molecules stuck to antibody and C1q, just as the anti-IgG
B cell can. In the follicle centre anti-C1q B cells will tend to find C1q
bound to antibody and the key rescuing complement molecule, called
C3d.
Another vicious cycle has already been mentioned. Anti-C1q antibodies help
to generate other autoantibodies by using up complement. These other
autoantibodies can also use up complement by binding to self molecules.
Lupus probably represents a whole series of vicious cycles which can feed
off each other and which have one thing in common - running out of
complement.
The importance of complement is supported by the fact that people who are
born lacking complement proteins often get lupus. A genetic lack of the C1q
protein almost always leads to a form of lupus. In most lupus sufferers
there is no genetic lack of complement but there is increasing evidence of
other genetic factors which may alter the balance of the immune system.
These may affect complement, messenger molecules known as cytokines, or
molecules on the surface of cells through which cells interact, such as
CD40.
The anti-phospholipid antibody syndrome is another condition which may
depend on very similar vicious cycles to those in lupus. Thrombosis and
miscarriage are the commonest features. The two conditions often occur
together. It may be that low complement levels can generate vicious cycles
either with the antibodies to DNA which are typical of lupus, or antibodies
to phospholipid.
In several autoimmune conditions autoantibodies are
found which bind to molecules found inside the nucleus of cells. These
molecules are often involved in the handling of DNA in chromosomes or the
rather similar RNA, which carries the genetic code to be translated for
making proteins. Most antibodies cannot get in to cells, but a few can. By
getting into cells and attaching to target proteins autoantibodies of this
type may damage the cell when it is dividing or making protein. The cells
lining blood vessels are bathed in a high concentration of antibody because
they are exposed directly to blood plasma. This may be why in the condition
scleroderma, in which there are antibodies to proteins used for cell
division, small blood vessels die and cannot be replaced. Antibodies to
molecules involved in protein synthesis are often linked to the muscle
disease polymyositis, which may be because muscle cells make large amounts
of protein all the time. In childhood dermatomyositis antibodies to other
molecules linked to DNA and RNA occur with both blood vessel damage similar
to scleroderma and muscle damage.
Antibodies to proteins found in cells are associated with other conditions.
In Sjögren's syndrome, which consist of dryness of the mouth and eyes, and often arthritis,
the molecule is called Ro. In one form of juvenile arthritis there are
antibodies to a molecule called DEK. It is not known why antibodies to
these molecules cause the patterns of illness we see. However, an important
point is that when illness is due to antibodies getting in to cells blood
tests for inflammation, such as the ESR, may well be normal. This may lead
to a delay in diagnosis because the physician may only take note of
autoantibodies if the ESR test is high.
How these antibodies might cause vicious cycles is not certain but several
possibilities exist. If autoantibodies cause cells to die while dividing,
cell proteins may be released alongside a "danger" signal which fools the T
cells into treating them as non-self. If autoantibodies were to interfere
with molecules involved in cell survival or activity they might also
persuade both T and B cells to allow the autoantibody producing B cells to
stay alive. In Sjögren's syndrome excessive survival of B cells is a major
feature - often with excessive production of all sorts of antibodies. In
some cases this gets to the stage that groups of B cells get completely out
of control and behave like a malignant lymphoma.
The most clearly understood effect of autoantibodies
is at the site where signals from nerves are transmitted to muscle - the
muscle end plate. Antibodies to the signalling receptor cause a condition
of muscle weakness known as myasthenia gravis. The antibodies prevent the
signalling chemicals from nerves reaching the muscle. Antibodies to
receptors on thyroid gland cells are probably also the cause of excessive
production of thyroid hormone, in Grave's disease, or
thyrotoxicosis.
The vicious cycles in these conditions may involve antibody binding to
receptors on other cells involved in the immune system. The muscle end
plate acetylcholine receptors are also present on cells in the thymus. The
thymus is the organ where T cells undergo their "education" i.e. they
choose which molecule they will recognise and are selected according to
whether they see self or non-self. The unique feature of myasthenia is that
the B cells making autoantibody take over the thymus gland where they
interact with T cells and also form follicle centres. It seems that the
antibodies to the receptor cause changes in the thymus that completely
confuse the normal rules about B cell survival. If the thymus gland is
removed myasthenia often gets better, but it is not a reliable
cure.
