J. theor. Biol.

(1996) 178, 381–385

Expansion of Mutant Stem Cell Populations in the Human Colon

M B†

Department of Anatomy and Cell Biology, Medical Sciences Building, University of Toronto,
Toronto, Ontario M5S 1A8, Canada

(Received on 14 July 1995, Accepted in revised form on 18 September 1995)

In general, it is presumed that colonic epithelial stem cells are the principal cell type at risk of incurring
the series of somatic mutations leading to carcinoma, since all other epithelial cell types are short-lived.
Mutant stem cell clonal expansion increases the risk for subsequent mutations and is therefore a
potentially important step in carcinogenesis. The stem cells reside in colonic crypts, simple tubular
foldings of the epithelium, and thus counting crypts provides an indirect means to determine stem cell
numbers. The normal crypt population is known to expand through a process of crypt replication and
this is thought to result in a corresponding expansion of the epithelial stem cell population. A simple
mathematical model of the population dynamics of normal and mutant crypts (crypts containing mutant
stem cells) is developed and used to estimate a lower bound on the relative rate of expansion of the
mutant stem cell population. The model predicts that if mutant and normal crypt populations expand
at the same rate, and if the mutation rate is small relative to the rate of growth, then the fraction of
clusters of mutant crypts composed of only a single mutant crypt should steadily decrease with age
towards one-half. Aberrant crypts are easily recognizable lesions in human colon which have frequently
been shown to contain cells with K-ras and occasionally APC gene mutations. Application of the model
to recent counts of aberrant crypt cluster sizes indicate that the aberrant crypt population, and the
contained mutant stem cell population, is expanding substantially faster than normal.
7 1996 Academic Press Limited

Introduction increased mutation risk. Simple methods capable

There is strong evidence that many colorectal
carcinomas result from accumulated somatic mu-
tations, presumably starting with the mutation of
an epithelial progenitor or stem cell (Fearon &
Vogelstein, 1990; Nishisho et al., 1991; Gordon et al.,
1991; Powell et al., 1991; Vogelstein & Kinzler, 1993).
Expansion of the clones derived from such mutant
stem cells is thought to be an important stage
in carcinogenesis for at least two reasons: (i)
the increasing numbers of mutant cells directly
increases the probability that at least one of these
cells will receive the additional mutations leading
to malignant transformation; and (ii) clonal expan-
sion entails DNA replication with its associated
† E-mail: bjerknes.crypt.med.utoronto.ca.
of demonstrating abnormalities in the relative rate
of expansion of clones of mutant stem cells in man
are therefore of interest, and will require techniques
both for the identification of mutant clones and
also for the extraction of useful information from
identified clones. It is the latter that will be dealt with
here.
A careful scan of the colonic epithelium will
occasionally reveal crypts of abnormal morphology
and histochemistry; these are the aberrant crypts
(Bird, 1987; McLellan & Bird, 1988a,b; Roncucci
et al., 1991a, b; Pretlow et al., 1994). They are often
found in clusters thought to represent clones derived
from a mutant stem cell. In fact, many aberrant crypts
contain cells with K-ras or APC mutations (Stopera
et al., 1992; Pretlow et al., 1993; Vivona et al., 1993;
Smith et al., 1994; Jen et al., 1994; Yamashita et al.,

