See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/8669446

A Mathematical Model of the Dynamics of Odontogenic Cyst Growth

Article in Analytical and quantitative cytology and histology / the International Academy of Cytology [and] American Society of Cytology · March 2004
Source: PubMed

CITATIONS READS

18 436

4 authors, including:

John Ward Vanesa Magar

Loughborough University Ensenada Center for Scientific Research and Higher Education
49 PUBLICATIONS 1,740 CITATIONS 83 PUBLICATIONS 751 CITATIONS

SEE PROFILE SEE PROFILE

Susan Franks
University of Nottingham
17 PUBLICATIONS 573 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

coastal erosion and protection View project

PhD Project View project

All content following this page was uploaded by Vanesa Magar on 20 May 2014.

The user has requested enhancement of the downloaded file.

 Report on a problem studied at the UK Mathematics-in-Medicine Study Group Nottingham 2002
< http://www.maths-in-medicine.org/uk/2002/dental-cysts/ >

2 A model on the dynamics of odontogenic cyst

growth
Vanesa Magar (Cambridge University), John Ward (Loughborough University), Susan
Franks (University of Nottingham), Gabriel Landini (The University of Birmingham).

Abstract
Odontogenic cysts are non-neoplastic lesions that form as a consequence of the prolifera-
tion of epithelial residues that remain embedded in the maxillary bones after formation
of dental tissues. The epithelial rests proliferate in response to activating cell mediator
signals released during inflammatory processes (for example as a consequence of a tooth
infection). In this report we discuss the pathogenesis of odontogenic cysts based on cur-
rent knowledge and formulate a simple mathematical model of cystic growth. The aim of
the modelling is to establish the dynamics (i.e. the long term behaviour) of cyst enlarge-
ment based on osmotic pressure differences between the cyst contents and the stomal
extracellular fluid. Such osmotic gradient results in water movement into the cyst across
the epithelium lining which acts as a semi-permeable membrane. The modelling assumes
a (spherical) cyst with a semi-permeable shell of living cells and a core consisting of wa-
ter and a generic osmotic material (fed by the continuous death of epithelial cells in the
shell). The lining cells are assumed to behave like a Maxwell fluid, reflecting the action
of physical stresses by the surrounding cyst capsule formed by fibroblasts and collagen
fibres. The model couples the cyst radius and the osmotic pressure differences resulting
in a system of two nonlinear ordinary differential equations. Using the combination of
asymptotic and numerical techniques it is shown that in all parameter regimes the long
time behaviour of the cyst is the same and that linear radial expansion is predicted.
Moreover, the model predicts that in the early and intermediate stages of cystic growth,
osmotic pressure differences play an important role; however, for very large cysts, this
role is negligible as cell birth dominates growth.

2.1 Introduction
Odontogenic cysts are in general slow growing and pose no major threat to human life.
However, since they grow within the maxillary bones, they may cause bone or tooth re-
sorption, bone expansion, fracture or tooth migration. Odontogenic cysts associated with
the roots of non-vital teeth are called radicular cysts. These arise from the proliferation
of the so-called epithelial rests of Malassez. Such epithelial rests (small clusters of cells)
are the remnants of the root sheath of Hertwig which is responsible for the formation of
the roots of teeth. The epithelial rests remain dormant during the life of the individual
and are located between the root and the jaw bone, in the space occupied by the peri-
odontal ligament. However the rests may proliferate in the presence of an inflammatory
process (most commonly due to complications of caries followed by infection of the tooth
pulp). Keratocysts, on the other hand, comprise about 10% of all diagnosed cysts and
are thought to be of developmental origin [3]. This type of cyst grows more rapidly than
the radicular cysts and cell division activity is more commonly observed in the epithelial
lining. The process of active division of cells, known as mitosis, is thought to be more
important in the growth of keratocysts than in radicular cysts. Furthermore, keratocyst
exhibit epithelial budding (growth into the stromal compartment of the cyst wall) with
two consequences: secondary (or satellite) cysts may develop from these localised pro-
liferation foci, and recurrences after surgery are high if the cyst wall is not completely
removed. In this report, we present a model of cystic growth dynamics and consider
briefly how the model includes both the effects of mitosis and cell lining loss. Then, we
describe the implications the model results suggest for cystic growth dynamics. In par-
ticular, during the early stages, the cyst growth rate depends on the physical parameters
included in the model; however after a sufficiently long time has elapsed, all cysts grow
linearly with time. Finally, we suggest some experiments which would test the assump-
tions of our modelling approach and describe some mechanisms the current model does
not include that may alter the conclusions drawn from this study.

