Human Studies of Calcium Supplementation and Colorectal ... · ofcalcium consumed is absorbed (3),...
Transcript of Human Studies of Calcium Supplementation and Colorectal ... · ofcalcium consumed is absorbed (3),...
Vol. 6. 971-980. November 1997 Cancer Epidemiology, Biomarkers & Prevention 97/
Review
Human Studies of Calcium Supplementation and Colorectal Epithelial
Cell Proliferation
Roberd M. Bostick’
Department of Public Health Sciences, Epidemiology Section, The Bowman
Gray School of Medicine, Wake Forest University. Winston-Salem, North
Carolina 27157
Abstract
The kinetics of coborectal epithelial cell proliferation(CECP) have been found to be altered in patients atincreased risk for colon cancer. Altered CECP kineticsinclude an increase within the colon crypts of the overallproportion of proliferating cells (labeling index; LI) andthe proportion of proliferating cells that are in the upper40% of the crypts (4h). Use of CECP as a biomarker tomeasure effects of calcium interventions on the colon hasbeen reported in five small uncontrolled clinical trials,nine small randomized placebo-controlled trials, andthree full-scale randomized placebo-controlled trials. Allfive uncontrolled trials indicated substantial andsignificant decreases in proliferation. Of the nine small
controlled trials, three found statistically significantdecreases in the LI, and the remainder were inconclusivebecause of insufficient sample size. Of the three full-scaletrials, one found a decrease in, or normalization of, the
4�h but no effect on the LI; a second found an increase inthe 4�h but no effect on the LI; and the third, which didnot measure the 4�h� also found no effect on the LI.Differences between the two full-scale trials thatmeasured the 4�, were that the negative trial wasmulticentered, used multiple types of bowel-cleansingpreparations, had no baseline measurements of CECP,and had low intrareader reliability for CECP scoring.The positive trial was single centered, used no bowel-
cleansing preparations, measured CECP prior to, and atthree precise intervals after, randomization, and had highintrareader reliability for CECP scoring. The currentliterature indicates that in humans, it is unlikely thatcalcium supplementation can substantially lower CECPrates, but it may normalize the distribution ofproliferating cells within colon crypts.
Introduction
Colon cancer is the second leading cause of cancer deaths inWestern countries and one of the leading causes of cancerdeaths in the world as a whole (1 ). World-wide, average cab-
Received 3/4/97; revised 8/ l/97; accepted 8/5/97.
The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement in
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
I To whom requests for reprints should be addressed, at Department of Public
Health Sciences. Epidemiology Section. The Bowman Gray School of Medicine.Wake Forest University. Medical Center Boulevard, Winston-Salem, NC 27157.
cium intake in modern times has been far lower than estimatedfor that in Paleobithic times when Honio sapiens evolved (wild
plant foods were a high source of calcium, thus providing
1500-2000 mg daily compared to the current American aver-age intake of740 mg; Ref. 2). Considering that only about 30%
ofcalcium consumed is absorbed (3), it is plausible that humans
evolved in an environment in which there was a generally highlevel of calcium in the colon. It is reasonable to speculate that
our modern dietary habits with low calcium intake and lowcalcium bevels in our enteral environments are having modernadverse consequences for cobonic diseases. Indeed. it is biobog-
icably plausible that a higher intake of calcium may preventcolon cancer. If calcium intake is in the range estimated for that
during the Pabeobithic period. then, unlike with the average
amounts consumed currently. there is enough calcium in thecolon left over after absorptive needs and binding with phos-
phate to bathe the colon mucosa with free calcium and to bindwith bile acids (4). These events may be beneficial because (a)
in cell culture, calcium slows cell proliferation and speeds up
cell differentiation, both thought to reduce the risk of cancer (5,
6), and (b) calcium binds bile acids, which have been shown todamage DNA (7) and, in animals, to have toxic effects on colon
epithelial cells, resulting in compensatory cobonic epithelial cellproliferation (8) and promotion of tumorigenesis (9. 10).
Thus, it is biologically plausible that raising calcium in-takes to 1500-2000 mg daily may reduce the risk of coloncancer in humans and that it may reduce risk by mechanisms
involving normalization of rates of cell proliferation and dif-ferentiation in the colon epithelium. Indeed, calcium has been
shown consistently to substantially normalize colorectal epithe-hal cell proliferation (I 1-13) and to reduce colon tumorigenesis
in rodents ( 14-20). However, studies in rodents may or may not
be of direct relevance to humans. In humans. ecological epi-demiobogicab studies show a correlation between per capita
calcium consumption and rates of colon cancer mortality ( 1):
however, these studies are subject to the ecological fallacy. Asreviewed elsewhere ( 1 , 2 1 ), the analytic observational epide-miobogicab literature on the association of calcium and colon
cancer is somewhat inconsistent, but inverse associations havemore frequently been found. However, several factors make it
difficult to draw strong conclusions from these studies. In
general. observational studies are hampered by the relativehomogeneity of diets within populations, the multitude of die-
tary variables, the problem of unmeasured nutrient-nutrientinteractions, and the back of precision in current dietary meas-
ures used in barge population studies. Although the majority ofthe published studies reported associations suggesting de-
creased risk of colon cancer with relatively high intakes ofcalcium, reporting and publication biases cannot be ruled out
(i.e., investigators finding no association may not have submit-ted their findings for publication or they may have been unable
to have them accepted for publicaton because ofjournal editors’back of interest in negative results). Residual confounding in the
reviewed studies also cannot be ruled out. Although the epide-
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
972 Review: Calcium and Colon Cell Proliferation
miologicab appearance of colon cancer is different from that of
rectal cancer, many of these studies combined the two forreported analyses, possibly attenuating observed associations.
Furthermore, few of the studies reported taking into accountsupplement intake, a factor that is especially important giventhe homogeneity of diets within most populations. Most of thestudies were reported before the emergence of strong theoret-ical and experimental support for calcium as a potential pro-tective factor against colon cancer and did not focus specificallyon calcium intake; consequently, they did not fully address all
confounding and interaction issues related to current hypothe-
ses regarding this dietary constituent.Thus, although there is animal and epidemiobogical sup-
port for the hypothesis that a higher calcium consumption maycontribute to at least a modest reduction in the risk of colon
cancer, clinical trials will be needed to ultimately fully addressthe hypothesis. However, at this time, the evidence is not strong
enough to support such an expensive undertaking. This is acommon predicament for many hypotheses of cancer etiology
and primary prevention. A recent trend in thought to understandcarcinogenesis is to view it as a mubtistep process occurringover a relatively long period of time (22). Such a view implies
that there may be opportunity for measuring early steps in thecarcinogenic process (e.g., the well-established late “interme-
diate end point” for colon cancer, adenomatous pobyps) andusing their measurements as “intermediate end points” in in-tervention trials. Intermediate end points selected for thesetrials generally should have the properties of having been
validated as true intermediate end points, being more common
than the cancer, being likely to be reversible with the interven-tion, and having been shown to be easily and reliably measured.When an intermediate end point can be treated successfully orreversed in randomized controlled trials, then further support is
provided for whether or not a trial using the same interventionbut with the cancer itself as the end point is warranted. Cancerprimary prevention research using intermediate end points hasbeen most extensively put to use in colon cancer investigations.These investigations have included studies to validate whethervarious measures of coborectal epithelial cell proliferation ki-
netics are valid and reliably measured intermediate end pointsfor coborectal cancer and whether or not they can be modulated
in clinical trials, most of which have used calcium as the tested
intervention.Herein, after a brief description of the basic measurements
of colorectab epithelial cell proliferation kinetics and a briefreview of the evidence of whether colorectab epithelial cellproliferation kinetics are valid intermediate end points for cob-rectal cancer, a full review of human studies of calcium sup-
plementation interventions on colorectal epithelial cell prolif-eration is provided with the aim of assessing whether this lineof study has been explored adequately, and if so, whether itsresults provide the support required forjustifying a calcium andcolon cancer intervention trial.