Certain other conditions, including multiple sclerosis, sarcoidosis and Crohn's disease give the impression of being autoimmune, but are not linked to typical autoantibodies. However, unusual types of antibody do occur in each. There is more of a suggestion that these conditions are triggered by immunity to non-self molecules from viruses, bacteria or foodstuffs, but antibody-based vicious cycles may still be important.
In multiple sclerosis, antibodies are made within the
fluid bathing the brain and spinal cord. The antibodies come from a small
group of B cells and tend to remain the same throughout the illness. Nobody
knows what they bind to. This may not be the point. The brain and spinal
cord cannot cope with large amounts of any type of antibody in the
surrounding fluid. Large amounts of antibody may cause phagocytic
microglial cells to destroy the myelin which surrounds nerves. Although it
has been thought that damage to the brain in multiple sclerosis is due to a
direct attack by T cells, recent studies suggest that when the damage first
starts antibody and active microglia may be present, but not T
cells.
Multiple sclerosis is more common in some places than others and sometimes
seems tooccur in epidemics. One explanation for this is that it depends on
a previous infection with a virus such as measles. The problem may be that
some of the B cells making antibody to the virus make antibodies which give
their B cell the ability to get into and live in the brain. The ability of
B cells to get into tissues and survive there depends on the binding of one
cell to another through surface proteins called adhesion molecules. Viruses
often mimic or bind to these adhesion molecules. It may be that the B cells
that survive in the brain make antibody which also mimics an adhesion
molecule. If so, surface antibody on the B cells may not only allow the
cells to survive in the brain, but may also lead to activation of microglia
or other brain cells. This would be not so much an autoantibody stimulating
its own production as an antibody creating the possibility of its own
production in the brain.
The cells that make the antibodies found in
the fluid around the brain in mutiple sclerosis appear to be only found in
the brain and spinal cord. However, they must come from cells in the
lymphoid system. There may be a pool of cells outside the nervous system
which do not make antibody most of the time, but every now and again some
may get into the brain or spinal cord and start making antibody. This may
result in a patch of myelin damage or "demyelination" which is the cause of
symptoms in multiple sclerosis.
In sarcoidosis, clusters of phagocytic macrophage
cells form in various body tissues, apparently stimulated by material they
have taken up and by nearby T lymphocytes. However, in people with
sarcoidosis, there is no reaction if non-self proteins are injected into
the skin, suggesting that the T cells have lost the ability to respond
properly to non-self proteins. Even more strangely, particles from the
tissue of one person with sarcoidosis injected into the skin of another
sufferer cause a cluster of macrophages to form (the Kveim
phenomenon).
It seems likely that in sarcoidosis some form of autoantibody is binding to
the surface of T cells and interfering with the way T cells interact with
cells such as macrophages and B cells, known as accessory cells
because they help T cells to function. People with sarcoidosis often have
antibodies to a type of sugar molecule (galactose) which is present
attached to proteins on cell surfaces. The presence of this sugar on the
surface of T cells is closely linked to their level of activity. The sugar
is also present on B cells, but only when they are in follicle
centres.
A number of plants make proteins called lectins which bind to specific
sugars. Galactose is bound by a lectin in peanuts called PNA. When lectins
bind to sugars on cells they can have powerful effects on the way the cell
behaves. As suggested by Pilatte and Lambre, the autoantibodies in sarcoid
may be interfering with the behaviour of T cells by behaving like lectins.
The effect would seem to be to cause T cells to overstimulate accessory
cells, but fail in other functions, and perhaps to cause follicle centre B
cells to stimulate each other. This would set the scene for a vicious cycle
of antibody production, explain the formation of macrophage clusters in the
tissues and also the lack of response to non-self proteins.