0022–5193/96/040381 + 05 $18.00/0 381 7 1996 Academic Press Limited

382 . 
1995). Although their role in human carcinogenesis
remains contentious (Jen et al., 1994; Yamashita
et al., 1995; Pretlow, 1995), aberrant crypts are none-
theless a good test case for the study of the expansion
of a population of readily identified mutant clones.
When first formed, aberrant crypts are single
isolated structures, but with time they branch
producing clusters of related aberrant crypts (McLel-
lan et al., 1991). This description of the growth of
clusters of aberrant crypts is reminiscent of the crypt
cycle, a continuous crypt replication process which
has been most thoroughly studied in the normal
mouse small intestinal epithelium (Bjerknes, 1986,
1994, 1995; Cheng & Bjerknes, 1985; Totafurno et al.,
1987), but which also seems to occur in human colon
(Cheng et al., 1986). In normal intestine, the crypt
cycle results in the slow expansion of the crypt
population through a process of crypt budding and
fission. It appears that newly formed aberrant crypts
expand their numbers through a similar process.
Thus, aberrant crypts arise from two sources; from
spontaneous somatic mutation of a stem cell in a
normal crypt, or from a crypt-cycle-like replication of
existing aberrant crypts. The latter mechanism has the
potential to expand the mutant stem cell pool rapidly
and thus may warrant serious concern.
It is a simple matter to write equations describing
the expansion of normal and mutant crypt popu-
lations and the conversion of normal into mutant
crypts. In this work I demonstrate that if mutant and
normal crypt populations expand at the same rate,
then these equations predict that with age, about
one-half of mutant crypt clusters should contain
only a single crypt. If fewer than one-half are
actually found to be unicryptal, then this is evidence
that the mutant crypt population is expanding faster
than normal. These ideas are applied to the analysis
of recently published data on the distribution of the
size of human aberrant crypt clones, where it was
found that fewer than one-half of clusters of
aberrant crypts were unicryptal; most were poly-
cryptal (Roncucci et al., 1991a,b; Pretlow et al.,
1994). It is concluded that the aberrant crypt
population, and hence the mutant stem cell popu-
lation contained in them, is expanding substantially
faster than normal.
A Mathematical Model of Aberrant Crypt Dynamics
Our goal is to determine if the population of
mutant stem cells found in aberrant crypts is growing
faster than the stem cell population found in normal
crypts. Since the rate of growth of a stem cell pool
is reflected in the rate of growth of the population of
crypts containing those stem cell (Bjerknes, 1986,
1994, 1995; Cheng & Bjerknes, 1985; Totafurno
et al., 1987), we will compare the expansion rates of
the populations of aberrant vs. normal crypts in
order to explore the relative expansion rates of the
populations of mutant vs. normal stem cells. We will
assume that the normal crypt population expands at
rate k1 proportional to the current population size,
and that l is their rate of conversion into aberrant
crypts (presumably through spontaneous somatic
mutations in a contained stem cell; Bjerknes, 1995).
We also assume that single isolated aberrant crypts
give rise to larger clusters of aberrant crypts at a
constant rate k2 , again in proportion to current
population size. Note that k1 and k2 may be
interpreted either as the respective rates of replication
of normal and mutant crypts, or in the presence of
significant crypt loss as the respective net population
expansion rates. Also note that we will ignore the
presumably negligible probability that stem cells in
adjacent normal crypts would undergo the same
mutation thus producing a mutant cluster without
crypt replication. Under these assumptions, the
growth of the populations of normal and aberrant
crypts at time t may be approximated by the system
of differential equations
N '(t) = (k1 − l)N(t) (1)
S '(t) = lN(t) − k2 S(t) (2)
C '(t) = k2 S(t), (3)
where N(t) is the number of normal crypts, S(t) is the
number of single aberrant crypts, and C(t) is the
number of clusters containing two or more aberrant
crypts. These equations have solutions [assuming
N(0) = N0 and S(0) = C(0) = 0]
N(t) = N0 eat (4)
lN0 at
S(t) = (e − e−k2 t ) (5)
b
lN0
C(t) = (k eat + a e−k2 t − b), (6)
ab 2
where a = k1 − l and b = k1 + k2 − l. Here, all
aberrant crypts were treated as though they behave
identically. This should be an adequate first
approximation, but as markers are developed which
allow subpopulations of aberrant crypts to be
identified (for example those containing different
mutations), this system of equations could easily be
expanded by addition of variables representing the
singleton and larger clusters of each of the subtypes,
each with their own mutation and replication rates
(Bjerknes, 1995).
      