Basic Definitions of Colorectab Epithelial CellProliferation Kinetics
There are two basic measurements of colorectab epithelial cellproliferation kinetics, one to indicate the rate of proliferation of
colon crypt epithelial cells and the second to indicate thedistribution of proliferating cells within the colon crypts. Themost commonly used methods to indicate these aspects ofcoborectal epithelial cell proliferation kinetics in humans usemeans of labeling proliferating cells and counting the labeled
and unlabeled cells in several colon crypts either in histological
cross-sections or in wet mounts of microdissected whole crypts.From these procedures, the resulting indicator of rate is termedthe LI,2 defined as the proportion of cells in the crypts that arelabeled (23). The most validated measurement of distribution isone that indicates a deviance from the proliferative zone of thecrypt being confined to the lower (basal) 60% of the crypt to
one that extends into the upper (luminal) 40% of the crypt. Thismeasurement has been termed the 4� and is defined as theproportion of labeled cells in the crypt that are in the upper 40%
of the crypt (23).
The most commonly used laboratory methods of meas-
uring colorectal epithelial cell proliferation have been the[3H]dThd method (23), the BrdUrd method (24), the PCNAmethod (25), and the CCPR method (26). In the [3H]dThd and
BrdUrd methods, biopsies are obtained and then incubated in
cell culture media for a specified length of time (e.g., 1 h), sothat cells in S-phase will incorporate the primary labeling agent
(i.e., the [3H]dThd or the BrdUrd). Following histological sec-tioning of the biopsies, the cells containing the labeling agent
(the proliferating cells) are identified as labeled by microauto-radiography in the [3H]dThd method and by immunohisto-chemical methods for the BrdUrd method. In the PCNAmethod, no incubation and incorporation of a labeling agent arerequired; rather, immunohistochemical methods are used to
identify an endogenous nuclear protein associated with the S
phase of the cell cycle. In the CCPR method, whole crypts aremicrodissected out, incubated in media to which vincristine isadded to arrest cells undergoing mitosis in metaphase, and then
stained with Feubgen’s stain to identify the cells arrested inmitosis. Most validation work in humans has been done with
the [3H]dThd method, some has been done with the BrdUrdmethod, a little with the PCNA method, and virtually none withthe CCPR method. The [3H]dThd and BrdUrd methods suffer
from variability of uptake of the labeling agent (27), and the
CCPR method suffers from not providing information on dis-tribution of proliferating cells and from not being validated inhumans. The PCNA method has been criticized for being moredifficult to count, because the method yields cells that are
labeled to varying intensity; however, (a) methods in which thelabeled cells are distinguished as darkly labeled (corresponding
to the S-phase cells) or lightly labeled (peri-S-phase cells) havefound that the dark cell count correlates very highly with, and
parallels, the all-labeled (dark plus bight) cell count; and (b) the
cell proliferation scoring reliability reported for this method has
actually been higher than that for any other method (25).
Coborectal Epitheliab Cell Proliferation as an IntermediateEnd Point for the Study of Coborectal Cancer
Several studies (28-44) have reported that, compared to pa-tients at low risk for colon cancer, patients with colon cancer (4,28-37) and patients in every category known to be at higher
risk for colon cancer [those with a history of sporadic adenoma(28-3 1 , 33-37), familial polyposis (32, 38), or ulcerative colitis
(28, 39, 40); those with a family history ofcobon cancer (32, 33,41); and the elderly (39, 42)], on average, exhibit in their
normal-appearing mucosa both an increased cobonic epithebial
cell proliferation rate and an extension of the colon crypt
proliferative zone from the lower (basal) 60% of the crypt to
2 The abbreviations used are: LI, labeling index; I3HldThd, tritiated thymidine;
BrdUrd, 5-bromodeoxyuridine; PCNA. proliferating cell nuclear antigen; CCPR.colon crypt production rate; BL - FU, baseline minus follow-up; bowel prep.
bowel cleansing preparation.
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
Cancer Epidemiology, Biomarkers & Prevention 973
Table I Summ ary of uncont rolled clinical trials of calci um supplem entation and colorectal epithelial cell proliferation
Reference, yearCalcium
Patients ndose (g)
Trial design
Biopsy
time points
(months)
Bowel
prep
Laboratory
method”
Readingreliability
(r)
LIproportional
change (%)
4�,
proportional
change (%)
Comments
Lipkin and Newmark Family history 10 1.25 Uncontrolled 0, 2 Tap water l3HldThd -4O�’ -19
(50), 1985 of sporadic
adenoma
enema
Buset et al. (51), Mixed 9 1.5 Uncontrolled 0, 1.5 Tap water [3HldThd -561986 enema
Lipkin et al. (52), Mixed 28 0.6-0.8 Uncontrolled 0. 3-4 Tap water l3HldThd -28 - I I1989 enema
Rozen et al. (53), Mixed 35 1.25, 1.5 Uncontrolled 0, 1, 1.5, None [3HldThd -36” Used two forms of
1989 2-3 calcium
Biopsies also I .5-2
months
posttreatment
O’Sullivan et al. Sporadic 30 1.0, 2.0 Randomized. 0, 1 ? BrdUrd 1.0 g: 47b 1.0 g Pilot; no real control
(54, 1993 adenoma parallel
group
2.0 g: -40” LI4’: -50
LI,”:
2.0 g
LI(: -44”
LI5”: -32”
group
,, [3H]dThd, tritiated thymidine uptake method of identifying S-phase cells; BrdUrd, BrdUrd uptake method of identifying S-phase cells.
S Statistically significant at P � 0.05.� LI4, LI of the fourth quintile of the crypt (the first quintile includes the base of the crypt).
,/ LI5, LI of the upper or luminal quintile of the crypt.
include the upper (luminal) 40% of the crypt. As an example ofthe magnitude of these differences, in sporadic adenoma, pa-
tients as compared to normal controls (across the eight studieshaving at least 10 persons per group), the mean LI was 43%higher (median, 24%), and the mean 4h was 582% higher(median, 94%). In patients with previous colon cancer or spo-radic adenomas, these changes also predict adenoma recurrence(43, 44). In large bowel tumors in humans, an upward shift in
the proliferative zone is found in colon cancers and adenomasbut not in hyperplastic polyps (45). As reviewed elsewhere (5,
6, 46), proliferative changes in normal-appearing mucosa havebeen shown to be a consequence of both cancer-initiating and
cancer-promoting agents; proliferative changes both precedeand accompany cobonic neoplasms in rodents given chemical
carcinogens, and a high-fat diet produces proliferative changesin both rodents and humans. Animal experimental evidence and
preliminary evidence in humans strongly suggest that these two
proliferation abnormalities (hyperproliferation and upward shiftof the proliferation zone) are reversible biomarkers or precur-
sors for colon neoplasia (5, 6, 46, 47). In humans, the twoproliferation abnormalities appear to be independent variables(35, 48), and rectal biopsy findings on both measures reflectthose throughout the colon (37, 49).
Human Trials of Calcium and Coborectab Epithebial CellProliferation
The results of the human trials of calcium and colorectalepithelial cell proliferation are summarized in Tables 1-3. Inviewing these tables, there are several points that should be
kept in mind: (a) the trials in Table 1 are all uncontrolled(50-54), and thus, although they represent seminal works,they should not be weighed as heavily as the controlledtrials; (b) the distinction between the smaller trials (46,55-62; Table 2) and the larger trials (25, 63, 64; Table 3) is
somewhat arbitrary, and it should be kept in mind that whatconstitutes a study of adequate size is related to the magni-tude of the size and clinical importance of the estimatedtreatment effect and the variability in the end point meas-
urement(s); (c) there are differences among the various trials
in the types of patients (some even have mixed-diagnosis
populations). Conceivably, different types of patients may be
affected differently by calcium supplementation; (d) thedifferent studies all used doses of calcium in approximately
the same range, most used calcium carbonate as the calcium
supplement form. and only one (53) deviated from optimal
clinical trial form by using a mixture of different forms anddoses of calcium supplements (i.e., there was not a random-
ization and comparison of the different forms and doses); (e)
the different trials took biopsies for cell proliferation meas-urements at different time points from one another. In the
tables, in the columns summarizing the results, for the stud-ies having more than one follow-up biopsy visit, only the
results involving the final follow-up visit are presented.
Results of studies with more than one follow-up indicate thatany response to calcium may vary by length of treatment,
and results do not stabilize in the first 2 months (25, 63).
