This study was approved by the Emory University IRB. Written informed consent was obtained from each study participant.
Participants were recruited from the patient population attending the Digestive Diseases Clinic at Emory University. Eligibility for the study included 30–75 years of age, in general good health, capable of informed consent, a history of at least one pathology-confirmed adenomatous colonic or rectal polyp within the past 36 months, no contraindications to calcium or vitamin D supplementation or rectal biopsy procedures, and no medical conditions, habits, or medication usage that would otherwise interfere with the study as described below. Detailed eligibility and exclusion criteria were previously reported (9
Clinical Trial Protocol
All age-eligible patients diagnosed with at least one pathology confirmed adenomatous colonic or rectal polyp within the past 36 months were identified as potential study participants. All patients passing initial chart screening for eligibility were sent an introductory letter, followed by a telephone interview. During the telephone interview, a few preliminary screening questions were asked and, if a person was willing, still appeared eligible, and could be available for the next eight months, an in-person eligibility visit was scheduled. Potential participants were asked to bring all medications and vitamins and minerals being taken to this appointment.
During the eligibility visit, potential participants were interviewed, signed a consent form, completed questionnaires (included questions on socio-demographics, medical history and medication use, nutritional supplement use, lifestyle, family history, and others), and provided a blood sample. Diet was assessed with a semi-quantitative Willett food frequency questionnaire (22
). Medical and pathology records were reviewed. Those still eligible and willing to participate then entered a 30-day placebo run-in trial. Only participants without significant perceived side effects and who had taken at least 80% of their tablets were eligible for randomized assignment. Adherence for the run-in trial was assessed by questionnaire, interview, and pill count.
Eligible participants then had their vital signs taken, underwent a baseline rectal biopsy and, if still willing to participate, were randomly assigned (stratified by sex and nonsteroidal anti-inflammatory drug [NSAID] use) to one of four treatment groups. Of patients who passed initial chart eligibility, 42% were contacted and 20% were eligible and consented to participate.
Participants (n=92) were randomly assigned to the following four treatment groups: a placebo control group, a 2.0 g elemental calcium supplementation group (as calcium carbonate in equal doses twice daily), an 800 IU vitamin D3 supplementation group (400 IU twice daily), and a calcium plus vitamin D3 supplementation group taking 2.0 g elemental calcium plus 800 IU of vitamin D3 daily. Each group consisted of 23 participants.
Study tablets were custom manufactured by Tishcon Corporation, NY, USA. The corresponding supplement and placebo pills were identical in size, appearance, and taste. The placebo was free of calcium, magnesium, vitamin D, and chelating agents.
Calcium carbonate was chosen because it delivers more elemental calcium for a given tablet than other forms, therefore, fewer tablets are required, enhancing adherence; it was the form used in the Calcium and Polyp Prevention adenoma recurrence (23
) and the Calcium and Colorectal Epithelial Cell Proliferation (24
) trials, and in the majority of the larger studies using long term calcium supplementation for other reasons, therefore, its safety record had been well established; and it was the least expensive and most widely available calcium supplement form.
was the chosen form of vitamin D for several reasons, the most important of which was to avoid the toxicity risks associated with 1,25(OH)2
-vitamin D or 25(OH)-vitamin D. Multivitamins and calcium/vitamin D supplements typically provide 400 IU per day of vitamin D3
, but numerous intervention studies show that this dose will not suppress PTH in the overwhelming majority of North American adults (25
). So, we chose a more effective dose of 800 IU per day, which raises serum 25-(OH) vitamin D levels toward the desired range, and leaves a substantial margin of safety, even after taking into consideration dietary intake.
The treatment period was six months to replicate the treatment period of the Calcium and Colorectal Epithelial Cell Proliferation trial (24
) and to ensure approximately 2 – 3 months of 25-OH-vitamin D steady state levels. Participants attended follow-up visits at 2 and 6 months after randomization and were contacted by telephone at monthly intervals between the second and final follow-up visits. At follow-up visits, pill-taking adherence was assessed by questionnaire, interview, and pill count. Participants were instructed to remain on their usual diet and not take any nutritional supplements not in use on entry into the study. At each of the follow-up visits participants were interviewed, filled out questionnaires, and had their vital signs taken. At the first and last visits all participants had their blood drawn and underwent a rectal biopsy procedure. All participants were asked to abstain from aspirin use for seven days prior to each biopsy visit. All visits for a given participant were scheduled at the same time of day to control for possible circadian variability in the outcome measures.
Factors hypothesized to be related to risk for colorectal neoplasms or to the expression of MMR proteins in normal colon mucosa (e.g., diet, medications, etc.) were assessed at baseline, several were reassessed at the first follow-up visit, and all were reassessed at the final follow-up visit. Participants did not have to be fasting for their visits and did not take a bowel cleansing preparation or enema.
