800837322372Neuro Endocrinol LettNeuro Endocrinol. Lett.Neuro endocrinology letters0172-780X235246213655703NIHMS462527ArticleAssociation of the Period3 clock gene length polymorphism with salivary cortisol secretion among police officersWirthMichael12BurchJames124ViolantiJohn5BurchfielCecil6FekedulegnDesta6AndrewMichael6ZhangHongmei2MillerDiane B.7YoungstedtShawn D.34HébertJames R.12VenaJohn E.8Cancer Prevention and Control Program, University of South Carolina, Columbia, SC, USA Department of Epidemiology and Biostatistics, University of South Carolina, Columbia, SC, USA Department of Exercise Science, University of South Carolina, Columbia, SC, USA WJB Dorn VA Medical Center, Columbia, SC, USA Department of Social and Preventive Medicine, School of Public Health and Health Professions, State University of New York at Buffalo, Buffalo, New York, USA Biostatistics and Epidemiology Branch, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, WV, USA Toxicology and Molecular Biology Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, WV, USA Department of Epidemiology and Biostatistics, College of Public Health, University of Georgia, Athens, GA, USA Correspondence to: Michael Wirth, MSPH, PhD., Cancer Prevention and Control Program, University of South Carolina, 915 Greene Street, Suite 200, Columbia, SC, USA. tel/fax: +01 (803) 576-5624; wirthm@mailbox.sc.edu752013201316520133412737© 2013 Neuroendocrinology Letters2013OBJECTIVE

This study evaluated whether measures of waking or diurnal cortisol secretion, or self-reported psychological disturbances differed among police officers with a Period3 (PER3) clock gene length polymorphism.

METHODS

The cortisol awakening response was characterized via the area under the salivary cortisol curve with respect to the increase (AUCI) or total waking cortisol (AUCG). Diurnal cortisol measures included the slope of diurnal cortisol and the diurnal AUCG. Psychological disturbances were characterized using the Center for Epidemiologic Studies Depression Scale, Impact of Events Scale, and Life Events Scale.

RESULTS

Officers with a 4/5 or 5/5 genotype had higher awakening AUCG and greater diurnal cortisol AUCG levels compared to officers with the 4/4 genotype. Among those working more afternoon or night shifts, waking AUCI and AUCG were greater among officers with a 4/5 or 5/5 genotype compared to the 4/4 referents.

CONCLUSION

Cortisol secretion was modified among police officers with different PER3 VNTR clock gene variants.

PER3 VNTRcircadiancortisolshiftworkpolice
INTRODUCTION

Clock genes are part of the body’s intrinsic timekeeping system. Their expression exhibits positive and negative transcription-translation feedback loops that help maintain tissue-specific circadian rhythmicity (Cermakian & Boivin 2009; Yu & Weaver 2011). Clock genes have been described in the body’s central pacemaker, the suprachiasmatic nuclei (SCN) (Fu & Lee 2003; Okamura et al. 2010), and in most peripheral tissues and major organ systems (Hastings et al. 2007). Clock gene expression helps maintain ~24-hour cyclical variation in numerous physiological and cellular processes in a manner that is synchronized with ambient light-dark cycles, and dysregulation of clock genes can impact sleep-wake cycles, cardiovascular, digestive and endocrine systems, and mental state (Hastings et al. 2007; Kripke et al. 2009; Takeda & Maemura 2010; Landgraf et al. 2012). Clock genes also influence cell cycle control (e.g. cell proliferation, DNA repair, apoptosis) (Khapre et al. 2010), and disrupted clock gene expression may increase susceptibility to several chronic diseases including cancer (Huang et al. 2011).

A variable number tandem repeat (VNTR) sequence within the human Period3 (PER3) clock gene (rs57875989) codes for 4 or 5 copies of a 54-base pair length polymorphism. Phenotypic differences in various physiological or psychological measures have been described among those with the 4/4 or 5/5 genotype, Some investigators have reported that the 4/4 genotype may be associated with evening diurnal preference (Archer et al. 2003; Ellis et al. 2009) and greater heroin dependence (Zou et al. 2008), whereas others suggest that the 5/5 or combined 4/5+5/5 genotypes are associated with morning preference (Archer et al. 2003; Ellis et al. 2009), poorer cognitive performance after short-term (40-hr) sleep deprivation (Groeger et al. 2008), higher circulating concentrations of insulin-like growth factor-1 or interleukin-6 (Chu et al. 2008; Guess et al. 2009), earlier age of onset of bipolar disorder (Benedetti et al. 2008), a tendency for depressive symptoms (Guess et al. 2009), or increased odds of breast cancer primarily among premenopausal women (Zhu et al. 2005). However, associations among those with different PER3 VNTR genotypes have not been consistent. For example, several studies have found no association between PER3 VNTR genotypes and diurnal preference or breast cancer (Dai et al. 2010; Barclay et al. 2011; Osland et al. 2011). Individuals with the 5/5 PER3 VNTR had strong correlations between PER3 expression and the timing of peak cortisol or melatonin levels in blood, whereas those with the 4/4 genotype did not (Boivin et al. 2003; Archer et al. 2008). Thus, the PER3 VNTR may serve as a marker of genetic susceptibility to the effects of sleep deprivation or circadian misalignment that can influence the timing of secretion of cortisol or melatonin, two hormones with typically robust circadian rhythms (Dijk & Archer 2010). However, the functional consequences of this polymorphism in terms of adaptation to work schedule or work-related stress are not completely understood, particularly among shiftworkers.

Cortisol is a well-described adrenal steroid “stress hormone” (Henry 1992). Salivary cortisol typically increases by 40-75% upon awakening (cortisol awakening response), and then decreases throughout the day (except for a common post-prandial spike after the mid-day meal) (Kudielka et al. 2006; Fries et al. 2009). Daily cortisol rhythms are controlled by the SCN through efferent connections with neurons of the paraventricular nucleus and through hypothalamic-pituitary-adrenal (HPA) axis-independent alterations in ACTH (adrenocorticotropic hormone) sensitivity in the adrenal cortex (Nader et al. 2010). Stressful circumstances stimulate an increase in cortisol secretion (Kudielka et al. 2006). If these stressful circumstances become chronic in nature, they may result in an inability of the HPA axis to self-regulate cortisol, referred to as the “exhaustion stage” of the general adaptation syndrome, or allostatic overload (Motzer & Hertig 2004). The extent to which diurnal cortisol secretion patterns are influenced by polymorphic variation in clock genes is not well understood (Hastings et al. 2007; Archer et al. 2008; Nader et al. 2010). However, cortisol dysregulation can perturb physiological processes controlling inflammation (Elenkov 2008), and possibly augment susceptibility to depression, post-traumatic stress disorder, cardiovascular disease (CVD), type II diabetes, or stroke (Zuzewicz et al. 2000; Neylan et al. 2005; Huber et al. 2006; Gidron & Ronson 2008; Scheer et al. 2008). An altered pattern of circadian cortisol secretion also is associated with poor cancer survival (Sephton et al. 2000; Sephton et al. 2012).