Only one study (64), unfortunately one of the larger studies,
deviated from optimal clinical trial form by not having a
baseline (pretreatment) biopsy. A few studies (52, 53, 64),including, unfortunately, the same larger trial that did not
obtain a baseline biopsy (64), deviated from optimal clinicaltrial form by not standardizing the length of time from onsetof treatment to obtaining the follow-up biopsy of greatest
interest; (f) the studies differed on use of bowel preps. At
least some bowel preps have been shown to affect cellproliferation (65). The studies that have suggested that some
bowel preps do not affect measurements (65) have been
small and may not be conclusive. Some studies deviated
from optimal clinical trial form by using different bowelpreps on different patients in the study (60, 63, 64) and even
by using different bowel preps at baseline and follow-up on
the same patients (60, 63). At this point in our knowledge, it
may be safest to assume that the ideal bowel prep forpurposes of measuring cell proliferation is no prep untilproven otherwise; (g) the different studies used different
laboratory methods for assessing cell proliferation. Pros and
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
974 Review: Calcium and Colon Cell Proliferation
of smaller controlled clinical trials of calcium supplementation and colorectal epithelial cell proliferation
.Reference, year Patients
Calcium
n dose
(g)
. .Tnal design
Biopsy. .
time points
(months)
LaboratoryBowel prep
method
Reading. . .
reliability
(r)
LI
proportional
change(%)
proportional
change(%)
Comments
Gregoire et (11. Postoperative 30 1 .2 Randomized. 0, 1 None L3HldThd +36 LI4”: - 16 Pilot
(55, 1989 colon cancer double blind. Ll3�: +50
placebo
controlled
Stem it al. (56). Postoperative 31 1.2 Randomized, 0, 3, 6. 9 None l3HldThd -21 LI4”: -50 Pilot
199)) FAt” double blind, LI5’: +50
placebo
controlled
Barsoum ci al. Spordic 14 1 .25 Randomized, 0. 2 ? CCPR - l8� Pilot; CCPR not
(57, 1992 adenoma double blind,
placebo
controlled,
validated
cross-over
Wargovich et al. Sporadic 20 2 Randomized, 0, I PEG p.ot [3HbdThd 0.88 �‘ Pilot; did not adhere
(58). 1992 adenoma single blind,
placebo
controlled,
cross-over
to intention-to-treat
principles
Thomas et al. Postoperative 28 0.6 Randomized, 0, 6 None CCPR -42W Pilot; did not adhere
(59), 1993 FAt” double blind,placebo
controlled
to intention-to-treat
principles
Bostick ci al. Sporadic 21 1.2 Randomized, 0, 2 None I3HldThd 0.66 +23 LI4”: +66 Pilot
(46. 1993 adenoma double blind, LI5’: 0
placebo
controlled
Cats el al. (60), FH of HNPCC” 30 1 .5 Randomized, 0, 3 PEG pot and/or BrdU Rectum: - 1 1 Rectum: Pilot
1995 double blind,
placebo
controlled
PEG enema Sigmoid: -32
Descending
colon: +8
- I l2�
Sigmoid:
-ll3�
Descending
colon: +8
Weisgerber et al. Sporadic 30 2 Randomized, -3. 9 7 BrdU and +6 LI4”: +2 Pilot;
(61 ), 1996 adenoma double blind,
placeco
PCNA LI5’: -40 Biopsies
from sigmoid at
colonoscopy; 5
patients BrdUrd,
25 PCNA
Bostick et a!. Ulcerative 31 2 Randomized, 0, 2 None PCNA 0.91 -3 - 10 Pilot; PCNA/BrdUrd’
(62). 1996 colitis double blind,
placebo
controlled
“ I’HbdThd. i’HldThd uptake method of identifying S-phase cells; CCPR, crypt cell production rate method of identifying mitotic cells; BrdUrd and PCNA. methods ofidentifying S-phase cells.
1’ LI4. LI of the fourth quintile of the crypt (the first quintile includes the base of the crypt).
‘ LI,. LI of the upper or luminal quintile of the crypt.
,, Postop FAP. has familial adenomatous polyposis and has received colectomy for it.� Statistically significant at P � 0.05.
1PEG, polyethylene glycol.a � , decrease in LI”; numerical value not provided.
‘, FH of HNPCC, family history of hereditary nonpolyposis colon cancer.‘ PCNA/BrdUrd, S-phase cells identified by the PCNA methods, but only after biopsies underwent incubation process, as if they were to be processed for the BrdUrd
method.
cons of the different laboratory methods are discussed aboveat the end of the previous section; (h) very few studies (25,
46, 58, 62, 64) adhered to optimal clinical trial form bysubmitting blinded biopsy slides to assess biopsy scoringreliability. Of those that did, the scoring reliability was goodexcept for, unfortunately, the same larger trial (64) that did
not obtain baseline biopsies; (i) because different trials useddifferent methods of measuring cell proliferation, and sometrials that used the same or similar methods obtained mark-edly different levels of cell proliferation, to standardize the
magnitude of the estimated treatment effect, the results of
the studies pertaining to LI and 4h are summarized as pro-portional change. For the uncontrolled trials, this was de-fined as follows:
BL - FU
lOO%X BL
For the controlled trials, the definition was the placebo group
proportional change in values minus the calcium group propor-tionab change in values. It should be pointed out that some trials(56-60, 63) did not properly analyze their results as in this
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
Cancer Epidemiology, Biomarkers & Prevention 975
Table 3 Summary oflarger controfled clinicaltrials ofcalcium supplementation and colorectal epithelial cell proliferation
Reference. year Patients nCalcium
dose (g)Trial design
Biopsy
time
points
(months )
Bowel prepL4boratory
method”
Readingreliability
(r)
LI
proportional
change
(%)
�,,
proportional
change
(%)
Comments
Armitage et al.
(63), 1995
Sporadic
adenoma
79 1.5 Randomized,
double
blind,
placebo
controlled
0, 3. 12 None or PEG
p.o.”
CCPR - 10 Full scale
Baron et (21. Sporadic 333 1.2 Randomized, 6-9 PEG poP and/ PCNA 0.40 - 30 +90’ Full scale:(64). 1995 adenoma double
blind.
placebo
controlled
or saline or
tap water
enema
multicentered,
no baseline
biopsy
Bostick ci al. Sporadic 193 1.0. 2.0 Randomized, 0, 1, 2, 6 None PCNA LI”: 0.91 -3 -100’ Full scale:
(25). 1995 adenoma double
blind,
placebo
controlled,
parallel
group
4,,’: 0.77 (1 g: -77
2g: -127’)
PCNAIBrdUrd”
“ CCRP. crypt cell production rate method of identifying mitotic cells: PCNA. proliferating cell nuclear antigen method of identifying S-phase cells.h PEG, polyethylene glycol.‘. Statistically significant at P � 0.05.d PCNA/BrdUrd, S-phase cells identified by the PCNA method but only after biopsies underwent incubation process as if there were to be processed for the BrdUrd uptake
method of identifying S-phase cells.
latter difference of differences [i.e., (BL - FU in the placebogroup) - (BL - Hi in the calcium group)], thus defeating the
point in having a control group. Thus, for these trials, it was
necessary to make the calculations from their published data.
Also, for some trials (50-52, 60), it was necessary to calculatethe LI and/or 4�h from data presented in the paper before making
the calculation for the proportional change. This was not alwayspossible for the 4�. In these cases, if possible, the compart-mental LIs for the fourth and fifth (luminal) compartments of
the crypts are presented as an approximation of the 4s�; and (j)in the “Comments” columns of the Tables, some of the com-ments point out potential concerns about the study summarized
on the intersecting row. Some of these concerns were pointed
out above. Additional concerns pointed out are that, to attainstatistical significance, two trials (58, 59) violated the clinical
trial “intention-to-treat” analysis principle; one trial was mul-
ticentered (64), a dubious design at this point for studies using
such a difficult measurement as cell proliferation; and one trial
(25) measured PCNA after incubating biopsies for BrdUrd, thusmaking the results intermediate in quality between those ofPCNA and BrdUrd studies.
Of the five uncontrolled trials (50-54), all suggested sub-
stantial (28-56%) decreases in the LI. Of the three that meas-
ured 4h in some form (50, 52, 54), all suggested substantial(1 1-44%) decreases in the 4h� One of the uncontrolled trials
(53) suggested that when the calcium intervention was stopped,the LI returned to baseline levels. The data of one uncontrolled
trial (54) suggested that there was no difference in response to1.0 g of calcium compared to 2.0 g of calcium (this interpre-
tation of the data, which is contrary to that of the authors’, was
a result of my own calculations that were based on the “differ-
ence of differences” described above). The major problem withthese seminal studies, of course, is that there was no control
group. Other problems with these small, short studies includedno evidence of blinding of treatment assignment to anyone
involved in the studies, and no reports of biopsy scoring reli-ability. The use in some studies of mixed patient groups, bowel
preps, and laboratory methods involving incubation steps were
potential problems.