Six sextant approximately one millimeter-thick biopsy specimens were taken from normal-appearing rectal mucosa 10 cm proximal to the external anal aperture through a rigid sigmoidocsope with a jumbo cup flexible endoscopic forceps mounted on a semi-flexible rod. No biopsies were taken within 4.0 cm of a polypoid lesion. The biopsies were then immediately placed in phosphate buffered saline and examined and reoriented under a dissecting microscope to ensure that they were not twisted or curled on the bibulous paper. The biopsies were then immediately placed in 10% normal buffered formalin.
The biopsies in formalin were left undisturbed for at least six hours, transferred to 70% ethanol 24 hours after being placed in formalin, embedded in paraffin blocks (two blocks of three biopsies each) within two weeks of the biopsy procedure, cut and stained within another four weeks, and analyzed within another four weeks. From one block, five slides with four section levels each taken 40 microns apart were prepared for each antigen, yielding a total of 20 levels per antigen.
Heat-mediated antigen retrieval was used to break the protein cross-links formed by formalin to uncover the epitope. To accomplish this, slides were placed in a preheated Pretreatment (PT) Module (Lab Vision Corp., CA) with 100× Citrate Buffer pH 6.0 (DAKO S1699, DAKO Corp., Carpinteria, CA; further referred to as DAKO) and steamed for 40 minutes. After antigen retrieval, slides were placed in a DAKO Automated stainer (DAKO) and rinsed with warm PT Module Buffer. The Autostainer was programmed for each immunohistochemistry (IHC) run and the following reagents were used: antibody (MLH1 antibody manufactured by BD Pharmingen, catalog no. 554072, dilution 1:15; or MSH2 antibody manufactured by Calbiochem, catalog no. NA27, dilution 1:500) diluted with Antibody Diluent (DAKO S0809 for MLH1 and S3022 for MSH2, DAKO), LSAB2 Detection System (DAKO K0675, DAKO) for MLH1 and Envision+ Detection System (DAKO K4007, DAKO) for MHS2, diaminobenzidine (DAB) (DAKO K3466 for MLH1 and K3438 for MSH2, DAKO), and TBS buffer (DAKO S1968, DAKO). The slides were not counterstained. After staining, the slides were coverslipped automatically with a Leica CV5000 Coverslipper (Leica Microsystems, Inc., IL) and placed in opaque slide folders. In each staining batch of slides, positive and negative control slides were included. A surgical specimen of normal colon tissue was used as a control tissue for both MMR biomarkers. The control tissue was processed in the same manner as the patient’s tissue, and the negative and the positive control slides were treated identically to the patient’s slides except that antibody diluent was used rather than primary antibody on the negative control slide.
Protocol for Quantifying Staining Density of Immunohistochemically Detected Biomarkers in NormalColon Crypts (“Scoring”)
The imaging and analysis unit was a “hemicrypt”, defined as one side of a colonic crypt bisected from base to colon lumen surface. Intact (at most two contiguous cells missing) hemicrypts extending from the muscularis mucosae to the colon lumen were considered eligible for quantitative image analysis (“scorable”; ). Before analysis, negative and positive control slides were checked for staining adequacy, and the patient’s slides were scanned to assess the adequacy of the biopsy specimen (i.e., whether “scorable” crypts were present).
Figure 1 Quantitative image analysis of MSH2 labeling optical density consists of several steps: a) finding eligible crypts (see text for details); b) manually tracing one side of the crypt (“hemicrypt”); c) automated division of the outline into (more ...)
The major equipment and software for the image analysis procedures were: personal computer, light microscope (Olympus BX40, Olympus Corporation, Japan) with appropriate filters and attached digital light microscope camera (Polaroid DMC Digital Light Microscope Camera, Polaroid Corporation, USA), digital drawing board, ImagePro Plus image analysis software (Media Cybernetics, Inc., MD), our in-house developed plug-in software for colorectal crypt analysis, and Microsoft Access 2003 relational database software (Microsoft Corporation, WA).
The following preparations were performed before starting the scoring program: 1) ensuring standardized settings on the microscope, digital camera, and imaging software; and 2) cleaning and visually scanning the slides. Then, participant ID number, scorer ID, visit number, and antigen, followed by the number of the first biopsy to be scored, whether it had “scorable” crypts, whether it was labeled, and if so, the section level number on the biopsy on which scoring was begun was recorded. Slides were oriented in a standardized fashion and the section levels on the slides were viewed in sequence using light microscopy. All images were taken at 200× magnification and stored as 16-bit grayscale 1,600 × 1,200 pixel images.