2.2 Background
Odontogenic cysts form epithelial residues of the tooth-forming organ. While many of
these residual epithelial cells remain quiescent throughout life, however, cell mediators
and signalling molecules released during an inflammatory process (for example tooth
microbial infection as a consequence of dental caries) may trigger the proliferation of the
epithelial cell rests.
Cyst formation. The epithelial proliferation may reach a critical size at which the
diffusion of nutrients within the mass of cells cannot keep pace with the metabolic util-
isation of the nutrients (epithelial tissue is not vascularised and nutrients reach all cells
by diffusion from the vascular stroma). The lack of nutrients results in degeneration and
death of the central cells in the proliferating mass with subsequent liquefaction, creating
a new cavity in the epithelial mass. However cells closer to the stroma remain vital.
Cyst growth. The degradation of the central cells in the cyst results in an increased
osmotic pressure in comparison to the osmotic pressure on the stroma surrounding the
cyst. This gradient is believed to draw a water flow into the cavity (to balance the osmotic
pressure), inducing an increase in hydrostatic pressure inside the cyst in comparison to the
stroma. This volume expansion stretches the epithelial layer and peripheral cells divide
further to maintain the epithelial lining intact. Since the epithelial lining regenerates, the
osmotic pressure difference will be maintained through the constant shedding of cells into
the lumen. At the same time, the stroma reorganises around the cyst wall and produces
a capsule relatively rich in collagen fibres.

2.2.1 Clinical data

Experimental data regarding cysts is hard to come by since investigation without treat-
ment provision raises a number of ethical issues. Cystic growth is slow, with a radial
expansion of only a few millimetres a year [2]. The mitotic rate is small in radicular
cysts, meaning that, only a small proportion of living cells are undergoing mitosis at
any one time. There have been a few studies aimed at comparing the osmolality, the
molar quantity of “osmotically active” molecules per litre, between cyst fluid and serum
(i.e. normal tissue fluid). The study of Toller (1970, [1]) showed that in most of the
cases tested the osmolality is higher in cyst fluid (mean 0.291 Osml ≡ 0.291 mol) than
in serum (mean 0.280 Osml); these values being consistent across several types of cyst.
The osmotic pressure difference ∆Po relates to the difference in osmolality, in this case
0.011Osml, via the formula
∆Po = ∆M RT,
where ∆M is the molar concentration difference of‘ “osmotically active” molecules, R is
the ideal gas constant and T is the absolute temperature. Using ∆M = 0.011mol/m3 ,
T = 310K and R = 8.31J/mol.K the osmotic pressure is, in this case, approximately
Po = 28.3 N/m2 . As will be evident in the modelling to come, the osmotic pressure will
change as the cyst develops.

2.3 Mathematical model

Given that the timescale of cystic growth of the order of years, water is drawn inside the
cavity almost instantaneously, so that the hydrostatic pressure difference, between the
cavity and the environment, balances the osmotic pressure difference between the same
two media,
Ph+ − Ph− = Po+ − Po− ,
where Ph+ and Po+ are the hydrostatic and osmotic pressures, respectively, in the cav-
ity, while Ph− and Po− are the hydrostatic and osmotic pressures, respectively, in the
surroundings.
Also, the supply of nutrients remains sufficient to maintain the epithelial cells alive due
to the living cells being in contact, or close to, the medium containing the nutrients (such
as the local vasculature). Moreover, we assume that the cell lining is a semi-permeable
membrane, which allows the passage of water and of nutrients to the cavity; but the
degraded material driving the osmosis does not cross the epithelial layer. Therefore, if s
is the total amount of degraded material inside the cyst, then the rate of change of s is
proportional to the surface area of the covering epithelium. So, if R is the cyst radius,
then
ds
= 4πβR2.
dt
The parameter β, a supply rate of degraded material into the cavity, will change according
to the type of cyst: the value of β will be smaller in a cyst which is loosing its cell lining
slower than in a “normal” radicular cyst – growing as a response to environmental signals
only, and larger for keratocysts in which the epithelial cells actively divide.
We assume, further, that the concentration of degraded cells is uniform inside the cyst.
If we consider that each epithelial cell contains the same amount of osmotically active
substance, the osmotic pressure becomes proportional to a concentration of degraded
material. Also, the concentration outside the cell is assumed constant. This implies that
the osmotic pressure in the surroundings is constant as well, and that the jump in osmotic
pressures across the epithelial lining is proportional to the difference in concentration of
degraded cells, that is,
Po+ − Po− = α (C − C0 ) ,
s
where C the concentration of material inside the cyst, that is, C = , and C0 the
4πR3 /3
concentration outside it, assumed constant; α is the proportionality constant.
It is worth noting that most cysts are roughly spherically shaped, independently on
the material that surrounds them. This could be due to the fact that their growth rate
is very low, so when they grow into a bony material, the bone is resorbed and the cyst
grows as if there was no obstacle stopping it from expanding. Thus, the stresses on the
cyst, seem to have little angular dependence.
Now, the hydrostatic pressure jump balances the stresses in the semi-permeable mem-
brane, and the stresses on the cyst from the surrounding,