Of the nine small controlled trials (46, 55-62), six sug-gested [three statistically significant (57-59)] decreases (3-
21%) in the LI, and three (46, 55, 61) suggested (none statis-
ticably significant) increases (6-36%). Of the two that
measured it (60, 62), both suggested decreases in the 4h (10 and1 12%, of which the larger decrease was statistically signifi-
cant). Four (46, 55, 56, 61) of the nine studies reported com-partmental LIs for the upper two quintile compartments sepa-rateby; of these, all yielded ambiguous results. The two studies
that used the CCPR laboratory method did not attempt to studythe distribution of proliferating cells in the colon crypts (57). In
the only trial that has investigated the effects of calcium inportions of the colon other than the rectum (60), small non-
statistically significant decreases in the LI were suggested to
have occurred in the rectum and sigmoid but not in the de-
scending colon, and large statistically significant decreases in
the 4h were found in the rectum and sigmoid but not in thedescending colon. The major problem with these smaller stud-ies, from the standpoint of statistical inference. of course, istheir small size and lack of statistical power. Despite this
problem, as noted, statistically significant reductions in the LIwere found in three (57-59), and in the 4h in one (60); none of
those that suggested increases in either measure of cell prolif-eration had statistically significant results (46, 55, 61). Otherproblems included, in three studies (55, 56, 59), patient popu-
lations composed of persons with colectomies, hardly a situa-
tion that physiologically replicates the average person at risk forcolon cancer; in two studies (57, 59), the CCPR laboratorymethod, an unvalidated method for human studies, was used; in
one study (58), single rather than double blinding was used, andstatistical significance was achieved in only one arm of the
cross-over and then only when the intention-to-treat principle ofanalysis of controlled trials was violated; in one other study(59), the intention-to-treat principle of analysis of controlledtrials was violated also; in one study (60). different bowel preps
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
0 2 4 6 8
Month
S-Phase
c)
V
Month
Fig. 1. Changes relative to baseline [relative change is presented on a scale like that
of the comparable relative risk or odds ratio (right verticalaxis) and as its equivalent
translation into proportional relative change on a pereentage scale (left vertical axis)l
in rectal mucosa crypt S-phase LI (fraction of crypt cells that are labeled as prolif-
crating) and S-phase 4’,, (fraction ofcrypt cells that are labeled as proliferating that are
in the upper 40% of the crypt) over the course of the Calcium and Colorectal
Epithelial Cell Proliferation clinical trial (n = 193; 25), by treatment group. (There
were no statistically significant differences in LI among treatment groups. Relative to
placebo, on average, the S-phase 4i�, was reduced by one-half(relative effect, 0.50 or
a proportional relative change of 100%) in the combined calcium groups (P = 0.05).
0. placebo; 0 . 1.0 g calcium; 0, 2.0 g calcium.
976 Review: Calcium and Colon Cell Proliferation
were used for different patients and for baseline and follow-up
biopsy visits in some patients; in one study (61), biopsies were
performed at cobonoscopy, but the colon prep was not specified;and only three studies (46, 58, 62) reported on biopsy scoringreliability. The use, in some studies, of bowel preps and labo-ratory methods involving incubation steps were potential prob-lems.
Of the three larger controlled trials (25, 63, 64), nonefound statistically significant reductions in the LI, althoughsmall reductions of 3-10% were suggested in all. One trial usedthe CCPR laboratory method and did not address the 4� (63).The other two trials diverged widely in their findings for the 4�h
One found a large. statistically significant reduction (25), and
the other a barge, statistically significant increase (64). Therewere also substantial differences in the designs and reported
quality assurance in these two trials. Differences between thetwo full-scale trials that measured the 4h were that the trialfinding an increase in the 4,, (64) was multicentered, used
multiple types of bowel preps, had bow intrareader reliabilityfor biopsy scoring, had no baseline measurements of cell pro-
biferation, and measured the 4,, once only (from biopsies ob-tamed during a cobonoscopy that occurred anywhere from 6 to9 months after randomization). The laboratory method for
measuring cell proliferation was the PCNA method. The trialfinding a decrease in the 4,, (25) was single centered; used nobowel preps; measured cell proliferation prior to, and at threeprecise intervals after, randomization; and had high intrareader
reliability for biopsy scoring. The laboratory method for meas-
uring cell proliferation was the PCNA method; however, the
immunohistochemical method for identifying PCNA was usedonly after the biopsies were incubated, as for the BrdUrdmethod. Thus, neither trial was ideal; however, the trial show-
ing an increase in the 4,, (64) had substantially more method-obogical problems, some of which potentially substantially aS-fected the results (i.e., the fasting involved with thecobonoscopy bowel preps, the multiple types of bowel preps,the poor intrareader reliability for biopsy scoring, and the lackof baseline end point measurements). Furthermore, the resultsof the smaller controlled trials were more in accord with the
larger trial showing the decrease in 4,,.The data from the larger trial showing a decrease in 4h
(25) are potentially instructive and are summarized in Table 3and in Fig. 1 . As stated earlier, some of the trials were uncon-
trolled, and some of the controlled trials did not take the controlgroup into account in their analyses. The purpose of the control
group is to account for any occurrence of the regression-to-the-mean phenomenon or any extraneous influences that may occur
and that presumably affect the treatment and control groups
equally, such as temporal influences, laboratory drift, or theHawthorne effect (i.e. , participants under the intense scrutinycharacteristic of the clinical trial setting unconsciously changetheir behavior and thus affect their end point measurements).Thus, changes that occur baseline to follow-up in the controlgroup must be subtracted from baseline to follow-up changes in
the treatment group (i.e., calculate a “difference of differences”to find the true treatment effect in the treatment group). The
data from the full-scale trial (see Fig. 1 ) showed that the LI of
the patients in the placebo group dropped just as much as thatof those in the calcium groups. (For an example of how errorsare propagated when differences of numbers, each with errors,are compared, see the tables in Ref. 25.) Without a randomlyassigned control group and using a proper difference of differ-ences analysis, the apparent drops in the LIs in the calciumgroups during the first 2 months of the trial could have beenfalsely interpreted as treatment effects. These results indicate
S-Phase Labeling Index
that the findings of uncontrolled trials of calcium and theoverall cell proliferation rate were likely the result of regressionto the mean and/or the Hawthorne effect. Alternatively, based
on the results of one uncontrolled trial, the authors of that trial
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
Cancer Epidemiology, Biomarkers & Prevention 977
speculated that patients with relatively high proliferation ratesmay be calcium responsive and that those with relatively low
rates may be unresponsive (52). However, in the full-scale trialshown in Fig. 1 , relative changes (calcium versus placebo) in
the LI were similar in both those with high and low prolifera-tion at baseline (data not shown), thus indicating that the
findings of the uncontrolled trial were likely the result ofregression to the mean. In another uncontrolled trial, there was
a drop in the LI 2 months into the study, which returned tobaseline 2 months after cessation of treatment (53). The authorstook this as further evidence of a calcium treatment effect.However, as noted in Fig. 1, the LI of patients in the placebogroup of the full-scale controlled trial dropped as much as those
in the treatment groups during the first 2 months and, as didthose in the treatment groups, rose back to baseline by the next
time point (6 months), thus indicating that the findings of theuncontrolled trial were likely the result of the Hawthorne effect,which was wearing off after the initial months of the study. Not
only do control groups that are properly taken into account inthe analyses prevent falsely seeing treatment effects, but theymay also help prevent missing treatment effects. For example,
as noted in Fig. 1, without a control group, the treatment effecton the 4�h would have been underestimated because the time-related influences producing an increase in the 4h from 2 to 6
months in all treatment groups (seen most prominently in theplacebo group but also to some extent in the two calciumgroups) would not have been accounted for. A final point is that
from a trial without a baseline measurement, it would not beknown whether the treatment and control groups started out at,
on average, the same end point values; thus, differences at
follow-up may merely be a perpetuation of differences at base-
line rather than a treatment effect.