For each patient the two biopsies with the greatest number of “scorable” hemicrypts were selected for quantitative image analysis (“scoring”). Intact hemicrypts were “scored” in order from the first section of the first biopsy from left to right. The goal was to score at least 16 “scorable” hemicrypts per biopsy (32 per patient). If the 16th hemicrypt was reached before the level was finished, the scorer continued scoring until either the level was finished or the 20th hemicrypt was scored, whichever came first. No more than 20 hemicrypts per biopsy were scored.
If the two best biopsies had less than 32 “scorable” biopsies, an attempt was made to cut more slides. If that did not solve the issue, scoring was completed if the two best biopsies had 16 or more “scorable” hemicrypts between them. All three biopsies were scored only if there was less than a total of 16 “scorable” hemicrypts between the two best biopsies.
To ensure adherence, a scorer was guided through the scoring protocol by the computer software. For each scored slide background correction images were obtained and controlled for by the computer program. Hemicrypts were manually traced by the scorer (). A traced hemicrypt was divided by the software into segments corresponding in width to that of an average normal crypt epithelial cell. Overall hemicrypt- and segment-specific optical signal densities were then calculated by the software and stored into a Microsoft Access database along with various dimensional parameters of the hemicrypt.
One slide reader analyzed all of the MLH1 and MSH2 stained slides throughout the study. A reliability control sample previously analyzed by the reader was re-analyzed during the course of the trial to determine intra-reader reliability.
Protocol for measuring serum 25-OH-vitamin D and 1,25-(OH)2-vitamin D levels
Laboratory assays for serum 25-OH-vitamin D and 1,25-(OH)2
-vitamin D were done by Dr. Bruce W. Hollis at the Medical University of South Carolina using a RIA method as previously described (27
). Serum samples for baseline and follow-up visits for all subjects were assayed together, ordered randomly, and labeled to mask treatment group, follow-up visit, and quality control replicates. The average intra-assay coefficient of variation for serum 25-OH-vitamin D was 2.3%, and for 1,25-(OH)2
-vitamin D, 6.2%.
Treatment groups were assessed for comparability of characteristics at baseline and at final follow-up by the Fisher’s exact test for categorical variables and analysis of variance (ANOVA) for continuous variables. Slide scoring reliability was analyzed using intra-class correlation coefficients.
Labeling optical densities for MLH1 and MSH2 for each study participant were adjusted for staining batch by dividing each person’s measurement by the mean value for everyone included in the staining batch in which the participant’s sample was run. We decided a priori
to investigate overall (total) crypt expression, expression in the upper 40% (differentiation zone) and lower 60% (proliferation zone) of the crypts, and the ratio of expression in the upper 40% to the full length of the crypts as a measure of within-crypt distribution (distribution index or DI) of the MMR markers (24
Treatment effects were evaluated by assessing differences in mean labeling optical densities from baseline to the 6-month follow-up visit between patients in each active treatment group relative to the placebo group using linear mixed models to account for correlated data. Primary analyses were based on assigned treatment at the time of randomization, regardless of adherence status (intent-to-treat analysis). Two continuous outcomes – MLH1 and MSH2 labeling optical density measurements—were analyzed separately. To provide perspective on the magnitude of the absolute treatment effects ([follow-up – baseline in the active treatment group] – [follow-up – baseline in the placebo group]) of each outcome variable, we also calculated relative effects, defined as: [treatment group follow-up mean/treatment group baseline mean] / [placebo follow-up mean/placebo baseline mean]. The interpretation of the relative effect is somewhat analogous to that of an odds ratio (e.g., a relative effect of 2.0 would mean that the relative proportional change in the treatment group was twice as great as that in the placebo group). No adjustment was made for other covariates in the primary intent-to-treat analyses.
Possible effects of calcium and/or vitamin D on the distribution of MLH1 and MSH2 in rectal crypts were also assessed graphically with the Loess procedure as implemented in SAS version 9 statistical software (31
). First, the number of cells within a hemicrypt was standardized to 50 segments (the average number of cells within a column of colonic crypt cells). Then, average Loess model predicted segment-specific levels of MSH2 for cases and controls by colon site were plotted in the graphs () along with smoothing lines to make graphical evaluation easier.
Expression of MSH2 protein at standardized positions within the crypts of normal-appearing rectal mucosa in four treatment groups.* The Calcium, Vitamin D vs. Markers of Adenomatous Polyps Trial
In sensitivity analyses, we also analyzed data without standardization for batch, by including batch as a covariate, and using different transformations; the results from these analyses did not differ materially from those reported.
Statistical analyses were done using SAS v.9.2 statistical software (Copyrightc 2002–2008 by SAS Institute Inc., Cary, NC, USA). A cutoff level of P ≤ 0.05 (2-sided) was used for assessing statistical significance.