Ph+ − Ph− = f (R, Ṙ) + fb (R, Ṙ),

where f , the physical stresses, is in general, a function of R and the time derivative of R;
and fb corresponds to stresses due to biological processes.
However, it is not clear which biological or physical stresses should be included. For
instance, the fact that in some cases the cells actively divide may affect the stresses
the surrounding material exerts on the cyst: it is not clear whether these stresses are
important, or whether they are proportional to the rate of change of R, or proportional
to the rate of change of the surface area of the cyst. So, in this work we only will consider
the physical stresses of the surrounding material. Also, the equations above allow us to
find a single equation relating the radius of the cyst and the stress function, of the form
1 R ˙
Ṙf + f + C0 Ṙ = β. (2.1)
α 3α
Even with this simplification, we still lack information on the properties of this material.
All we know is that it is a mixture of fluid, collagenous capsule, and crystalline structures.
Thus, we will assume that the material may be modelled as a Maxwell fluid, so the stresses
satisfy
Ṙ
τ f˙ + f = µ , (2.2)
R
where τ is the relaxation time, and µ the viscosity of the external medium.
Considering the scalings R = R0 R, b f = µ f, b t = Tbt, τ = T τb, assuming that T =
T
C0 R0 µβ
, and defining CB = , equations (2.1) and (2.2) become
β αR0 C02
CB ˙
CB Ṙf + Rf + Ṙ = 1, (2.3)
3
and
Ṙ
τ f˙ + f = , (2.4)
R
with initial conditions R(0) = 1, and f (0) = f0 . In equations (2.3) and (2.4), we have
dropped all the hats for the sake of simplicity. It is worth remarking that, using the
osmotic pressure equation, the material concentration in the core C is related to the
dimensionless function f via C = C0 (1+CB f ). The solution of these two coupled ordinary
differential equations may be found analytically in limiting cases of the parameters CB , τ
and f0 , for which some terms in those equations dominate. However, the general solution
can be found numerically only.

2.3.1 Asymptotic analysis

The coupled nonlinear system of ordinary differential equations (2.1) and (2.2) are not
solvable in closed form and in order to understand the model predictions for the evolution
of R and f we resort to asymptotic methods. Of particular interest is the determination
of timescales for the shifting of importance of the main mechanisms involved in cystic
growth. There are three parameters in the model, namely Cb , τ and f0 , for which there
is insufficient data to determine their values. Hence, we seek solutions in the six cases in
which one parameter is set to be large and small, whilst the other two are of O(1) size.
It turns out that in all cases the (very) large time attractor for equations (2.1) and (2.2)
is given by the system

Ṙ 1
Ṙ ∼ 1, f∼ ∼ , (2.5)
R R
implying that R ∼ t and f ∼ 1/t as t → ∞. The model therefore predicts in large time
that the radial expansion of the cyst will be linear and the concentration of the internal
osmotic material will equilibrate with the external concentration. Hence cystic growth
will eventually be governed by cell birth. We also note that in large time the outer fibrous
capsule layer behaves like a viscous material with negligible elastic properties.
The details of the analysis are omitted for brevity and summarised in Table 1 (at the
end of the report). The Table shows for each limit the appropriate rescalings on time for
each of the important timescales (column 2), the rescalings on R and f for each timescale
(columns 3 and 4) and the “large” time behaviour of R and f for each timescale (columns
5 and 6). We note the last line for each limit in the table indicates the timescale and
appropriate rescalings for the large time system (2.5) to become dominant. For example,
in the case of large CB , i.e. CB → ∞, we expect the behaviour given by (2.5) to dominate
in a timescale of t = O(CB ), likewise for the large τ limit, we expect (2.5) to result in a
timesale of t = O(τ ln(τ )). Below is a short discussion of the salient features drawn out
by the analysis for each case.