The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS) cohort study provides a prospective framework for examining biological processes through which stressors associated with police work may mediate adverse health outcomes. The protocol combines the characterization of stress biomarkers, subclinical CVD measures, psychosocial factors, and shiftwork to examine their potential associations with psychological disturbances and chronic diseases afflicting police officers (Violanti et al. 2006; Violanti et al. 2009). Shiftwork can result in circadian rhythm dysregulation, sleep insufficiency, and chronic stress (Shields 2002; Burch et al. 2009), which could eventually lead to cortisol dysregulation and allostatic overload (McEwen & Stellar 1993). Previously, we found that recent night work (within 3-14 days), or a high number of cumulative shift changes over a period of years was associated with reductions in the salivary cortisol awakening response among officers in this cohort (Wirth et al. 2011). The possible implications for cortisol dysregulation, or changes in the human stress response due to PER3 genetic variation among police or other workers in stressful occupations remains to be determined. Our objective in the present analysis was to test the hypotheses that police officers with a 4/5 or 5/5 PER3 genotype have a more robust awakening (waking AUCI or AUCG) or diurnal cortisol rhythm (diurnal AUCG, slope) compared to those with the 4/4 genotype, and whether this relationship may be modified by shiftwork. Potential differences in depressive or stress-related symptoms among those with different PER3 VNTR genotypes were also examined.

METHODSStudy population

Police officers from the Buffalo, New York Police Department enrolled in the BCOPS cohort were selected using a computer-generated random sample (42 females, 58 males) (Violanti et al. 2006; Wirth et al. 2011). The study received Institutional Review Board approval and all subjects provided informed consent. Participants were examined at a health clinic on a scheduled training day or day off, and saliva collection occurred the day after the clinic visit at the officers’ homes. Data collection included a peripheral white blood cell sample for DNA recovery, serial saliva collection over a single non-work day, long-term shiftwork history, basic demographic characteristics, and the completion of several validated instruments that ascertained stress related to traumatic events (Impact of Events Scale or IES) (Horowitz et al. 1979), significant life events (Life Events Scale) (Paykel et al. 1971), or depressive symptoms (Center for Epidemiologic Studies Depression scale or CES-D) (Radloff 1977). A majority of the participating officers (88%) were either working a day shift or had a day off prior to their clinic visit.

Shiftwork history

Daily work histories were obtained for each participant from 1994 or initiation of employment (if it occurred after 1994) to the date of study examination between 2001 and 2003 using electronic payroll records (Wirth et al. 2011). The typical work schedule after 1994 consisted of four work days, four days off, four work days, three days off, and then the cycle was repeated. Long-term shiftwork variables included the cumulative number of shift changes and a categorical shift status variable defined as the shift on which each participant spent the majority of her or his time working during the study period. Shifts were classified based on start times as day (between 04:00-11:59 h), afternoon (between 12:00-19:59 h), or night (between 20:00-03:59 h). Most subjects (>85%) spent a majority (≥70%) of their total work time on their designated shift. This process of classifying officers into a shift status has shown good consistency summarized over 30, 60, or 90 days, and after 5 years of employment (Violanti et al. 2009; Wirth et al. 2011). Those on night or afternoon shifts were combined for the stratified analyses. Although officers worked fixed shifts during the study period (i.e., day, afternoon, or night only), they occasionally worked for an absent colleague or were temporarily assigned to a different shift schedule. The frequency of shift changes was defined as the number of times a participant switched between any two shift types during the study period.

Salivary cortisol

Salivary cortisol measurements are noninvasive, thereby reducing participant burden and facilitating protocol compliance; and the biologically active hormone can be readily quantified via a sensitive and specific immunoassay (Violanti et al. 2009). Participants collected serial saliva samples: upon first awakening, then at 15-, 30-, and 45-minute intervals after waking and at 12:00h (before lunch), 17:00h (before dinner), and at bedtime (Wirth et al. 2011). Cortisol levels obtained based on collection from time of awakening have a higher test-retest stability compared to samples collected at specific clock times (Coste et al. 1994; Pruessner et al. 1997; Neylan et al. 2005). On the morning of participation, sample collection was achieved by placing a Salivette cotton roll (Sarstedt Aktiengesellschaft & Company, Numbrecht, Germany) into the mouth for three minutes to allow for saturation. The saturated roll was then refrigerated until delivery at the research laboratory for processing. Samples were shipped to the National Institute for Occupational Safety and Health (Toxicology and Molecular Biology Branch, Health Effects Laboratory, Morgantown, WV) where they were centrifuged and archived at −20 °C (Fekedulegn et al. 2007). Cortisol determinations were performed at the Technical University of Dresden, Dresden, Germany, using a chemiluminescence immunoassay (CLIA, IBL-Hamburg, Germany) (Fekedulegn et al. 2007; Violanti et al. 2009; Wirth et al. 2011). Quality control samples were quantified at low and high concentrations for each assay plate, and analyses were repeated if control samples were outside the range of the expected concentration. The intra- and inter-assay coefficients of variation were 8% or below for either the high (25 nmol/l) or low (3 nmol/l) control concentrations. Blind replicate samples of participants’ salivary cortisol (10%) had a coefficient of variation of 15%.

Waking cortisol was summarized using the area under the curve with respect to increase above baseline (i.e. first waking sample), and AUC above the assay detection limit or ground concentration (Fekedulegn et al. 2007). The AUCI represents the change in cortisol secretion after awakening, or its reactivity (Fekedulegn et al. 2007). If cortisol levels decrease relative to the first waking value, it is possible to obtain a negative AUCI value. The AUCG measures the total amount of cortisol secreted during the sampling period (Fekedulegn et al. 2007). The diurnal slope represents the change in cortisol secretion across the day, which was estimated by fitting the initial waking, noon, dinner, and bed time salivary cortisol sample concentrations to a linear equation and estimating the line of best fit (Kraemer et al. 2006; Heaney et al. 2012). All cortisol time points were required to calculate diurnal AUCG measures, but this requirement was not true for diurnal slope. To maintain consistency, we restricted all main diurnal analyses only to participants with both diurnal AUCG and diurnal cortisol slope measures.