The total number of aberrant crypt clusters will
grow as
T(t) = S(t) + C(t) (7)
lN0 at
= (e − 1), (8)
a
and, therefore, the fraction of aberrant crypt clusters
which should be found as singletons is
S(t) a eat − e−k2 t
s(t) 0 = . (9)
T(t) b eat − 1
There are two observations about this equation
which will lead us to our final result. First, with
time, the fraction of single aberrant crypts tends to the
limit
s 0 lim s(t) (10)
t4a
= a/b (11)
= (k1 − l)/(k1 + k2 − l) (12)
1 k1 /(k1 + k2 ), (13)
where eqn (13) follows from the additional, and
probably reasonable, assumption that the spon-
taneous mutation rate, l, is small relative to the crypt
population expansion rates k1 and k2 (Bjerknes, 1995).
Second, is the observation that the fraction of single
aberrant crypts, s(t) is a decreasing function of time
(see Appendix).
Therefore, the fraction of aberrant crypt clusters
containing a single crypt starts at 1 (at t = 0) and
decreases steadily towards the asymptotic value
k1 /(k1 + k2 ). Equation (13) may be rewritten as
s = 1/(1 + r), (14)
where r = k2 /k1 . This tells us that if the rates of
expansion of normal and aberrant crypts are equal
(r = 1) then asymptotically one-half of aberrant
crypts should be single. Since measurements of
aberrant crypt clusters in human colon revealed that
far fewer than one-half of the clusters were composed
of single crypts (Roncucci et al., 1991a,b; Pretlow
et al., 1994), we may safely conclude that the aberrant
crypt population is growing faster than is the
population of normal crypts, and that in turn the
mutant stem cell population contained in the aberrant
crypts expands at an increased rate. This conclusion
holds even though the samples may not have been
obtained at asymptotic ages because s(t) is steadily
decreasing in time.
383
Discussion
Since fewer than one-half of aberrant crypt clusters
are single (Roncucci et al., 1991a,b; Pretlow et al.,
1994), we concluded that the aberrant crypt
population is expanding at an increased rate in
comparison with normal crypts. We may estimate a
lower bound on the magnitude of the increase from
recent data. Pretlow et al. (1994; their fig. 4) found
in a study of 15 patients that only one of 42 clusters
of aberrant crypts were single. Thus, an estimate of
an upper bound on the asymptotic fraction of clusters
of single aberrant crypts is s E 1/42. This is an
inequality because s(t) is a decreasing function of
time, and we do not know whether the data were
collected from individuals whose crypt dynamics had
reached the asymptotic state. Rearranging eqn (14)
we obtain
1
r= −1 (15)
s
1
e −1 (16)
1/42
= 41. (17)
Thus, given this admittedly rough estimate of s,
the mutant stem cell pool contained in the aberrant
crypt population is growing q40 times faster than
normal. This may be an overestimate if single
aberrant crypts are more easily missed during scoring
than are large clusters, and thus, the reader should be
cautioned that the data analysed here was not
collected with our purpose in mind. Vagaries of the
experimental methods used may have biased the
distribution of cluster-sizes obtained; an issue which
future measurements of cluster-size distributions must
address.
It is important to stress that the conclusion that
the mutant stem cell pool is growing q40 times
faster than the normal stem cell pool does not
necessarily mean that the mutant stem cells are
cycling 40 times faster than normal stem cells, a feat
which might not even be possible. Rather, the
magnitude of the increase in the rate of growth of the
mutant stem cell pool suggests that increased
retention of stem cell offspring, which normally would
be lost through cell differentiation or death, could be
a major factor in the increased rate of expansion.
Recruitment into the cell cycle of non-cycling mutant
stem cells, if such exist, offers an additional source.
However, increased proliferation probably does make
a contribution because aberrant crypts are known to
have proliferative abnormalities (Roncucci et al.,
1993; Pretlow et al., 1994; Yamashita et al., 1994), but
384 . 
it is unclear from these previous studies whether the
stem cells are affected.
Crypt loss is also a potential concern, but it should
be recalled that the model parameters k1 and k2 may
be interpreted as representing the net expansion rates
of normal and mutant crypts, respectively. Further-
more, if significant loss of aberrant crypts occurs, say
as a result of immune surveillance, then the mutant
stem cells must be replicating at an even higher
rate than is indicated above, in order to counter the
loss.
Another issue, about which little is known, is the
actual stem cell content of normal and mutant crypts.
There are strong theoretical reasons for suspecting
that the stem cell content of normal crypts is variable
(Bjerknes, 1986, 1994, 1995; Totafurno et al., 1987),
but nothing is known about possible effects of