Large CB The dimensionless parameter CB consist of a number of parameters so there

are a number of combinations of relative sizes that can generate a “big” CB . How-
ever, the value of CB will increase if the material supply rate β and fibrous capsule
“viscosity” µ are increased or the initial cyst diameter R0 , external material con-
centration C0 (thereby enhancing osmotic pressure) and osmotic pressure constant
α are decreased. There are three main timescales. Growth in the first timescale is
dominated by the osmotic pressure difference and the resulting absorption of water
forces expansion; the amount of new material added to the core by cell death is
negligible. There follows two longer timescales in which cell death now contributes
to the amount of material in the core and the elastic properties of the capsule col-
lagen fibres become negligible. Eventually, the balance given by (2.5) is reached in
an O(CB ) timescale.

Small CB As with the large CB case, small CB may result in a number of ways and
its value will decrease by reversing the parameter relationships described in the
large CB case. This case effectively implies that the osmotic pressure difference has
negligible effect on growth and cell birth dominates growth. As a result, the large
time scenario (2.5) occurs in O(1) time.

Large τ This case implies that the fibrous capsule responds to the growing cyst, at least
in the short term, like an elastic layer (the viscous properties being negligible).
There are a number of timescales in this case, although the latter ones are technical
 rather than have any biological significance. Initially, the osmotic pressure remains
constant, the material concentration in the core being held stable due to amount of
new material introduced by cell death is exactly compensated by the influx of water;
consequently linear growth is predicted. Eventually a second, longer timescale
ensues in which the viscous properties of the fibrous capsule come into play; this
has the effect of “taking the slack” with influx of water, hence the internal material
becomes diluted and the osmotic pressure decreases exponentially. The large time
of linear radial expansion evolves in this timescale, i.e. t = O(τ ). There follow two
minor timescales of little interest before evolving to the balance given by (2.5).

Small τ The case of τ → 0 effectively assumes that the fibrous capsule behaves as a
viscous material, i.e. it readily responds to cystic growth by taking the slack rather
than by stretching. Water influx by the osmotic pressure difference contributes
initially to cystic growth, however, the balance given by (2.5) dominates as equili-
bration of material concentration occurs in O(1) time.

Large f0 Large initial hydrostatic/osmotic pressure may arise, for example, if there is a
very high concentration of material in the core of the cyst compared to surrounding
tissue. Growth initially starts very rapidly due to a massive influx of water to
balance the osmotic pressure difference. There follows two long timescales in which
there is apparently negligible cystic growth and the osmotic pressure seems to settle
to some fixed, non-zero level. However, there is a small amount of cystic growth
1/3
and in large time, t = O(f0 ), we get the balance given by (2.5).

Small f0 Here we assume that initially there is a negligible hydrostatic/osmotic pressure

across the cystic epithelial lining, which may arise, for example, when the concen-
tration of material inside and outside the cyst are nearly equal. There are two
phases of interest. Initially, there is a rapid increase in osmotic pressure due to the
water influx being too slow to compensate for the additional material being added
to the core by cell death; in time, this material concentration will build up suffi-
ciently for the osmotic pressure difference to become non-negligible. Consequently,
a full balance, i.e. (2.1) and (2.2), results in O(1) time and soon the balance given
by (2.5) ensues. The very small initial timescale suggests that this case is unlikely
to be observed.

There are of course a number of combinations of these limits that could be considered. The
more interesting cases arise when the parameters are large, for example 1 ≪ CB = O(τ )
and 1 ≪ CB = O(τ 5/3 ) seem to be distinguished limits. However, these cases offer no
further insights and will not be discussed further.

2.3.2 Numerical results

Figures 8 and 9 show log-log plots of the evolution of R and f for CB = τ = f0 = 1. The
dashed lines are the large time solutions given by (2.5) and in both cases these solutions
agree well (although not exactly) with the full solutions from about t = 1. It can be
shown that the approach to these large time solutions is given by

2CB 1 2CB ln(t)

R ∼ t− ln(t), f∼ + ,
3 t 3 t2
as t → ∞, explaining the apparent slowness in convergence of the curves. Figures 10
and 11 show two examples of the asymptotic approximations compared to the numerical
solutions. In both cases good agreement is obtained between the numerical solution
(solid curves) and the asymptotic solutions (various dotted and dashed curves) in the
appropriate timescales. We note that it is the asymptotic solutions shown in the figure
are the full, leading order asymptotic solutions for each timescale, rather than the “large
time” expansions given in the last two columns of Table 1.
1000

100
radius of cyst, R

10

0.1

0.01
0.01 0.1 1 10 100 1000
time, t

Figure 8: Log-log plot showing the evolution of the cyst radius, with CB = τ = f0 = 1.