Genotyping

Peripheral blood samples were collected on the day of the clinic visit, centrifuged using a Ficoll gradient to separate WBCs, then stored in capillary tubes at −80 °C for recovery of DNA. Genomic DNA was extracted using the DrGentle protocol (Takara, Japan) and DNA pellets (50-100 μg) were dissolved in 100-200 μL of TE buffer. About 200ng was subjected to polymerase chain reaction (PCR) using a Perkin Elmer GeneAmp System 9700 (Waltham, MA) according to the manufacturer’s protocol. The PER3 VNTR repeat polymorphism was amplified using the following two primers: (forward) 5′-CAAAATTTTATGACACTACCAGAATGGCTGAC-3′ and (reverse) 5′-AACCTTGTACTTCCACATCAGTGCCTGG-3′ (Zhu et al. 2005). The PCR was performed in a reaction mixture of 25 μl containing standard PCR buffer, 5% DMSO, 1.0 mM MgCl2, 0.2 mM dNTP, 1 unit Taq polymerase (Gibco-Invitrogen), and 0.4 μM of each oligonucleotide primer. The reactions were heated to 94 °C for 2 minutes followed by 35 cycles of 94 °C for 30 seconds, 60 °C for 30 seconds, and 72 °C for 5 seconds. Reactions were extended for 7 minutes at 72 °C, and PCR products were then separated by electrophoresis on 3% agarose gel. Laboratory personnel were blinded to the identity and characteristics of the participants. Quality control re-analyses of 10% of the genotypes indicated 100% concordance.

Statistical analysis

Analyses were performed using SAS analytical software package (version 9.2, Cary, NC)®. Relationships between each dependent variable (waking AUCI, waking AUCG, diurnal AUCG, diurnal slope, IES, CES-D, and Life Events Scale) and potential confounding factors (i.e. age, gender, race, education, marital status, rank, and years of police work) were evaluated univariately using the generalized linear models (PROC GLM) procedure in SAS. Variables were selected for further evaluation as potential confounders if their statistical significance was p≤0.15. A backward elimination procedure was then used to develop final models that included all variables that were statistically significant (p≤0.05) or, when removed from the model, changed the beta coefficient of the PER3 VNTR genotype by at least 10%. One diurnal AUCG observation was removed due to a studentized residual of 4.02 and a Cook’s D of 0.48, which is greater than the suggested cut-point (4/sample size included in analysis [n=54] or 0.07). The GLM procedure in SAS was used to compute adjusted (least squares or LS) means of each dependant variable among those with different PER3 VNTR genotypes, after adjustment for the selected covariates. A square root transformation was used to obtain normally distributed values and normalized model residuals of the Life Events and CES-D scores; the LS means were back-transformed for presentation in the tables. A priori comparisons included differences in mean dependent variables among the 4/5, 5/5 or combined 4/5+5/5 genotypes compared to the 4/4 genotype. In separate analyses, differences in LS mean cortisol measures among the PER3 genotypes were stratified by shift status (night+afternoon vs. day shifts) or cumulative shift changes (a median split of <17 vs. >17). Individuals with 4/5 or 5/5 genotypes were similar with respect to all covariate and exposure variables, and were therefore combined in stratified analyses (Zhu et al. 2005; Dai et al. 2010). Ancillary logistic regression analyses indicated that participants with missing data did not differ with respect to the PER3 VNTR, shiftwork, CES-D, IES, Life Events Scale, or any covariate data, and were thus considered missing at random. Additional adjustment for time of awakening (first saliva collection) did not alter the interpretation of the results presented below.

RESULTS

Complete waking cortisol data were available for 57 officers (32 missing waking cortisol samples, 11 missing PER3 VNTR) and diurnal cortisol data were available for 54 participants (37 missing diurnal cortisol, 9 missing PER3 VNTR). The distribution of PER3 VNTR genotypes among participants was in Hardy-Weinberg Equilibrium (χ2=0, p-value=1.0). The average number of years of police work in this population (± standard deviation) was 14±9 years (range: 1-33 years). The mean age was 43±8 years (range: 29-63 years). Males comprised 60% of the study group and European Americans 75%. There were no statistically significant differences in age, gender, race, education, marital status, rank, years worked, or cumulative shift changes among the 4/5, 5/5, or 4/5+5/5 genotypes compared to the 4/4 group (Table 1). Similarly, there we no statistically significant differences in the mean time of first saliva sample collection among participants with the 4/4, 4/5 or 5/5 genotypes (08:16±112, 07:45±94, and 07:12±129 minutes, respectively) on the day of study participation.

Mean waking cortisol AUCG levels were greater among those with the 4/5+5/5 genotype (775 vs. 448 nmol/L-minute, p<0.01) compared to the 4/4 group. The 4/5 and 5/5 groups showed a tendency for higher waking AUCI values compared to the 4/4 group, although the differences were not statistically significant. The 4/5+5/5 group (7201 vs. 4996 nmol/L-minute, p<0.01) and those with the 5/5 genotype (7279 vs. 4996 nmol/L-minute, p=0.02) had greater mean diurnal AUCG values compared to the 4/4 group (Table 2). Note that there were no differences between the 5/5 and 4/5 groups for any of the cortisol measures evaluated, which provides a reasonable rationale for combining 4/5 and 5/5 groups in the analysis. Only those with both a diurnal AUCG value and a diurnal slope were included in the analyses. A post-hoc analysis of all individuals with a diurnal slope value (n=76) did not change the interpretation of the results.

Regardless of shiftwork status, officers with the 4/5 or 5/5 genotype had greater mean diurnal AUCG values than officers with a 4/4 genotype (Table 3). After stratification by shift status, those participating in afternoon or night shiftwork who also possessed the 4/5+5/5 genotype had elevated mean waking AUCI (202 vs. −8 nmol/L-minute, respectively, p=0.03) and mean waking AUCG values (791 vs. 361 nmol/L-minute, respectively, p=0.01) relative to shiftworkers with the 4/4 genotype (Table 3). However, these values did not differ by genotype among day workers. When cumulative shift changes were examined, officers with the 4/5 or 5/5 genotype tended to have greater mean waking AUCI, waking AUCG, or diurnal AUCG values compared to those with the 4/4 genotype, regardless of whether they were above or below the median number of cumulative shift changes (Table 4). There were no statistically significant differences in the mean diurnal cortisol slopes (Tables 2-4) or mean scores for IES, Life Events Scale, or CES-D among those with different PER3 VNTR genotypes (Table 2).