Conclusions
Taking the results of all of the published studies of calciumsupplementation and coborectab epithelial cell proliferation ab-together and considering all their strengths and weaknesses, thedata do not support the hypothesis that calcium supplementa-
tion reduces the LI, or rate of proliferation of the colon epithe-hum, in humans. If calcium does affect the LI in humans, itwould appear that the effect is small (say, on average, in adults
consuming an average intake of calcium, 5%). The data are notconclusive but do suggest that calcium supplementation mayreduce the 4sh (i.e., normalize the distribution of proliferating
cells in colon crypts) and may do so substantially (say, on
average, in adults consuming an average intake of calcium,100%). Ifcalcium affects the 4� and not the LI, then this is stillconsistent with the hypothesis that a higher consumption ofcalcium may reduce the risk of colorectal cancer. Previous
studies (35, 48) have shown that the LI and 4,,, are statisticallyindependent variables, and other preliminary controlled trials
testing other agents hypothesized to normalize colorectal pro-liferative kinetics and to reduce risk of colon cancer have found
reductions in the 4�h without reductions in the LI (66, 67).If calcium normalizes the distribution of proliferating cells
without affecting the overall proliferation rate in humans, themechanistic hypotheses for how calcium affects cell prolifera-
tion in viva and may reduce colon cancer needs to be reexam-med. The first and most cited hypothesis for how calcium mightreduce colorectal cancer risk has been that calcium binds in-traluminal bile acids, thus preventing their toxic effects with
their resultant promotion of compensatory hyperproliferation(4). However, the data provide little evidence for this explan-
atory hypothesis. First, the evidence does not support an effect
on the overall cell proliferation rate. Second, the hypothesiswould predict that, for a person consuming a level of fat and
calcium in the Western-style diet range, the 1 .0-g calcium dose(resulting in a total calcium intake of 1 .5-2.0 g daily) would
have provided as great as an effect as the 2.0-g dose in thelarger trial that found a reduction in 4h#{149} The data, however,were consistent with a greater effect provided by the higherdose. Furthermore, recent reports of studies that have examined
the effects of calcium supplementation on stool bile acids inhumans are inconsistent with one another (68 -7 1 ). These linesof evidence do not rule out a beneficial effect of calcium via
bile acid binding but do suggest that, if this is involved, themechanism is more complex than previously thought.
On the basis of in vitro data showing that calcium directlyaffects the cell cycle, modulating cell proliferation and inducingterminal differentiation, it has also been hypothesized that cal-
cium may exert a similar influence in vivo in the cells of thecolon mucosa (5, 46, 72). The data from the larger trial that
found a decrease in the 4h (25) are consistent with this hypoth-esis. Even when proliferation rates are fast, if differentiation
occurs rapidly as cells migrate up the crypt, they are more likelyto have completed proliferation bower in the crypt. A conse-quence of this may be that, because the DNA of a cell under-going replication is more vulnerable to damage by variousagents and because cells proliferating lower in the crypt may be
less likely to be exposed to injurious intraluminal agents, such
cells may be less likely to be involved in colorectal carcino-
genesis.The mechanisms by which calcium affects cell cycle are
not clear; however, several lines of research indicate that cal-
cium may exert such effects by interacting with cyclic AMP(73), calmodulin (6, 74), tyrosmne kinase (75), and ornithine
decarboxylase (5, 75). In addition, calcium may influence othermechanisms, e.g., cell adhesion mechanisms involving E-cad-hem, a calcium-dependent cell adhesion molecule that inter-acts in complex fashion with the adenomatous polyposis cobi
gene product (76, 77).Although studies in humans have found that cell prolifer-
ation kinetics found on rectal biopsies reflect those foundthroughout the colon (37, 49) and studies in rodents have found
that calcium affects cell proliferation kinetics throughout thecolon (1 1-13), there are insufficient data to assess whether
calcium affects cell proliferation throughout the colon in hu-mans. This remains an important question, especially given that
the epidemiology of rectal cancer appears different in severalrespects from that of colon cancer ( 1).
Another point is that coborectal epithelial cell proliferationkinetics remain unproven but logical and well-supported inter-mediate end points for colon cancer. The existing cell prolif-eration studies in humans therefore cannot prove that, because
calcium may normalize one of the cell proliferation end pointsin the rectum, calcium can reduce the risk of colon cancer; theydo, however, provide justification for additional study of thecalcium-colon cancer association. More work needs to be done
to determine whether calcium supplementation can normalizecolorectab epithelial cell proliferation kinetics in humans. A
clinical trial to test the efficacy of calcium on reducing the
incidence of colon cancer is not justified at this time.
Recommendations
Although it is not reviewed herein, there needs to be an ac-
knowledgment that measuring coborectal epithebiab cell prolif-eration is difficult and susceptible to numerous sources ofvariability due to error (27). More work is needed on quanti-
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
978 Review: Calcium and Colon Cell Proliferation
tating coborectal epithebial cell proliferation. More work isneeded to investigate the effects of bowel preps on coborectalepithelial cell proliferation.
Although it is not reviewed herein, more work is needed to
establish whether coborectab epithebiab cell proliferation is a trueintermediate end point for colon cancer. More work also needs
to be done on the number of crypts that should be counted perperson in clinical trials. One study (46) indicated that approx-imateby eight crypts should be scored in sporadic adenomapatients assessed using [3HJdThd; however, the number may
differ by cell proliferation measurement methodology, the lab-
oratory peforming the measurements, the patient populationunder study, and the sample size.
Improvements are needed in intermediate end point trials
in general and in coborectab epithebiab cell proliferation endpoint trials in particular.
(a) It is acknowledged that uncontrolled seminal studiesare often done on small budgets, are often focused on method-ology, and are legitimate for establishing whether additional,more definitive study is justified. However, such studies shouldprovide little weight in weighing the evidence of whether agiven intervention produces a treatment effect. Studies that
should be given the most weight should be randomized and
controlled, and where possible, double-blinded.(b) End points such as colorectal epithelial cell prolifera-
lion, should be measured at baseline and follow-up.(c) All participants of clinical trials should be treated
exactly alike, and the conditions at end point ascertainmentshould be exactly alike at baseline and follow-up for all par-
ticipants. For example, for coborectal epithelial cell prolifera-lion trials. because bowel preps may affect coborectal epithebiabcell proliferation, either no or only one proven type of bowelprep should be used for all patients at all biopsy visits.
(d) Quality assurance should be a priority in all trials,regardless of their size. A high level of end point (e.g.. cob-rectal epithelial cell proliferation) measurement reliability
should be attained before a trial begins, and intra- (and whereappropriate, inter-) rater reliability should be monitored
throughout the trial by submission of blinded repeat or splitsamples and presented in the final manuscript reporting the
results of the trial. Measures of adherence to treatment andclinical trial protocol should also be monitored and reported.
(e) Extreme caution should be used in attempting to use inmulticentered studies an end point that is so difficult to measure
and that is so sensitive to extraneous influences as is thecoborectab epithelial cell proliferation end point.
(11Proper controlled trial analysis methods should be used.Proper methods for controlled trials with baseline and fob-low-up end point measurements do not include control group
follow-up versus treatment group follow-up comparisons, ortreatment group baseline versus follow-up measurements.
Rather, in principle, the proper method should estimate thetreatment effect by taking the control group BL - FU minus the
treatment group BL - FU and then test this result for statisticalsignificance by comparing the control group BL - FU versus
the treatment group BL - FU. Of course, there are variationsof this. For example. to account for the need to include datafrom trials with multiple follow-up time points in repeated
measures mixed models in which the treatment effects areexpressed as odds ratios or relative effects, the following cab-culation is employed: (treatment group follow-up/treatmentgroup baseline):(control group follow-up/control group base-line). In addition, for the inferential statistical tests used, the
end point measures may need to be transformed to meet theassumptions of normality (25, 64, 78). For example, for cob-
rectal epithelial cell proliferation, the data usually need to betransformed by the natural logarithm before the data are enteredinto the tests of statistical significance (25, 78). Few of thereported calcium and coborectal epithelial cell proliferationstudies even mentioned whether the data were tested to seewhether they met the assumptions for the statistical tests used.