100

10
physical stress, f

0.1

0.01

0.001
0.01 0.1 1 10 100 1000
time, t

Figure 9: Log-log plot showing the evolution of the osmotic pressure, with CB = τ =
f0 = 1.

2.4 Conclusions
In this report we present a simple mathematical model to describe cystic growth, based on
cell birth (and death) and osmotic-hydrostatic pressure balances. We assumed that the
 100000

10000

1000
radius of cyst, R

100

10

0.1
1e-05 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000
time, t

Figure 10: Log-log plot showing the evolution of the cyst radius, with CB = 1000,
τ = f0 = 1. The solid curve is the numerical solution of (2.1) and (2.2) and the dotted
and dashed curves are the asymptotic approximations.
1

0.1

0.01
physcial stress, f

0.001

0.0001

1e-05

1e-06

1e-07
0.1 1 10 100 1000 10000 100000 1e+06
time, t

Figure 11: Log-log plot showing the evolution of the osmotic pressure difference, with
τ = 1000, CB = f0 = 1. The solid curve is the numerical solution of (2.1) and (2.2) and
the assorted dotted and dashed curves are the asymptotic approximations.

cyst consists of a thin shell of living cells, behaving as a Maxwell fluid (reflecting action of
a fibrous capsule), surrounding a fluid core. Material feeding into the core by cell death
leads to osmotic pressure differences between the cyst core and its surroundings, which,
together with cell birth, drives cystic growth. The dimensionless version of the model
consists of 3 parameters, namely CB , τ and f0 , the values of which cannot be determined
accurately from existing data. Nevertheless, the asymptotic and numerical solutions
described in Sections 3.3 and 3.4 demonstrate that, regardless of the parameters, the
cysts radius will eventually grow linearly with the osmotic pressure differences decaying
to zero. In dimensional terms the long time behaviour is given by
β µ
R ∼ t, f ∼ , (2.6)
C0 t
recalling that β is the supply rate of degraded material, C0 is the external concentration
of the osmotic material and µ is the viscosity of the external material. We note that the
manner and timescales in which this behaviour occurs is very much dependent on the
parameter values.
Cysts when diagnosed and treated can be up to a few centimeters in diameter; in
the context of the model, this would probably represent cysts at a fairly mature stage
of development. However, these cysts still have measurable osmotic pressure differences,
which suggests that the large time growth pattern, given by 2.6, has yet to be achieved.
This has two possible implications. Firstly, at least one of the dimensionless CB , τ and f0
is large, otherwise negligible osmotic pressure differences would be expected. Secondly,
radial growth may not be linear at the time of diagnosis. Nevertheless, despite the model
indicating that, in the long term, osmotic pressure differences do not influence growth,
it does suggest that it plays an important role in the early and intermediate stages of
cyst development. From this we can conclude that when a cyst is diagnosed, the osmotic
pressure differences are still an important factor.
The model should be applicable for any kind of cyst if an osmotic pressure drop is
driving the growing process, but only if the growth rate is very slow. For the particular
application which led to this study, we predict that dental cysts will grow indefinitely,
unless we include other physical or biological processes that may affect cystic growth.
For instance, here the supply rate β implicitly includes any mitosis or lining loss effects;
however, basal cells on a Keratocyst surface are four times more densely packed than
the cells lining a radicular cyst, and thus it is not clear whether mitotic cells will be the
source of important stresses on a Keratocyst surface. These stresses probably ought to
be included in the force balance equation as biological stresses, such as a mitotic stress
function (proportional to the rate of change of either, the cyst radius, fm ∝ Ṙ, or the
cyst surface area, that is fm ∝ RṘ), a cell lining loss stress function, and so on.
Also, physical stresses, of the surroundings on the cyst, clearly will be smaller if the
cyst is growing in soft tissues rather than in bone, unless the process is slow enough for
the bone to be resorbed by the body at a similar (or larger) rate than that of the cyst
growth, so that the cyst is not deformed from its spherical shape. If this were not true,
the stresses would have spatial dependence on the angular direction, and the cyst would
not be spherically shaped; in that case, the model would have to be modified accordingly.
However, experimental observations support our assumptions in this respect.
Finally, it is difficult to assess whether the concentration of material inside the cyst is
uniform or not, and it would be important to carry out experiments to test this hypothesis,
and to improve the model. It may be necessary, as well, to divide the material inside
the cavity into more subunits so as to consider biological-chemical processes particular to
certain species in more detail. For instance, both the osmotically active substance, and
the water drawn into the cavity, are fundamental in this problem and may be separated
from the rest of the cavity contents. In fact, if the osmotically active substance could
diffuse out of the epithelial layer, the flux of water into the cell would decrease, and so
would the growth rate of the cyst; this proposed mechanism could even stop the growing
process completely. Also, the role in cystic growth of other substances in the cavity, such
as albumin, or cholesterol crystals, should be defined more clearly in future experiments
and models.