DISCUSSION

The role of the PER3 length polymorphism in the regulation of sleep and circadian processes in human populations has not been fully elucidated. The extra copy of the 5-repeat PER3 VNTR sequence contains several potential casein kinase Iε (CKIε) phosphorylation motifs (Archer et al. 2003). Phosphorylation of Period clock genes by CKIε is required for translocation of the period and cryptochrome protein complex into the cell nucleus so that it can exert its influence on the negative arm of the clock gene transcriptional-translational feedback loop. CKIε also facilitates metabolic degradation of this complex (Nader et al. 2010). In the PERIOD 2 clock gene, a CKIε binding site mutation has been associated with gene hypophosporylation and familial advanced sleep phase syndrome (Toh et al. 2001). Polymorphic variation in the PER3 gene has been associated with differences in the homeostatic regulation of sleep and the timing of circadian hormone secretion (Archer et al. 2008; Dijk & Archer 2010). When participants with different PER3 VNTR genotypes were subjected to a 40-hour sleep deprivation protocol, those with the 5/5 genotype experienced greater changes in EEG activity and REM sleep, and increased sleep pressure compared to those with the 4/4 genotype (Cajochen et al. 1995; Dijk et al. 1997; Viola et al. 2007). Individuals with this genotype also performed poorly (relative to baseline) on a waking performance test after sleep deprivation, whereas impacts among those with the 4/4 genotype were much less apparent (Viola et al. 2007). Other studies suggest that PER3 may only have a minimal role in regulating circadian processes (Shearman et al. 2000; Bae et al. 2001; Costa et al. 2011), and that the influence of the PER3 VNTR on sleep homeostasis may vary depending on the duration of sleep deprivation, or the strategy used to adjust to sleep loss (Goel et al. 2009; Gamble et al. 2011). Whether the PER3 VNTR evokes differences in circadian endocrine secretion, for example among shift workers or those with altered sleep-wake timing, remains to be determined. Positive correlations between PER3 expression and the timing of peak melatonin or cortisol secretion (but not amplitude) were previously observed among those with the 5/5 genotype, whereas these measures were not well correlated among subjects with the 4/4 genotype (Archer et al. 2008) . Although no cause-effect relationship was established in that study, the results suggest that the 5/5 variant may facilitate coupling of cortisol secretion to the circadian system through PER3 expression. In addition to its influence on the timing of cortisol secretion (Archer et al. 2008), the PER3 VNTR has been associated with adverse health outcomes that might overlap with cortisol dysregulation, including delayed sleep phase syndrome (Ebisawa et al. 2001), bipolar disorder (Dallaspezia et al. 2011), and increased cancer risk (Zhu et al. 2009; Dai et al. 2010). However, the functional consequences of this relationship remain to be determined.

Salivary cortisol values among officers participating in the present study were generally representative of those observed in other populations, with higher levels in the morning and declining values throughout the day (Kudielka et al. 2007; Griefahn & Robens 2008). Consistent with the hypothesis that the 5/5 genotype may be linked with circadian cortisol secretion, officers with a 4/5 or 5/5 genotype had an adjusted mean waking AUCI that was 106% greater, a waking mean AUCG that was 73% greater, and mean cortisol output across the day that was 44% greater than those with a 4/4 genotype (Table 2). The reason that no difference in the mean diurnal cortisol slope was observed among those with different PER3 genotypes is uncertain, but suggests that the PER3 VNTR had a stronger influence on morning cortisol secretion than secretion occurring throughout the day. The primary mechanism for cortisol’s robust circadian rhythm may be an increase in morning secretion due to enhanced light sensitivity occurring at that time of day (Clow et al. 2004). If so, then individuals with a morning circadian preference who typically exhibit earlier wake times would be expected to have an increased cortisol secretion due to an increased probability of elevated light exposure after awakening. This is consistent with studies that observed elevated cortisol levels among morning types compared to evening types (Bailey & Heitkemper 2001; Kudielka et al. 2006). Because the 5/5 PER3 genotype tends to be more frequently associated with morningness (Archer et al. 2003; Ellis et al. 2009), our findings are consistent with these observations. However, diurnal preference has not always been associated with the PER3 VNTR (Barclay et al. 2011; Osland et al. 2011), and chronotype was not characterized in this study. Nonetheless, we found that collection of the first saliva sample occurred about one hour earlier, on average, among those with the 5/5 genotype compared to the 4/4 genotype (08:16±112, 07:45±94, and 07:12±129 minutes for the 4/4, 4/5 and 5/5 genotypes, respectively). Although these differences were not statistically significant, it is possible that they contributed to the salivary cortisol measures that were observed.

The SCN synchronizes the circadian rhythms of ACTH and cortisol secretion, as well as clock gene expression within the pituitary gland and adrenal cortex (Dickmeis 2009; Girotti et al. 2009; Nader et al. 2010; Tonsfeldt & Chappell 2012). There are currently several models describing pathways whereby light-entrained SCN activity, in conjunction with neural and endocrine effectors of the HPA axis, regulate diurnal cortisol secretion. Although, the role of PER3 in these processes remains to be determined, one possibility is that PER3 may influence sensitivity of the circadian system to ambient light exposure. A genetic predisposition to light-induced suppression of melatonin, another hormone with a robust circadian rhythm, was recently observed among individuals with the 5/5 PER3 VNTR genotype, whereas 4/4 homozygotes were less responsive (Chellappa et al. 2012). Based on these observations, we speculate that light-induced morning cortisol secretion may occur in a similar, PER3 genotype-dependent manner.