In Crohn's disease, similar clusters of macrophages,
(called epithelioid granulomas) occur to those in sarcoidosis, but in the
bowel. There is, therefore reason to think that the two diseases have a
similar mechanism. Perhaps not surprisingly, people with Crohn's disease
have antibodies to sugar molecules of a different type (mannose) in a
specific grouping. Interestingly, the bacteria that cause rather similar
clusters of macrophages (tuberculous granulomas or "tubercles") in
tuberculosis use mannose and galactose containing sugar molecules on their
surface to interfere with the waymacrophages and T cells behave.
Mannose sugars are present on the T cells' surface receptor molecules. The
lectin concanavalin A binds to these sugars and is a very powerful
stimulator of T cells. Anti-mannose antibodies in Crohn's disease may well
behave like concanavalin A. The effect would be similar to that with
anti-galactose antibodies in sarcoidosis. In both conditions the disease
looks as if it is driven by T cells but the T cells are probably in turn
being driven by the lectin-like antibodies. B cells do not carry the same
mannose-bearing receptors as T cells. Continued production of anti-mannose
antibodies requires stimulation of B cells and it is likely that this
requires mannose from bacteria, which is probably only available in the
gut. This would explain why the anti-mannose B cells develop mostly in the
lining of the gut and the abnormal behaviour of T cells is mostly seen in
the gut.
Diabetes in young people is often caused by a lack of insulin due to damage to the islet cells of the pancreas. Autoantibodies to these cells are often found. However, both in diabetes and some forms of thyroid disease it may be that the damage to the tissue is done by T cells directly. Experiments in animals suggest that under unusual circumstances excessive numbers of T cells can accumulate in the pancreatic islets, rather in the way that excessive numbers of B cells can accumulate in the thymus, brain or joint in different autoimmune diseases. This suggests that diabetes may involve a vicious cycle involving T cells alone. Nevertheless it is possible that the autoantibodies present contribute to this cycle.
It appears that many autoimmune diseases may result
from the chance production of antibodies which stimulate their own
production in various ways. It is clear that only a tiny number of
antibodies can do this - perhaps one in a million million possible types of
antibody. We make many million types every day. Over seventy years our
chances of getting autoimmunity are about one in twenty to one in fifty. It
is likely that the antibody not only has to bind to the wrong molecule, it
must be of the right subtype and bind in exactly the right way. Making one
of these antibodies is just bad luck.
If so, autoimmune diseases may be a bit like bugs in a computer programme.
Most of the time the programme can take cope with any numbers you put into
it and it will do its job. However, if you happen to press certain keys in
a particular order (i.e. make a dangerous antibody) the programme crashes -
often going in to a "loop" in which it keeps doing the same thing for ever.
The solution is to turn everything off, wipe out all the numbers you have
put in and start up afresh. For autoimmune disease this may mean getting
rid of all the B cells in the body, which will include the ones making the
dangerous antibodies causing the loop, and start again. Fortunately, this
may be possible.
Most autoimmune diseases are being treated at present
with drugs which damp down the activity of immune cells. There may be
improvement, but in the long term the treatment has to go on being given.
It is a bit like keeping the weeds in the lawn down by mowing - it works
quite well but the weeds always come back.
In the last few years a number of people have suggested that we should
think in terms of curing autoimmune disorders. This is partly because we
feel we know enough about the immune system to think this might be possible
and partly because successful cures have been achieved in some other
conditions. Hodgkin's disease, for instance, which is a disease of immune
system cells, can quite often be cured. Moreover, certain autoimmune
conditions seem to be cured by the very high doses of chemotherapy given to
patients who need bone marrow transplants. The evidence is not clear, but
some people with rheumatoid arthritis who have had bone marrow transplants
for other reasons find that their arthritis may go away for long periods
and perhaps for good.
If it is accepted that it may be possible to produce permanent remission in
autoimmune disorders then it seems fair to say that we should direct our
efforts to this rather than simply to damping the condition down. The key
question is how to achieve long term remission safely and surely. We
believe that clearing away B cells is the best hope, for the reasons given
above, but other ideas have already been tried.