mutations on either the variability or the average stem
cell content. We have largely avoided this issue here
by focusing on the relative rates of expansion, rather
than the absolute sizes of the mutant and normal stem
cell populations.
It is also important to stress that the somatic
mutations afflicting aberrant crypts need not necess-
arily affect the stem cells directly in order to yield an
increased growth rate. This could result, for example,
if the mutation affected cell adhesion in a way which
stimulated crypt replication (Augenlicht, 1994).
Stimulation of the replication process would, in turn,
stimulate growth of the stem cell population.
Regardless of whether the mutation impacts directly
or indirectly on stem cell growth rates, the mutant
stem cell pool in aberrant crypts does appear to be
expanding at an abnormal rate, a potentially
dangerous situation.
Much effort is being spent in the search for dietary
factors which might alter the rate of progression
towards tumors (Bruce et al., 1993; Magnuson et al.,
1993; Kawamori et al., 1994; Sutherland & Bird,
1994; Takahashi et al., 1994; Thorup et al., 1994)
and thus, there is considerable interest in the design
of simple methods of analysis of the results from
these trials. The application of variations of the
analysis introduced here may offer another useful
approach.
The approach introduced above is not limited
only to the study of the rate of expansion of
aberrant crypts. The ideas may be applicable to the
study of any mutation for which markers are
developed allowing identification of the afflicted
crypts.
This work was supported by a grant from the MRC of Canada.
REFERENCES
A, L. H. (1994). Adhesion molecules, cellular differen-
tiation, and colonic crypt architecture. Gastroenterology 107,
1894–1898.
B, R. P. (1987). Observation and quantification of aberrant
crypts in the murine colon treated with a colon carcinogen:
preliminary findings. Cancer Lett. 37, 147–151.
B, M. (1986). A test of the stochastic theory of stem cell
differentiation. Biophys. J. 49, 1223–1227.
B, M. (1994). Simple stochastic theory of stem-cell
differentiation is not simultaneously consistent with crypt
extinction probability and the expansion of mutated clones.
J. theor. Biol. 168, 349–365.
B, M. (1995). The crypt cycle and the asymptotic dynamics
of clusters of mutant crypts in the intestinal epithelium. Proc.
Roy. Soc. Lond. B 260, 1–16.
B, W. R., A, M. C., C, D. E., M, A.,
M, S., S, D., et al. (1993). Diet, aberrant crypt foci and
colorectal cancer. Mutat. Res. 290, 111–118.
C, H. & B, M. (1985). Whole population cell kinetics
and postnatal development of the mouse intestinal epithelium.
Anat. Rec. 211, 420–426.
C, H., B, M., A, J. & G, G. (1986). Crypt
production in normal and diseased human colonic epithelium.
Anat. Rec. 216, 44–48.
F, E. R. & V, B. (1990). A genetic model for
colorectal tumorigenesis. Cell 61, 759–767.
G, J., T, A., S, W., C, M., G,
L., A, H. et al. (1991). Identification and characteriz-
ation of the familial adenomatous polyposis coli gene. Cell 66,
589–600.
J, J., P, S. M., P, N., S, K. J., H,
S. R., V, B. & K, K. W. (1994). Molecular
determinants of dysplasia in colorectal lesions. Cancer Res. 54,
5523–5526.
K, T., T, T., K, T., S, M., O, M. &
M, H. (1994). Suppression of azoxymethane-induced rat
colon aberrant crypt foci by dietary protocatechuic acid. Jpn.
J. Cancer Res. 85, 686–691.
M, B. A., C, I. & B, R. P. (1993). Ability of aberrant
crypt foci characteristics to predict colonic tumor incidence in
rats fed cholic acid. Cancer Res. 53, 4499–4504.
ML, E. A. & B, R. P. (1988). Specificity study to evaluate
induction of aberrant crypts in murine colons. Cancer Res. 48,
6183–6186.
ML, E. A., & B, R. P. (1988). Aberrant crypts: potential
preneoplastic lesions in the murine colon. Cancer Res. 48,
6187–6192.
ML, E. A., M, A. & B, R. P. (1991). Sequential
analysis of the growth and morphological characteristics of
aberrant crypt foci: putative preneoplastic lesions. Cancer Res.
51, 5270–5274.
N, I., N, Y., M, Y., M, Y., A,
H., H, A. et al. (1991). Mutations of chromosome 5q21 genes
in FAP and colorectal cancer patients. Science 253, 665–669.
P, S. M., Z, N., B-B, Y., B, T. M.,
H, S. R., T, S. N., V, B. & K,
K. W. (1991). APC mutations occur early during colorectal
tumorigenesis. Nature 359, 235–237.
P, T. P. (1995). Aberrant crypt foci and K-ras mutations:
earliest recognized players or innocent bystanders in colon
carcinogenesis? Gastroenterology 108, 600–603.
P, T. P., B, T. A., F, C., C, C. &
K, E. L. (1993). K-ras mutations in putative preneoplastic
lesions in human colon. J. Natl. Cancer Inst. 85, 2004–2007.
P, T. P., C, C. & O’R, M. A. (1994). Aberrant
crypt foci and colon tumors in F344 rats have similar increases
in proliferative activity. Int. J. Cancer 56, 599–602.
P, T. P., R, E. V., O’R, M. A., C, J. C.,
A, S. B. & S, T. A. (1994). Carcinoembryonic antigen
       385