Examination of cortisol and psychometric measures in conjunction with the PER3 VNTR among police officers in this study provided an opportunity to evaluate the influence of genotype and shiftwork on these parameters in a real-world setting. A strength of the BCOPS cohort is that long term shiftwork histories were quantified among participants via reconstruction of payroll records. We previously reported that officers working short-term night or afternoon shifts (3-14 days prior to saliva collection) had reduced waking cortisol AUCI or AUCG compared to day workers, consistent with several other studies among shiftworkers (Zuzewicz et al. 2000; Kudielka et al. 2007; Griefahn & Robens 2008). In addition, officers with more cumulative shift changes over periods of years also had reduced waking AUCI values (Wirth et al. 2011). Thus, we hypothesized that shiftwork may modify the relationship between the PER3 VNTR and cortisol secretion. When stratified by shift status, those with a 4/5 or 5/5 genotype who were working night or afternoon shifts had the highest waking AUCI values, more than double what was observed among day workers with the 4/5 or 5/5 genotype, which suggests a possible gene-environment interaction. Caution in the interpretation of these findings is warranted given the relatively limited sample size among strata of shiftwork and genotype. Also, results obtained for waking or diurnal AUCG indicated that, regardless of shift status, those with a 4/5 or 5/5 genotype had cortisol secretion patterns that were elevated compared to those with the 4/4 genotype. Overall, the results support the possibility that an extra PER3 VNTR copy enhances cortisol secretion. If this effect is modified by shiftwork, it most likely influences the absolute increase in cortisol after awakening (AUCI) rather than the total amount secreted after awakening (AUCG), or cortisol secretion throughout the day.

In the present study, the PER3 VNTR was not associated with stress-related psychological symptoms including depression or life events, in contrast with previous studies (Guess et al. 2009; Dallaspezia et al. 2011). Only about 6% of officers were depressed based on the CES-D definition (score ≥16), thus there may not have been enough variation in CES-D scores to determine differences between the PER3 VNTR genotypes. Stress or depressive symptoms were also a potential source of bias related to cortisol in this study since these symptoms can be associated with both cortisol secretion and the PER3 genotype (Chida & Steptoe 2009). However, there was no change in the interpretation of the results when the cortisol analyses were adjusted for the effects of the Life Events Scale, CES-D, or IES. Although we adjusted for these and other important confounding factors, information on other covariates was unavailable, for example ambient light exposures (Sephton & Spiegel 2003; Clow et al. 2004). Similarly, poor sleep or diets high in fat and low in fruit and vegetable intake may be associated with disrupted or flattened diurnal cortisol slopes (Kumari et al. 2011; Heaney et al. 2012). Thus, the possibility of residual confounding by these or other factors cannot be entirely eliminated.

In conclusion, the cross-sectional nature of this study precludes the ability to infer causation, although the results suggest that the 5-repeat sequence of the PER3 VNTR may facilitate increased cortisol secretion, particularly in the morning. Individuals with this genotype may be more susceptible to factors that can cause circadian rhythm disruption, such as shiftwork, poorly timed light exposures, or changes in sleep-wake timing. However, only modest evidence for a PER3-related influence of shiftwork on cortisol secretion was obtained in the present study. Cortisol dysregulation may have long-term health implications. Reduced cortisol secretion or a flattened slope has been associated with poor sleep quality (Backhaus et al. 2004), chronic fatigue syndrome and symptoms of burnout (Roberts et al. 2004), PTSD (Rohleder et al. 2004), depression (Stetler & Miller 2005), adverse cardiovascular health and mortality (Hurwitz Eller et al. 2001; Kumari et al. 2011), increased all-cause mortality (Kumari et al. 2011), and decreased breast or lung cancer survival (Sephton et al. 2000; Sephton et al. 2012). Although linkages between circadian clock gene expression and the HPA axis have been identified, the pathological implications of dysregulation of these processes await further characterization (Nader et al. 2010; Mavroudis et al. 2012).

ACKNOWLEDGEMENTS

This project was supported by Contract No. 212-2008-M-24052 from the Biostatistics and Epidemiology Branch of the National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, West Virginia. Michael Wirth’s participation in this research was supported by the University of South Carolina’s Behavioral-Biomedical Interface Program which is funded in part by training grant T32-5R18CE001240 from the National Institute of General Medical Sciences. Dr. Hébert is supported by an Established Investigator Award in Cancer Prevention and Control from the Cancer Training Branch of the National Cancer Institute (K05 CA136975). The findings and conclusions in this report are those of the author(s) and do not necessarily represent the views of the National Institute for Occupational Safety and Health.

Declaration of Interest

The authors report no conflict of interest.