(g) The relationships among sample size, size of the treat-ment effect, and amount of variability in the end point meas-
urements should be kept in mind, both for planning samplesizes for studies and in interpreting the results of studies. A
common problem noted in this review of the human studies ofcalcium and colorectal epithelial cell proliferation was authors
authoritatively, unequivocally stating conclusions of no effectbecause the results were not statistically significant, when in
fact the estimated treatment effect was large, but the statisticalpower quite low due to the small sample size and/or the bargevariability of the LI or 4� (55, 56, 60). Such studies should beregarded as pilot studies that actually suggest that there may be
a treatment effect and that a full-scale trial is justified todefinitively address the tested hypothesis. Current evidence isthat full-scale clinical trials using colorectal epithelial cell pro-liferation as the end point need to be larger than originallyhoped for but still much smaller than trials of cancer or colonpolyp occurrence. For example, it has been estimated that for a
two-group controlled trial using current PCNA methods to havesufficient statistical power to detect a proportional reduction in
LI of 20% at P � 0.05 (recall that across eight validation
studies, the LI was proportionately higher in sporadic adenomapatients than in normal controls by a mean and median, respec-lively, of 43% and 24%), the sample size needs to be about 100
patients per group (79).Another calcium and coborectal epithelial cell proliferation
epitheliab cell proliferation clinical trial is needed. Such a trialshould only be carried out after sufficient improvement in
measurement and validation of coborectal epithelial cell prolif-eration as a true intermediate end point for colorectal cancer.
The trial should be full scale, i.e., with sufficient sample size/statistical power to answer the question, and should use the
suggestions outlined above. It would also be desirable for thetrial to address calcium effects proximal to the rectum as well
as in the rectum and to address mechanisms of action of
calcium on cell proliferation.
Summary
In summary, there are biologically plausible mechanisms ofaction for protective effects of calcium against colon cancer,
and it is plausible that these mechanisms involve normalizationof coborectal epitheliab cell proliferation kinetics. Currently,
animal experimental data are strongly supportive. Epidemio-logical data are inconsistent but, overall, are weakly supportive
as well. A review of the 17 human trials (5 small and uncon-trolled, 9 small and controlled, and 3 larger and controlled) ofcalcium and colorectal epithelial cell proliferation suggests that
calcium supplementation, without affecting the proliferationrate, may normalize the distribution of proliferating cells in the
rectal mucosa. This supports the hypothesis that higher calciumconsumption may reduce the risk of colon cancer. It alsosupports the hypothesis that the possible chemoprotective ac-lion of calcium may not be simply binding bile acids andthereby reducing compensatory hyperproliferation, but it isconsistent with the hypothesis that calcium exerts its possiblechemoprotective effect by directly affecting cell cycle andincreasing rates of cell differentiation. The mechanism(s) bywhich calcium may affect coborectal epithelial cell proliferation
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
Cancer Epidemiology, Biomarkers & Prevention 979
in humans remain(s) to be resolved. Not only do the currently
available data indicate that a causal relationship between cal-cium intake and colorectal cancer incidence, although increas-
ingly supported, cannot be considered established, but theyeven indicate that a calcium and colon cancer prevention trial isnot yet justified. Much work needs to be done to improve themethodology of measuring colorectal epithelial cell prolifera-tion and to validate whether it is a true intermediate end point
for colorectal cancer. The quality of the conduct and analysis ofcoborectal epithelial cell proliferation trials needs to be im-
proved. When these issues are addressed adequately, another
full-scale calcium and colorectal epithelial cell proliferationclinical trial that addresses mechanisms of action and effects inmultiple sites of the colon will be needed.
References
1. Potter. J. D., Slanery, M. L., Bostick, R. M., Gapstur SM. Colon cancer: a
review of the epidemiology. Epidemiol. Rev., 15: 499-545, 1993.
2. Eaton, S. B., Shostak, M., and Konner, M. The Paleolithic Prescription. New
York: Harper and Row, 1988.
3. Welt, L. G., and Blythe, W. B. Cations: calcium. magnesium. barium, lithium,
and ammonium. In: Goodman, L. S., and Oilman, A. (eds.), The Pharmacologic
Basis of Therapeutics, Ed. 4. New York: MacMillan Publishing Co., 1970.
4. Newmark, H. L., Wargovich, M. J., and Bruce, W. R. Colon cancer and dietary
fat, phosphate and calcium: a hypothesis. J. Nail. Cancer Inst., 72: 1323-1325,
1984.
5. Newmark, H. L., and Lipkin, M. Calcium, vitamin D. and colon cancer. Cancer
Res., 52 (Suppl.): 2067s-2070s, 1992.
6. Wargovich, M. J., and Baer, A. R. Basic and clinical investigations of dietary
calcium in the prevention of colorectal cancer. Prey. Med., 18: 672-679. 1989.
7. Kandell, R. L., and Bernstein, C. Bile salt/acid induction of DNA damage in
bacterial and mammalian cells: implications for colon cancer. Nutr. Cancer, 16:
227-238, 1991.
8. Ranken, R., Wilson, R., and Bealmear. P. M. Increased turnover of intestinalmucosal cell ofgermfree mice induced by cholic acid. Proc. Soc. Exp. Biol. Med..
138: 270-242, 1971.
9. Chomchai, C., Bhadrachari, N., and Nigro, N. D. The effect of bile on the
induction of experimental intestinal tumors in rats. Dis. Colon Rectum. 17:
310-312, 1974.
10. Narisawa, T., Magadia, N. E., Weisburger, J. H., and Wynder. E. L. Pro-
moting effect of bile acids on colon carcinogenesis after intrarectal instillation of
N-methyl-N’-nitro-N-nitroguanidine in rats. J. Natl. Cancer Inst., 53: 1093-1097,
1974.
11. Wargovich. M. J., Eng, V. W. S.. and Newmark, H. Calcium inhibits the
damaging and compensatory proliferative effects of fatty acids on mouse colon
epithelium. Cancer Lett., 23: 253-258, 1984.
12. Bird, R. P.. Schneider, R., Stamp, D., and Bruce, W. R. Effect of dietary
calcium and cholic acid on the proliferative indices of murine colonic epithelium.
Carcinogenesis (Land.), 7: 657-661, 1986.
13. Wargovich, M. J., Eng. W. W. S., Newmark, H. L., and Bruce, W. R. Calcium
ameliorates the toxic effect of deoxycholic acid on colonic epithelium. Carcino-
genesis (Land.), 4: 1205-1207, 1983.
14. Appleton, G. V. N., Davies, P. W., Bristol, J. B., and Williamson, R. C. N.
Inhibition of intestinal carcinogenesis by dietary supplementation with calcium.
Br. J. Surg.. 74: 523-525, 1987.
15. Pence. B. C., and Buddingh, F. Inhibition of dietary fat-promoted colon
carcinogenesis in rats by supplemental calcium or vitamin D3. Carcinogenesis
(Lond.), 9: 187-190, 1988.
16. Behling. A. R.. Kaup, S. M., Choquette, L. L., and Greger. J. L. Lipid
absorption and intestinal tumour incidence in rats fed on varying levels of calcium
and butterfat. Br. J. Nutr., 64: 505-513, 1990.
17. Wargovich, M. J., AlInun, D., Palmer, C., Anaya, P., and Stephens, L. C.
Inhibition of the promotional phase of azoxymethane-induced colon carcinogen-
esis in the F344 rat by calcium lactate: effect of simulating two human nutrient
density levels. Cancer Lett., 53: 17-25, 1990.
18. McSherry. C. K., Cohen, B. 1., Bokkenheuser, V. D., Mosbach, E. H., Winter,
J., Matoba, N., and Scholes, J. Effects of calcium and bile acid feeding on colon
tumors in the rat. Cancer Res., 49: 6039-6043, 1989.
19. Sitrin, M. D., Halline, A. 0., Abrahams, C., and Brasitus, T. A. Dietary
calcium and vitamin D modulate 1,2-dimethylhydrazine-induced colonic carci-
nogenesis in the rat. Cancer Res., 51: 5608-5613. 1991.
20. Kaskare, M. R., Clark, T. D., and Glauert, H. P. Effect of dietary calcium on
colon carcinogenesis induced by a single injection of 1,2-dimethylhydrazine in
rats. J. Nuft., 121: 568-577, 1991.
21. Bostick, R. M., Potter, J. D., Sellers, T. A., McKenzie, D. R., Kushi, L. H..and Folsom, A. R. Relation of calcium, vitamin D, and dairy food intake to
incidence of colon cancer among older women. Am. J. Epidemiol., 137: 1302-
1317, 1993.
22. Vogelstein, B., and Knizler, K. W. The multi-step nature of cancer. Trends
Genet., 9: 138-141, 1993.