One approach has been to try to get rid of T cells. However, if the
explanation given here is correct this is not very logical. It is also
likely to be dangerous because T cells carry our immune memory and are not
readily replaced. Removing T cells can produce some benefit, but now that
many studies have been done, the general impression is that it has no
lasting effect - rather as we would expect.
A second approach is to get rid of all immune cells, on the basis that we
do not know for sure which ones are most important. This is what happens
when people have high dose chemotherapy before getting a bone marrow
transplant. The same high dose chemotherapy is now being used in patients
with autoimmunity. The difference is that they are being given their own
bone marrow back - so there is no risk of rejection of the marrow. So far,
several patients seem to have had benefit, but no clear picture has
emerged. Very often patients are given rather smaller doses of chemotherapy
to avoid too much risk. Unfortunately this may mean that they do not get
lasting benefit.
The approach we are favouring, getting rid of B cells,
has not been tried for very long in autoimmunity, but has been tried in other diseases.
It has some enormous advantages, quite apart from being logical. If only B
cells are removed, there is no immediate risk from infection, of the sort that occurs
if all immune cells are killed. The phagocytic cells that destroy bacteria
are unharmed. There is enough antibody in the blood to last for a good
while. This means that it is theoretically feasible to get rid of every
single B cell without any danger. This is quite different from T cells. If
a high proportion of T cells are removed immunodeficiency results. (The
AIDS virus causes illness by removing T cells.) What is more, B cells
regrow very quickly. After about 6 months the numbers are back to normal.
If all B cells can be removed,
new B cells will be making a different set of antibodies and, hopefully, will be
directed by T cells only to make useful antibodies.
If the theory is right, after removing all B cells, the chance of rheumatoid arthritis
occurring a second time should be about the same as the chance in an identical twin, which we know
to be about one in four. It may well be less, since it may take many years
for the conditions to develop which will allow an autoimmune vicious cycle
to occur. The situation for some of the other autoimmune disorders may be a
bit different. If the T cells have developed immunity to a particular virus
or bacteria this may help set a vicious cycle going again.
Depletion of B cells has been used extensively in people with malignant growths
of B cells called non-Hodgkin lymphoma. Taking away normal B cells does not seem
to be a major problem. B cell depletion is now increasingly being tried in autoimmune diseases,
(update 2001) including rheumatoid arthritis, IgM-associated neuropathy (a disease of
the nerves), lupus, myasthenia, dermatomyositis, and several blood disorders.
Interestingly the best way to get specific depletion of B cells is to use a drug which is
itself an antibody to B cells, such as rituximab. Rituximab on its own removes many B cells, but
the results are even better if it is
combined with other drugs which have a more general effect on immune cells.
Lymphoma sufferers receiving this combination have a good chance of
their disease being put into remission for many months and perhaps
years.
There are reasons for thinking that treatment of autoimmunity may be easier than
treatment of malignant growths. To cure a malignant disease it may be necessary to kill every
single diseased cell. In autoimmunity it may only be necessary to kill
sufficient cells to break the vicious cycle. However, an important question at present
is just how many cells we have to get rid of to get a long lasting effect in autoimmunity.
It may be 90%, 95%, or even 99.5%. It will
probably be different in different autoimmune conditions. 90% is probably
well within the scope of current technology. 95% may be at the limit, but
more efficent techniques should come along fairly soon. Treatment of autoimmunity has so far been
cautious, in that we have tried not to use a lot of potentially toxic drugs. Rituximab has been
used either on its own or with steroids or cyclophosphamide in fairly small doses. Although some
patients remain well for up to two and a half years, it looks as if, at least in rheumatoid arthritis,
we are not yet using a powerful enough mixture of drugs to get really long term remission.
Repeating the treatment seems to be succesful but we are still very much looking for long
term benefit from a single course of treatment. Work is continuing at present and some further
detail is given on the linked page "Update on B cell depletion therapy".