AbbreviationsSCN

suprachiasmatic nuclei

VNTR

variable number tandem repeat

PER3

Period 3 gene

HPA

hypothalamic-pituitary-adrenal

ACTH

adrenocorticotropic hormone

CVD

cardiovascular disease

BCOPS

Buffalo Cardio-Metabolic Occupational Police Stress

AUCI

area under the curve with respect to increase

AUCG

area under the curve with respect to ground

CES-D

Center for Epidemiologic Studies Depression scale

IES

Impact of Events Scale

LS

least squares

CKIε

casein kinase I epsilon

EEG

electroencephalography

REM

rapid eye movement

PTSD

post-traumatic stress disorder

REFERENCESArcherSNRobilliardDLSkeneDJSmitsMWilliamsAArendtJA length polymorphism in the circadian clock gene Per3 is linked to delayed sleep phase syndrome and extreme diurnal preferenceSleep200326413512841365ArcherSNViolaAUKyriakopoulouVvon SchantzMDijkDJInter-individual differences in habitual sleep timing and entrained phase of endogenous circadian rhythms of BMAL1, PER2 and PER3 mRNA in human leukocytesSleep2008316081718517031BackhausJJunghannsKHohagenFSleep disturbances are correlated with decreased morning awakening salivary cortisolPsychoneuroendocrinology20042911849115219642BaeKJinXMaywoodESHastingsMHReppertSMWeaverDRDifferential functions of mPer1, mPer2, and mPer3 in the SCN circadian clockNeuron2001305253611395012BaileySLHeitkemperMMCircadian rhythmicity of cortisol and body temperature: morningness-eveningness effectsChronobiol Int2001182496111379665BarclayNLEleyTCMillJWongCCZavosHMArcherSNSleep quality and diurnal preference in a sample of young adults: associations with 5HTTLPR, PER3, and CLOCK 3111Am J Med Genet B Neuropsychiatr Genet2011156B6819021714069BenedettiFDallaspeziaSColomboCPirovanoAMarinoESmeraldiEA length polymorphism in the circadian clock gene Per3 influences age at onset of bipolar disorderNeurosci Lett2008445184718789374BoivinDBJamesFOWuACho-ParkPFXiongHSunZSCircadian clock genes oscillate in human peripheral blood mononuclear cellsBlood20031024143512893774BurchJBTomJZhaiYCriswellLLeoEOgoussanKShiftwork impacts and adaptation among health care workersOccup Med (Lond)2009591596619270043CajochenCBrunnerDPKrauchiKGrawPWirz-JusticeAPower density in theta/alpha frequencies of the waking EEG progressively increases during sustained wakefulnessSleep19951889048746397CermakianNBoivinDBThe regulation of central and peripheral circadian clocks in humansObes Rev 10 Suppl200922536ChellappaSLViolaAUSchmidtCBachmannVGabelVMaireMHuman melatonin and alerting response to blue-enriched light depend on a polymorphism in the clock gene PER3J Clin Endocrinol Metab201297E433722188742ChidaYSteptoeACortisol awakening response and psychosocial factors: a systematic review and meta-analysisBiol Psychol2009802657819022335ChuLWZhuYYuKZhengTChokkalingamAPStanczykFZCorrelation between circadian gene variants and serum levels of sex steroids and insulin-like growth factor-ICancer Epidemiol Biomarkers Prev20081732687318990770ClowAThornLEvansPHucklebridgeFThe awakening cortisol response: methodological issues and significanceStress20047293715204030CostaMJSoAYKaasikKKruegerKCPillsburyMLFuYHCircadian rhythm gene period 3 is an inhibitor of the adipocyte cell fateJ Biol Chem201128690637021228270CosteJStrauchGLetraitMBertagnaXReliability of hormonal levels for assessing the hypothalamic-pituitary-adrenocortical system in clinical pharmacologyBr J Clin Pharmacol19943847497893592DaiHZhangLCaoMSongFZhengHZhuXThe role of polymorphisms in circadian pathway genes in breast tumorigenesisBreast Cancer Res Treat20101275314020978934DallaspeziaSLorenziCPirovanoAColomboCSmeraldiEBenedettiFCircadian clock gene Per3 variants influence the postpartum onset of bipolar disorderEur Psychiatry2011261384021316201DickmeisTGlucocorticoids and the circadian clockJ Endocrinol200920032218971218DijkDJArcherSNPERIOD3, circadian phenotypes, and sleep homeostasisSleep Med Rev2010141516019716732DijkDJShanahanTLDuffyJFRondaJMCzeislerCAVariation of electroencephalographic activity during non-rapid eye movement and rapid eye movement sleep with phase of circadian melatonin rhythm in humansJ Physiol 505 ( Pt199738518EbisawaTUchiyamaMKajimuraNMishimaKKameiYKatohMAssociation of structural polymorphisms in the human period3 gene with delayed sleep phase syndromeEMBO reports20012342611306557ElenkovIJNeurohormonal-cytokine interactions: implications for inflammation, common human diseases and wellbeingNeurochem Int200852405117716784EllisJvon SchantzMJonesKHArcherSNAssociation between specific diurnal preference questionnaire items and PER3 VNTR genotypeChronobiol Int2009264647319360490FekedulegnDBAndrewMEBurchfielCMViolantiJMHartleyTACharlesLEArea under the curve and other summary indicators of repeated waking cortisol measurementsPsychosom Med200769651917766693FriesEDettenbornLKirschbaumCThe cortisol awakening response (CAR): facts and future directionsInt J Psychophysiol200972677318854200FuLLeeCCThe circadian clock: pacemaker and tumour suppressorNat Rev Cancer200333506112724733GambleKLMotsinger-ReifAAHidaABorsettiHMServickSVCiarleglioCMShift work in nurses: contribution of phenotypes and genotypes to adaptationPLoS One20116e1839521533241GidronYRonsonAPsychosocial factors, biological mediators, and cancer prognosis: a new look at an old storyCurr Opin Oncol20082038639218525332GirottiMWeinbergMSSpencerRLDiurnal expression of functional and clock-related genes throughout the rat HPA axis: system-wide shifts in response to a restricted feeding scheduleAm J Physiol Endocrinol Metab2009296E8889719190255GoelNBanksSMignotEDingesDFPER3 polymorphism predicts cumulative sleep homeostatic but not neurobehavioral changes to chronic partial sleep deprivationPLoS One20094e587419516903GriefahnBRobensSThe cortisol awakening response: a pilot study on the effects of shift work, morningness and sleep durationPsychoneuroendocrinology200833981818650024GroegerJAViolaAULoJCvon SchantzMArcherSNDijkDJEarly morning executive functioning during sleep deprivation is compromised by a PERIOD3 polymorphismSleep20083111596718714788GuessJBurchJBOgoussanKArmsteadCAZhangHWagnerSCircadian disruption, Per3, and human cytokine secretionIntegr Cancer Ther200983293619926609HastingsMO’NeillJSMaywoodESCircadian clocks: regulators of endocrine and metabolic rhythmsJ Endocrinol20071951879817951531HeaneyJLPhillipsACCarrollDAging, health behaviors, and the diurnal rhythm and awakening response of salivary cortisolExp Aging Res20123829531422540384HenryJPBiological basis of the