Acknowledgements
The authors would like to thank, in no particular order, John Chapman, John King,
Colin Please, Iain Barrass, Michael Simmonds, Aisling Metcalfe, Paul O’Shea, Matthias
Ngwa, Helen Byrne, Chris Breward and anyone else who poked their noses into the “cysts
suite” for their mathematical input.

References
[1] Toller, P.A. (1970) The osmolality of fluids from cysts of the jaw. Brit. Dent. J., 129, 275-278.
[2] Soames, J.V. and Southam, J.C. (1998) Oral Pathology, 2nd. Edition. Oxford Univ. Press, Oxford.
[3] Shear, M. (1992) Cysts of the Oral Regions. 3rd. Edition. Wright, Oxford.
View publication stats

scalings t(i) → ∞ behaviour

(i)
limit t = R = f = R ∼ f (i) ∼
1/3
CB → ∞ t(1) R(1) f (1) (3f0 )1/3 t(1) 1/3t(1)
3/5 (2) 1/5 (2) −3/5 (2) 3 (2) 2
CB t CB R CB f 10 t 2/t(2)
CB t(3) CB R(3) CB−1 (3)
f t(3) 1/t(3)
CB → 0 t(1) R(1) f (1)
t(1) 1/t(1)
τ →∞ t(1) R(1) f (1) t(1) /(1 + f0 CB ) f0 − f0 t(1) /τ
(2)
τ t(2) τ R(2) f (2) t(2) f0 e−t
τ (ln(τ ) + ln(ln(τ )) + t(3) ) τ (ln(τ ) + ln(ln(τ )) + R(3) ) f (3) /τ ln(τ ) t(3) 1 − ln(ln(τ )) (3)
ln(τ ) − t / ln(τ )
ln(ln(τ )) (4)
τ (ln(τ ) + ln(ln(τ ))(1 + t(4) )) τ (ln(τ ) + ln(ln(τ ))(1 + R(4) )) f (4) /τ ln(τ ) t(4) 1− ln(τ ) t
τ ln(τ )(1 + t(5) ) τ ln(τ )(1 + R(5) ) f (5) /τ ln(τ ) t(5) 1/t(5)
τ →0 t(1) R(1) f (1) t(1) 1/t(1)
t(1) /3τ (1)
/τ
f0 = 1/ε → ∞ t(1) R(1) ε−1 f (1)  e  e−t
Zτ (2)
1/3 CB −ZCB /3 −Zt(2)
τ ln(1/ε) + t(2) ε−1/3 R(2) f (2) CB 1 − 3 e−Zt CB e
12τ +CB 1/3 −ZC /3 (2)
1/3 (3) 1/3 (3)
ln(1/ε) + t(3) ε R (3)
ε f CB +ε t CB + CB B e−Zt
−1/3 −1/3
9
12τ +CB
9 ln(1/ε) + ε−1/3 t(4) ε −1/3 (4)
R ε1/3 f (4) t(4) 1/t(4)
(1)
f0 = ε → 0 εt R(1) εf (1) 1 + εZt(1) 1 + Zt(1)
t(2) R(2) f (2) t(2) 1/t(2)

Table 1: Summary of the asymptotic analysis on equations (2.1) and (2.2) in the limits shown in the first column (the other parameters
being O(1)). The Table shows for each limit the appropriate rescalings on t, R and f for the various timescales and the large time
behaviour of R and f in their respective timescale. Where the leading order solutions are constant, the correction term(s) is(are) also
given. The constant Z is given by Z = 3/(3τ + CB ).