23. Lipkin. M., Enker, W. E., and Eilers, G. A. M. Tritiated-thymidine labeling
of rectal epithelial cells in “non-prep” biopsies of individuals at increased risk for
colonic neoplasia. Cancer Leu.. 37: 155-161, 1987.
24. Risio, M., Coverlizza, S., Poccardi, G., Candelaresi, G., and Gaiola, 0. In
vitro immunohistochemical localization of S-phase cells by a monoclonal anti-
body to bromodeoxyuridine. Basic AppI. Histochem., 30: 469-477, 1986.
25. Bostick, R. M., Fosdick, L., Wood, J. R., Grambsch, P., Grandits, G., and
Lilliemoe, J. T. Calcium and colorectal epithelial cell proliferation in sporadic
adenoma patients. J. Natl. Cancer Inst., 87: 1307-1315, 1995.
26. Allan A., and Jewell, D. P. In vitro model for the assessment of luminal
factors on rectal mucosa. Gut, 24: 812-817, 1983.
27. Bostick, R. M., Fosdick, L., Lillemoc, T. J., Overn, P., Wood, J. R.,
Grambsch. P.. Elmer, P.,T. A. and Potter, J. D. Methodologic findings and
considerations for measuring colorectal epithelial cell proliferation in humans.
Cancer Epidemiol. Biomark. Prey., 6: 1997.
28. Bleiberg, H., Buyse, M., and Galand, P. Cell kinetic indicators of premalig-
nant stages of colorectal cancer. Cancer (Phila.), 56: 124-129, 1989.
29. Paganelli. G. M., Santucci, R., Biasco, G., Miglioli. M., and Barbara, L.
Effect of sex and age on rectal cell renewal in humans. Cancer Len.. 53: 1 17-121,
1990.
30. Paganelli, G. M., Biasco, G., Santucci, R., Brandi. G.. Lalli, A. A., Miglioli,
M., and Barbara, L. Rectal cell proliferation and colorectal cancer risk level in
patients with nonfamilial adenomatous polyps of the large bowel. Cancer (Phila.),
68: 2451-2454, 1991.
31. Ponz de Leon, M., Roncucci, L., Di Donato, P., Tassi, L., Smerieri, 0.,Grazia. M., Malagoli. A. G., Dc Maria. D., Antonioli, A., Chahin, N., Perini, M.,
Rigo, G., Barberini. G., Manenti, A., Biasco. G., and Barbara, L. Pattern of
epithelial cell proliferation in colorectal mucosa of normal subjects and of patientswith adenomatous polyps or cancer of the large bowel. Cancer Res., 48: 4121-
4126, 1988.
32. Lipkin. M., Blauner, W. E.. Fraumeni. J. F.. Jr., Lynch, H. T., Deschner, E.,
and Winawer, S. Tritiated thymidine (44’,,) labeling distribution as a marker for
hereditary predisposition to colon cancer. Cancer Res., 43: 1899-1904, 1983.
33. Lipkin. M., Uehara, K., Winawer, S., Sanchez, A., Bauer, C., Phillips. R.,
Lynch, H. T., Blattner, W. A., and Fraumeni, J. F., Jr. Seventh-Day Adventist
vegetarians have quiescent proliferative activity in colonic mucosa. Cancer Lett.,
26: 139-144, 1985.
34. Lipkin, M., Enker, W. E., and Eilers, G. A. M. Tritiated-thymidine labeling
of rectal epithelial cells in “non-prep” biopsies of individuals at increased risk for
colonic neoplasia. Cancer Len., 37: 155-161, 1987.
35. Risio, M., Lipkin. M., Candelaresi, G., Bertone, A., Coverlizza, S., and
Rossini, F. Correlations between rectal mucosa cell poliferation and the clinical
and pathological features of nonfamilial neoplasia of the large intestine. Cancer
Res., 51: 1917-1921, 1991.
36. Stadler, J., Yeung. K. S., Furrer, R., Marcon, N., Himal, H. S., and Bruce,w. R. Proliferative activity of rectal mucosa and soluble fecal bile acids in
patients with normal colons and in patients with colonic polyps or cancer. Cancer
Lett., 38: 315-320, 1988.
37. Terpstra, 0. T., Strautenstein. M. V., Dees. J.. and Eilers. G. A. M. Abnormal
pattem of cell proliferation in the entire mucosa of patients with colon adenoma
or cancer. Gastroenterology, 92: 704-708, 1987.
38. Deschner, E. E., Lewis, D. M., and Lipkin. M. In vitro study of human rectalepithelial cells. 1. Atypical zone of H3 thymidine incorporation in mucosa of
multiple polyposis. J. Clin. Invest., 42: 1922-1928, 1963.
39. Biasco. G., Lipkin, M.. Minarmni. A., Higgins, P., Miglioli, M., and Barbara,
L. Proliferative and antigenic properties of rectal cells in patients with chronic
ulcerative colitis. Cancer Res., 44: 5450-5454, 1984.
40. Bleiberg, H., Mainguet, P., Galand, P., Chretien, J., and Dupont-Mairesse, N.
Cell renewal in the human rectum: in vitro autoradiographic study on active
ulcerative colitis. Gastroenterology, 58: 851-855, 1970.
41. Gerdes, H., Gillin, J. S., Zimbalist, E., Urmacher, C., Lipkin, M., andWinawer, S. J. Expansion of the epithelial proliferative compartment and fre-
quency of adenomatous polyps in the colon correlate with the strength of familyhistory of colorectal cancer. Cancer Res.. 53: 279-282, 1993.
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from
980 RevIew: Caldum and Colon Cell Proliferation
42. Roncucci. L., Ponz de Leon, M., Scalmati, A., Malgoli, G., Pratissoli, S.,Perini. M.. and Chahin, N. J. The influence of age on colonic epithelial cell
proliferation. Cancer (Phila.). 48: 235-245. 1988.
43. Anti, M., Marra, G., Armelao, F., Percesepe, A., Ficarelli, R., Ricciuto,G. M.. Valenti, A.. Rapaccini. G. L.. Dc Vitis. I., D’Agnostino, G., Brighi, S., and
Vecchio, F. M. Rectal epithelial cell proliferation patterns as predictors of
adenomatous colorectal polyp recurrence. Gut, 34: 525-530, 1993.
44. Scalmati, A., Roncucci, L., Ghidini, G., Biasco, G., and Ponz de Leon, M.Epithelial cell kinetics in the remaining colorectal mucosa after surgery for cancer
of the large bowel. Cancer Res., 50: 7937-7941, 1990.
45. Risio, M., Coverlizza, M., Ferrari, A., Candelaresi, G., and Rossini, F.
Immunohistochemical study of epithelial cell proliferation in hyperplastic polyps,adenomas, and adenocarcinomas of the large bowel. Gastroenterology, 94: 899-
906. 1988.
46. Bostick. R., Potter, J. D.. Fosdick, L., Grambsch, P., Lampe. J., Wood. J.,
Lauis, T., Ganz. R., and Grandits, G. Calcium and colorectal epithelial cell
proliferation: a preliminary randomized, double-blinded, placebo-controlled din-ical trial. J. NatI. Cancer Inst., 85: 132-141, 1993.
47. Rozen, P. An evaluation of rectal epithelial proliferation measurement asbiomarker of risk for colorectal neoplasia and response in intervention studies.
Eur. J. Cancer Prey., I: 215-224, 1992.
48. Grambsch, P. M., Randall, B. L., Bostick, R. M., Potter, J. D., and Lauis,T. A. Modeling the labeling index distribution: an application of functional data
analysis. J. Am. Stat. Assoc.. 90: 813-821. 1995.
49. Potten, C. S., Kellen, M., Roberts, S. A., Rew, D. A., and Wilson, 0. D.
Measurement of in vivo proliferation in human colorectal mucosa using bromode-
oxyuridine. Gut, 33: 71-81. 1992.
50. Lipkin, M., and Newmark, H. Effect of added dietary calcium on colonic
epithelial cell proliferation in subjects at high risk for familial colonic cancer.N. EngI. J. Med., 313: 1381-1384, 1985.
51. Buset, M., Lipkin. M., Winawer, S.. Swaroop, S., and Friedman, E. Inhibition
of human colonic epithelial cell proliferation in vivo and in vitro by calcium.
Cancer Res., 46: 5426-5430. 1986.