stress responseIntegr Physiol Behav Sci19922766831576090HorowitzMWilnerMAlvarezWImpact of event scale: a measure of subjective stressPsychosomatic Med197941209218HuangXLFuCJBuRFRole of circadian clocks in the development and therapeutics of cancerJ Int Med Res2011392061622289520HuberTJIssaKSchikGWolfOTThe cortisol awakening response is blunted in psychotherapy inpatients suffering from depressionPsychoneuroendocrinology200631900416707227Hurwitz EllerNNetterstromBHansenAMCortisol in urine and saliva: relations to the intima media thickness, IMTAtherosclerosis20011591758511689219KhapreRVSamsaWEKondratovRVCircadian regulation of cell cycle: Molecular connections between aging and the circadian clockAnn Med2010424041520568980KraemerHCGiese-DavisJYutsisMO’HaraRNeriEGallagher-ThompsonDDesign decisions to optimize reliability of daytime cortisol slopes in an older populationAm J Geriatr Psychiatry2006143253316582041KripkeDFNievergeltCMJooEShekhtmanTKelsoeJRCircadian polymorphisms associated with affective disordersJ Circadian Rhythms20097219166596KudielkaBMBuchtalJUhdeAWustSCircadian cortisol profiles and psychological self-reports in shift workers with and without recent change in the shift rotation systemBiol Psychol2007749210317101207KudielkaBMFederenkoISHellhammerDHWustSMorningness and eveningness: the free cortisol rise after awakening in “early birds” and “night owls”Biol Psychol200672141616236420KumariMShipleyMStaffordMKivimakiMAssociation of diurnal patterns in salivary cortisol with all-cause and cardiovascular mortality: findings from the Whitehall II studyJ Clin Endocrinol Metab20119614788521346074LandgrafDShostakAOsterHClock genes and sleepPflugers Arch201246331421833490MavroudisPDScheffJDCalvanoSELowrySFAndroulakisIPEntrainment of peripheral clock genes by cortisolPhysiol Genom20124460721McEwenBSStellarEStress and the Individual – Mechanisms Leading to DiseaseArch Intern Med1993153209321018379800MotzerSAHertigVStress, stress response, and healthNurs Clin North Am20043911715062724NaderNChrousosGPKinoTInteractions of the circadian CLOCK system and the HPA axisTrends Endocrinol Metab2010212778620106676NeylanTCBrunetAPoleNBestSRMetzlerTJYehudaRPTSD symptoms predict waking salivary cortisol levels in police officersPsychoneuroendocrinology2005303738115694117OkamuraHDoiMFustinJMYamaguchiYMatsuoMMammalian circadian clock system: Molecular mechanisms for pharmaceutical and medical sciencesAdv Drug Deliv Rev2010628768420620185OslandTMBjorvatnBRSteenVMPallesenSAssociation study of a variable-number tandem repeat polymorphism in the clock gene PERIOD3 and chronotype in Norwegian university studentsChronobiol Int2011287647021919721PaykelESPrusoffBAUhlenhuthEHScaling of life eventsArch Gen Psychiatry19712534075116988PruessnerJCWolfOTHellhammerDHBuske-KirschbaumAvon AuerKJobstSFree cortisol levels after awakening: a reliable biological marker for the assessment of adrenocortical activityLife Sci1997612539499416776RadloffLSThe CES-D Scale: A Self-Report Depression Scale for Research in the General PopulationsAppl Psychol Meas19771385401RobertsADWesselySChalderTPapadopoulosACleareAJSalivary cortisol response to awakening in chronic fatigue syndromeBr J Psychiatry20041841364114754825RohlederNJoksimovicLWolfJMKirschbaumCHypocortisolism and increased glucocorticoid sensitivity of pro-Inflammatory cytokine production in Bosnian war refugees with posttraumatic stress disorderBiol Psychiatry2004557455115039004ScheerFASheaTJHiltonMFSheaSAAn endogenous circadian rhythm in sleep inertia results in greatest cognitive impairment upon awakening during the biological nightJ Biol Rhythms2008233536118663242SephtonSSpiegelDCircadian disruption in cancer: a neuroendocrine-immune pathway from stress to disease?Brain Behav Immun200317321812946654SephtonSELushEDedertEAFloydARRebholzWNDhabharFSDiurnal cortisol rhythm as a predictor of lung cancer survivalBrain Behav Immun2012SephtonSESapolskyRMKraemerHCSpiegelDDiurnal cortisol rhythm as a predictor of breast cancer survivalJ Natl Cancer Inst200092994100010861311ShearmanLPJinXLeeCReppertSMWeaverDRTargeted disruption of the mPer3 gene: subtle effects on circadian clock functionMol Cell Biol20002062697510938103ShieldsMShift work and healthHealth Rep200213113315069802StetlerCMillerGEBlunted cortisol response to awakening in mild to moderate depression: regulatory influences of sleep patterns and social contactsJ Abnorm Psychol200511469770516351390TakedaNMaemuraKCircadian clock and vascular diseaseHypertens Res2010336455120448639TohKLJonesCRHeYEideEJHinzWAVirshupDMAn hPer2 phosphorylation site mutation in familial advanced sleep phase syndromeScience20012911040311232563TonsfeldtKJChappellPEClocks on top: the role of the circadian clock in the hypothalamic and pituitary regulation of endocrine physiologyMol Cell Endocrinol201234931221787834ViolaAUArcherSNJamesLMGroegerJALoJCSkeneDJPER3 polymorphism predicts sleep structure and waking performanceCurr Biol200717613817346965ViolantiJMBurchfielCMFekedulegnDAndrewMEDornJHartleyTACortisol patterns and brachial artery reactivity in a high stress environmentPsychiatry Res2009169758119616310ViolantiJMBurchfielCMHartleyTAMnatsakanovaAFekedulegnDAndrewMEAtypical work hours and metabolic syndrome among police officersArch Environ Occup Health20096419420119864222ViolantiJMBurchfielCMMillerDBAndrewMEDornJWactawski-WendeJThe Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS) pilot study: methods and participant characteristicsAnn Epidemiol2006161485616165369WirthMBurchJViolantiJBurchfielCFekedulegnDAndrewMShiftwork duration and the awakening cortisol response among police officersChronobiol Int2011284465721721860YuEAWeaverDRDisrupting the circadian clock: genespecific effects on aging, cancer, and other phenotypesAging (Albany NY)201134799321566258ZhuYBrownHNZhangYStevensRGZhengTPeriod3 structural variation: a circadian biomarker associated with breast cancer in young womenCancer Epidemiol Biomarkers Prev2005142687015668506ZhuYStevensRGHoffmanAEFitzgeraldLMKwonEMOstranderEATesting the circadian gene hypothesis in prostate cancer: a population-based case-control studyCancer Res20096993152219934327ZouYLiaoGLiuYWangYYangZLinYAssociation of the 54-nucleotide repeat polymorphism of hPer3 with heroin dependence in Han Chinese populationGene Brain Behav200872630ZuzewiczKKwareckiKWaterhouseJCircadian rhythm of heart rate, urinary cortisol excretion, and sleep in civil air traffic controllersInt J Occup Saf Ergon200063839211082552

Population Characteristics by PER3 Variable Number Tandem Repeat, BCOPS Study, Buffalo, NY, USA, 2001–2003.

CharacteristicTotal(n=57)4/4(n = 19)4/5(n = 28)5/5(n = 10)4/5 + 5/5(n = 38)
Age Group
 <4020 (35%)6 (32%)12 (43%)2 (20%)14 (37%)
 40-4924 (42%)7 (37%)11 (39%)6 (60%)17 (45%)
 >5013 (23%)6 (32%)5 (18%)2 (20%)7 (18%)
Gender
 Male34 (60%)14 (74%)14 (50%)6 (60%)20 (53%)
 Female23 (40%)5 (26%)14 (50%)4 (40%)18 (47%)
Race
 European American43 (75%)14 (74%)22 (79%)7 (70%)29 (76%)
 African American14 (25%)5 (26%)6 (21%)3 (30%)9 (24%)
Education
 ≤ High School9 (16%)3 (16%)5 (18%)1 (10%)6 (16%)
 College18 (32%)5 (26%)9 (32%)4 (40%)13 (34%)
 > College30 (53%)11 (58%)14 (50%)5 (50%)19 (50%)
Marital Status
 Single13 (23%)4 (21%)6 (21%)3 (30%)9 (24%)
 Married37 (65%)13 (68%)18 (64%)6 (60%)24 (63%)
 Divorced7 (12%)2 (11%)4 (14%)1 (10%)5 (13%)
Rank
 Police Officer38 (67%)13 (68%)19 (68%)6 (60%)25 (66%)
 Sergeant/Lieutenant10 (18%)2 (11%)6 (21%)2 (20%)8 (21%)
 Captain/Detective9 (16%)4 (21%)3 (11%)2 (20%)5 (13%)
Years Worked
 1-5 (n=13)13 (23%)5 (26%)6 (21%)2 (20%)8 (21%)
 6-10 (n=7)7 (12%)0 (0%)6 (21%)1 (10%)7 (18%)
 11-15 (n=15)15 (26%)5 (26%)7 (25%)3 (30%)10 (26%)
 >15 (n=22)22 (39%)9 (47%)9 (32%)4 (40%)13 (34%)
Shift Changes per Year
 Mean ± SD35.6 ± 52.432.6 ± 45.938.5 ± 58.554.8 ± 72.842.7 ± 61.8

Abbreviations: SD - Standard deviation; Column percentages not totaling 100% are due to rounding.

Mean Waking and Diurnal Salivary Cortisol Levels (95% Confidence Intervals) by PER3 Variable Number Tandem Repeat Genotype, BCOPS Study, Buffalo, NY, USA, 2001–2003

Dependent Variables4/4(n=19)4/5(n=28)5/5(n=10)4/5+5/5(n=38)p-values
4/4 vs. 4/54/4 vs. 5/54/5 vs. 5/54/5+5/5 vs. 4/4
Waking AUCI63(−45–172)122(36–209)151(6–296)130(56–203)0.400.330.740.31
Waking AUCG448(254–641)826(660–991)646(395–897)775(630–921)<0.010.190.21<0.01
Diurnal AUCG4,996(3,833–6,162)7,172(6,215–8,130)7,279(5,768–8,789)7,201(6,368–8,033)<0.010.020.90<0.01
Diurnal Slope−0.0026(−0.0035–−0.0018)−0.0030(−0.0037–−0.0024)−0.0020(−0.0031–−0.0010)−0.0028(−0.0033–−0.0022)0.470.380.130.84
CESD5.9(3.5–9.1)5.8(3.7–8.3)6.1(3.0–10.3)5.9(4.0–8.1)0.910.950.880.95
IES20.8(13.0–28.5)18.4(11.8–25.0)18.4(8.3–28.4)18.4 (12.6–24.1)0.610.700.990.59
Life Events Scale2.4(1.3–3.7)2.9(1.8–4.2)1.9(0.7–3.7)2.6(1.7–3.7)0.510.660.320.73

95% confidence intervals are in parentheses. Units for AUCI and AUCG are nmol/L-minutes. Abbreviations: AUCG – Area Under the Curve with respect to ground; AUCI – Area Under the Curve with respect to increase; CESD – Center for Epidemiologic Studies Depression scale; and IES – Impact of Events. Adjustments: Waking AUCG, CESD, and IES adjusted for rank; Waking AUCI adjusted for gender; Diurnal AUCG adjusted for education and age; Diurnal Slope adjusted for gender and age group; Life Events Scale adjusted for race.

Mean Waking and Diurnal Cortisol (95% Confidence Intervals) by Shift Type and PER3 VNTR Genotype, BCOPS Study, Buffalo, NY, USA, 2001–2003.

Risk Factor4/44/5 + 5/5p-value 4/4 vs. 4/5 + 5/5
Waking AUCI
Shift Type
Day126 (−14–267)n=1061 (−43–166)n=180.46
Night + Afternoon–8 (–164–148)n=9202 (93–311)n=170.03
p-value Day vs. Night + Afternoon0.200.07
Waking AUCG
Shift Type
Day512 (261–762)n=10736 (544–928)n=180.16
Night + Afternoon361 (52–670)n=9791 (573–1009)n=170.01
p-value Day vs. Night + Afternoon0.440.69
Diurnal AUCG
Shift Type
Day4965 (3177–6753)n=97120 (5900–8340)n=170.04
Night + Afternoon5003 (3222-6783)n=97347 (6078-8615)n=160.03
p-value Day vs. Night + Afternoon0.980.79
Diurnal Slope
Shift Type
Day−0.0033 (−0.0044–−0.0021)n=9−0.0024 (−0.0032–−0.0015)n=170.22
Night + Afternoon−0.0021 (−0.0034–−0.0007)n=9−0.0032 (−0.0041–−0.0023)n=160.12
p-value Day vs. Night + Afternoon0.210.19

Units for AUCI and AUCG are nmol/L-minutes. Abbreviations: AUCG – Area Under the Curve (Ground); AUCI – Area Under the Curve (Increase). Adjustments: Waking AUCG adjusted for rank; Waking AUCi adjusted for gender; Diurnal AUCG adjusted for education and age; Diurnal Slope adjusted for gender and age group.

Mean Waking and Diurnal Cortisol (95% Confidence Intervals) by Shift changes and PER3 VNTR Genotype, BCOPS Study, Buffalo, NY, USA, 2001–2003.

Risk Factor4/44/5 + 5/5p-value 4/4 vs. 4/5 + 5/5
Waking AUCI
Cumulative Shift Changes
<1760 (−79–199)n=12123 (2–244)n=150.50
≥1768 (−110–247)n=7134 (28–241)n=200.52
p-value <17 vs. ≥170.940.89
Waking AUCG
Cumulative Shift Changes
<17430 (192–667)n=12877 (651–1104)n=15<0.01
≥17494 (189–800)n=7708 (527–888)n=200.22
p-value <17 vs. ≥170.730.22
Diurnal AUCG
Cumulative Shift Changes
<175,040 (3490–6590)n=117,290 (5975–8604)n=140.03
≥174,911 (3078–6745)n=77,181 (6027–8335)n=190.03
p-value <17 vs. ≥170.910.90
Diurnal Slope
Cumulative Shift Changes
<17−0.0022 (−0.0033–−0.0011)n=11−0.0031 (−0.0041−–0.0021)n=140.24
≥17−0.0033 (−0.0046–−0.0019)n=7−0.0026 (−0.0034–−0.0017)n=190.35
p-value <17 vs. ