The study population consisted of male veterans scheduled for a screening or diagnostic colonoscopy at the Dom Veterans Affairs Medical Center (DVAMC) in Columbia, South Carolina. Data were collected from March through November 2007 between 9:00 AM and 5:00 PM. Following informed consent, participation included a personal interview and collection of a peripheral blood sample for recovery of serum and DNA. All data were collected prior to completion of the colonoscopy; therefore, participants had no knowledge of the procedure outcome at the time of interview. The questionnaire included information on sociodemographic characteristics, lifestyle, diet, employment and shift work, health status, medications and supplements, mental and physical well-being, physical activity,28
time spent outdoors, risk factors for colorectal polyps and cancer, sleep habits, major life events, stress and coping strategies, social support, and validated instruments targeting sleep disruption (Pittsburgh Sleep Quality Index [PSQI]),29
fatigue (Multidimensional Fatigue Inventory [MFI]),30
and depressive symptoms (Beck Depression Inventory II [BDI]).31,32
The study was approved by the institutional review boards of the DVAMC and University of South Carolina.
Whole blood samples used for DNA recovery were collected in ethylenediaminetetraacetic acid–preserved vacutainers and stored in 0.5-mL aliquots at –80°C prior to analysis. Genomic DNA was extracted using the DrGentle protocol (Takara, Japan). After extraction, genomic DNA pellets (50-100 μg) were dissolved in 100 to 200 μL of TE buffer, of which about 200 ng was subjected to polymerase chain reaction (PCR) using a Perkin Elmer GeneAmp System 9700 (Waltham, MA) according to the manufacturer's protocol. The Per3 variable number tandem repeat (VNTR) DNA sequence was amplified using the following primers (forward) 5′-CAAAATTTTATGACACTACCAGAATGGCTGAC-3′ and (reverse) 5′-AACCTTGTACTTCCACATCAGTGCCTGG-3′, with a reaction mixture consisting of 25 μL standard PCR buffer, 5% DMSO, 1.0 mM MgC12, 0.2 mM dNTP, 1 unit Taq polymerase (Gibco-Invitrogen, Carlsbad, CA), and 0.4 μM of each oligonucleotide primer. The reactions were heated to 94°C for 2 minutes followed by 35 cycles at 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 45 seconds. Finally, the reactions were extended for 7 minutes at 72°C. PCR products were then separated by electrophoresis on a 3% agarose gel to identify homozygous (4-repeat: 4/4), 5-repeat (5/5), or heterozygous (4/5) individuals. Laboratory personnel were blinded to the identity and characteristics of participants, and a 10% random sample was reanalyzed to assess genotyping concordance.
Following collection of a nonfasting venous whole blood sample in a red-top vacutainer, serum samples were clotted at room temperature for 15 to 30 minutes, centrifuged (5000 × g for 5 minutes at 4°C), and 0.5-mL serum aliquots were stored at –80°C until analysis. Serum cytokine concentrations (interferon [IFN]-γ, tumor necrosis factor [TNF]-α, interleukin [IL]-6, IL-1ra, IL-1-β, vascular endothelial growth factor [VEGF]) were determined using a Beadlyte human multicytokine detection system kit (Bio-Rad, Hercules, CA). Filter bottom ELISA plates were rinsed with 100 μL of Bio-plex assay buffer. The assay buffer was then removed using a Millipore Multiscreen Separation Vacuum Manifold System (Bedford, MA) set at 5 mm Hg. Analyte beads in assay buffer were then added to the wells followed by 50 μL of serum or standard solution. The plates were incubated for 30 minutes at room temperature with continuous shaking using a Lab-Line Instrument Titer Plate Shaker (Melrose, IL). The filter bottom plates were washed and centrifuged as before at 300 × g for 30 seconds. Subsequently, 50 μL of antihuman IFN-γ, TNF-α, IL-6, IL-1ra, IL-1-β, or VEGF antibody–biotin reporter solution was added to each reaction well, after which the plates were incubated with continuous shaking for 30 minutes followed by centrifugation and washing. Following this, 50 μL streptavidin–phycoerythrin solution was added, and the plates were incubated with continuous shaking for 10 minutes at room temperature. Next, 125 μL of Bio-plex assay buffer was added, and Beadlyte readings were measured using a Luminex System (Austin, TX) and calculated using Bioplex software. Samples below the limit of detection were assigned a value of one half the limit of detection (0.25) pg/mL for data analyses.
Data analyses were performed using the SAS statistical software package (Version 9.l, SAS Institute Inc, Cary, NC). Values for IL-6, TNF-α, INF-γ, IL-1ra, and IL-l-β were log-transformed prior to analysis, and mean concentrations were transformed back to original units for presentation. VEGF and psychosocial measures (BDI, MFI, and PSQI scores) were analyzed without transformation. One participant had elevated TNF-α and INF-γ concentrations, which were evaluated as potential outliers using the method described by Cook.33
These concentrations met the criteria for Cook's d
and were within a physiological range; thus, the values were retained in the analyses. A 2-step variable selection procedure was used to identify potential confounding factors associated with each cytokine and psychometric (MFI, BDI, PSQI) variable. First, each independent variable obtained from the questionnaire was compared with each dependent variable using the generalized linear models (GLM) procedure in SAS. Second, screened variables (P
< .05) were combined in another GLM model, and factors that were no longer associated with the dependent variable (P
> .10) were removed sequentially from further analysis. This step was repeated until a set of statistically significant (P
< .05) covariates was identified for each dependent variable, and these were included in subsequent hypothesis testing analyses. To test the study hypotheses, the GLM procedure was used to determine whether mean serum cytokine concentrations (IL-6, TNF-α, IFN-γ, IL-1-β, IL-1ra, and VEGF) differed among individuals grouped according to different circadian disruption symptoms (using MFI, BDI, or PSQI scores) or among those with different Per3
genotypes. Adjusted (least squares [LS]) mean cytokine concentrations were calculated, and differences among symptom categories or genotype were determined using the least significant differences (LSD) statistic, after adjustment for selected potentially confounding factors. BDI scores were used to group participants into categories of normal (≥10), mild (11-20), or moderate to severe (≥21) depressive symtoms.32,34
Poor sleepers were defined as those with PSQI scores ≥5,29,35
and MFI cutpoints were chosen to obtain a distribution of 10 to 15 individuals in several fatigue categories. The relationship between cytokine concentrations and circadian disruption symptoms was also evaluated using each score as a continuous variable in separate GLM analyses. The GLM procedure was used to determine whether circadian disruption symptoms were associated with different Per3
variants by comparing adjusted mean MFI, BDI, or PSQI scores among the Per3
genotypes. Inclusion of polyp status (no polyp detected, benign or hyperplastic polyp, adenoma, or unknown—colonoscopy not yet performed), time of day, or month of sample collection in the statistical analyses did not affect the results obtained.