52. Lipkin, M., Friedman, E., Winawer, S. J., and Newmark, H. Colonic epithe-
hal cell proliferation in responders and nonresponders to supplemental dietary
calcium. Cancer Rca., 49: 248-254, 1989.
53. Roam, P., Fireman, A., Fine, N., Wax, Y., and Ron, E. Oral calcium
suppresses increased rectal epithelial proliferation of persons at risk of colorectal
cancer. Gut, 30: 650-655, 1989.
54. O’Sullivan, K. R.. Mathias, P. M., Beanie, S., and O’Morain, C. Effect of oral
calcium supplementation on colonic crypt cell proliferation in patients withadenomatous polyps of the large bowel. Eur. J. Gastroenterol. Hepatol., 5: 85-89,
1993.
55. Gregoire, R.. Stem, H. S., Yeung, K. S.. Stadler, J., Langley, S., Furrer, R.,
and Bruce, W. R. Effect of calcium supplementation on mucosal cell proliferation
in high risk patients for colon cancer. Gut, 30: 376-382, 1989.
56. Stem, H. S., Gregoire, R. C., Kashtan, H., Stadler, J., and Bruce, R. W.
Lang-term effects of dietary calcium on risk markers for colon cancer in patientswith familial polyposis. Surgery (St. Louis), 108: 528-533. 1990.
57. Barsoum, G. H., Hendrickse, C., Winslet, M. C., Youngs, D., Donovan, I. A.,
Neoptolemos, J. P., and Keighley, M. R. B. Reduction of mucosal crypt cell
proliferation in patients with colorectal adenomatous polyps by dietary calciumsupplementation. Br. J. Surg.. 79: 581-583, 1992.
58. Wargovich, M. J., Isbell, G.. Shabot, M.. Winn, R., Lanza, F., Hochman, L.,
Larson, E., Lynch, P., Roubein, L., and Levin, B. Calcium supplementation
decreases rectal epithelial cell proliferation in subjects with sporadic adenoma.
Gastroenterology, 103: 92-97, 1992.
59. Thomas, M. G., Thomason, J. P. 5., and Williamson, R. C. N. Oral calcium
inhibits rectal epithelial cell proliferation in familial adenomatous polyposis.
Br. J. Surg., 80: 499-501, 1993.
60. Cats, A., Kleibeuder, J. H., van dci Meer, R., Kuipers, F., Sluiter, W. J.,
Hardonk, M. J., Oremus, E. T. H. G. J., Mulder, N. H., and de Vries, E. G. E.
Randomized, double-blinded, placebo-controlled intervention study with supple-
mental calcium in families with hereditary nonpolyposis colorectal cancer. J. Nail.
Cancer Inst., 87: 598-603, 1995.
61. Weisgerber, U. M., Boeing, H., Owen, R. W., Waldherr, R., Raedsch, R., and
Wahrendorf, J. Effect of longterm placebo controlled calcium supplementation on
sigmoidal cell proliferation in patients with sporadic adenomatous polyps. Gut,
38: 396-402, 1996.
62. Bostick, R. M., Boldt, M., Darif, M., Fosdick, L., Wood, J. R., Ovem, P., and
Lillemoe, T. Calcium and colorectal epithelial cell proliferation in ulcerative
colitis. Cancer Epidemiol. Biomark. Prey., 5: 235, 1996.
63. Armitage, N. C., Rooney, P. S., Gifford, K-A., Clarke, P. A., and Hardcastle,
J. D. The effect of calcium supplements on rectal mucosal proliferation. Br. J.
Cancer, 71: 186-190, 1995.
64. Baron, J. A., Tosteson, T. D., Wargovich, M. J., SandIer, R., Mandel, J.,
Bond, J., Haile, R., Summers, R., van Stolk, R., Rothsstein, R., and Weiss, J.
Calcium supplementation and rectal mucosal proliferation: a randomized con-
traIled trial. J. Natl. Cancer Inst., 87: 1303-1307, 1995.
65. Alberta, D. S., Einsphar, J., Alckin, M., Hixson, L., Earnest, D., Roe, D., and
Powell, M. Validation of proliferation indices as surrogate endpoint markers.
J. Cell. Biochem., 19: 76-83, 1994.
66. Anti, M., Marra, G., Armelao, F., Percesepe, A., Ficarelli, R.. Ricciutu,
G. M., Valenti, A.. Rapaccini, 0. L., Dc Vitis, I., D’Agostino, G., Brighi, S., and
Vecchio, F. M. Effect of s�,-3 fatty acids on rectal mucosal cell proliferation in
subjects at risk for colon cancer. Gastroenterology, 102: 883-891, 1992.
67. Paganelli, G. M., Biasco, G., Brandi. G., Santucci, R., Gizzi, G., Villani, V.,Cianci, M., Miglioli, M., and Barbara, L. Effect of vitamin A, C. and E supple-
mentation on rectal cell proliferation in patients with colorectal adenomas. J. Nail.
Cancer Inst., 84: 47-51, 1992.
68. Alder, R., McKeown-Eyssen, 0., and Bright-See, E. Randomized trial of the
effect of calcium supplementation on fecal risk factors for colorectal cancer.
Am. J. Epidemiol.. 138: 804-814, 1993.
69. Lapr#{233},J. A., Dc Vries, H. T., Termont, D. S. M. L., Kleibeuker, I. H., Dc
Vries, E. G. E., and Van der Meer, R. Mechanism of the protective effect of
supplemental dietary calcium on cytolytic activity of fecal water. Cancer Res., 53:248-253, 1993.
70. Van der Meer, R., Welberg, J. W. M., Kuipers. F.. Kleibeuker, J. H., Mulder,
N. H., Termont, D. S. M. L., Vonk, R. J., Dc Vries, H. T., and Dc Vries, E. G. E.
Effects of supplemental dietary calcium on the intestinal association of calcium,phosphate, and bile acids. Gastroenterology, 99: 1653-1659, 1990.
71. Van der Meer, J. W. M., Kleibeuker, J, H., Van der Meer, R., Kuipers, F.,
Cats, A., Van Rijsbergen, H., Termont, D. S. M. L, Boersma-Van Ek, W., Vonk,
F. J., Mulder, N. H., and Dc Vries, E. G. E. Effects of oral calcium supplemen-
tation on intestinal bile acids and cytolytic activity of fecal water in patients with
adenomatous polyps of the colon. Eur. J. Clin. Invest., 23: 63-68, 1993.
72. Yang, K., Cohen, L., and Lipkin, M. Lectin soybean agglutinin: measure-ments in colonic epithelial cells of human subjects following supplemental dietary
calcium. Cancer Left., 56: 65-69, 1991.
73. Whitfield, J. F., Boynton, A. L., MacManus, J. P., Rixon, R. H., Sikorska, M.,
Tsang, B., and Walker, P. R. The roles of calcium and cyclic AMP in cell
proliferation. Ann. NY Acad. Sci., 339: 216-240, 1980.
74. Rasmussen, CD., and Means, A. R. Calmodulin is involved in regulation of
cell proliferation. EMBO J., 6: 3961-3968, 1987.
75. Arlow, F. L. Walczak, S. M., Luk, 0. D., and Majumdar, A. P. N. Attenu-
ation of azoxymethane-induced colonic mucosal ornithine decarboxylase and
tyrosine kinase activity by calcium in rats. Cancer Res.. 49: 5884-5888, 1989.
76. Su, L-K., Vogelstein, B., and Kinzler, K. W. Association of the APC tumor
suppressor protein with catenins. Science (Washington DC). 262: 1734-1737,
1993.
77. Rubinfeld, B., Souza, B., Albert, I., Muller, 0., Chamberlain, S. H., Masiarz,
F. R., Munemitsu, S., and Polakis, P. Association of the APC gene product with
fI-catenin. Science (Washington DC), 262: 1731-1734, 1993.
78. Freedman, L., Graubard, B., and Schatzkin, A. Statistical validation of
intermediate endpoints for chronic diseases. Stat. Med.. 11: 167-178, 1992.
79. Lyles, C. M., Sandler, R. S., Keku. T. 0., Kupper, L. L., Mitlikan, R. C.,
Murray, S. C., Bangdiwala, S. I., and Ulshen, M. H. Reproducibility and varia-
bility of the rectal mucosal proliferation index using proliferating cell nuclear
antigen immunohistochemistry. Cancer Epidemiol. Biomark. Prey., 3: 597-605,1994.
Association for Cancer Research. by guest on September 7, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from