in human colonic aberrant crypt foci. Gastroenterology 107, of p53 protein accumulation in aberrant crypt foci of the colon.
1719–1725. Gastroenterology 108, 434–440.
R, L., M, A. & B, W. R. (1991a). Classification Y, N., M, T., O, M. & E, H. (1994).
of aberrant crypt foci and microadenomas in human colon. Increased cell proliferation of azoxymethane-induced aberrant
Cancer Epidemiol. Biomarkers Prev. 1, 57–60. crypt foci of rat colon. Jpn. J. Cancer Res. 85, 692–698.
R, L., P, M., F, R., D G, C. & P 
L, M. (1993). Cell kinetic evaluation of human colonic
aberrant crypts. Cancer Res. 53, 3726–3729.
R, L., S, D., M, A., C, J. B. & B, APPENDIX
W. R. (1991b). Identification and quantification of aberrant
crypt foci and microadenomas in the human colon. Hum. Path. In order to show that s(t) = S(t)/T(t) is a
22, 287–294. decreasing function of t is suffices to show that its
S, A. J., S, H. S., P, M., H, K., M, A., B,
B. V. & G, S. (1994). Somatic APC and K-ras codon 12
derivative is negative for t q 0. But
mutations in aberrant crypt foci from human colons. Cancer Res.
54, 5527–5530. s'(t) = [S '(t)T(t) − S(t)T '(t)]/T(t)2 (A.1)
S, S. A., M, L. C. & B, R. P. (1992). Evidence for
a ras gene mutation in azoxymethane-induced colonic aberrant = a(b eat e−k2 t − a eat − k2 e−k2 t )/b(eat − 1)2 .
crypts in Sprague-Dawley rats: earliest recognizable precursor
lesions of experimental colon cancer. Carcinogenesis 13,
2081–2085. (A.2)
S, L. A. M., & B, R. P. (1994). The effect of
chenodeoxycholic acid on the development of aberrant crypt foci Then, since 0 Q a Q b, it suffices to show that
in the rat colon. Cancer Lett. 76, 101–107.
T, M., M, T., Y, M., K, T., b eat e−k2 t − a eat − k2 e−k2 t Q 0, which in turn is equiv-
Y, K. & E, H. (1994). Effect of docosahexaenoic acid alent to showing that b eat − a ebt − k2 Q 0 (multiply
on azoxymethane-induced colon carcinogenesis in rats. Cancer through by ek2 t q 0. Expanding the exponentials and
Lett. 83, 177–184.
T, I., M, O. & K, E. (1994). Influence of a combining terms yields
dietary fiber on development of dimethylhydrazine-induced
aberrant crypt foci and colon tumor incidence in Wistar rats. b eat − a ebt − k2 = ab[(a − b)t 2/2! + (a 2 − b 2 )t 3/3!
Nutr. Cancer 21, 177–182.
T, J., B, M. & C, H. (1987). The crypt cycle. (A.3)
Evidence for crypt and villus production in the adult mouse small
intestinal epithelium. Biophys. J. 52, 279–294.
V, A. A., S, B., M, A., B, W. R., H, K., + ···+(a m − 1 − b m − 1 )t m/m!+···].
W, M. A., S, H. S. & G, S. (1993). K-ras
mutations in aberrant crypt foci, adenomas and adenocar- (A.4)
cinomas during azoxymethane-induced colon carcinogenesis.
Carcinogenesis 14, 1777–1781. But this is negative since 0 Q a Q b implies that
V, B. & K, K. W. (1993). The multistep nature of (a m − b m ) Q 0 for all m e 1. To summarize, we have
cancer. Trends Genet. 9, 138–141.
Y, N., M, T., O, A., O, M. & E, H. shown that s'(t) Q 0 which in turn implies that s(t)
(1995). Frequent and characteristic K-ras activation and absence is a decreasing function of time.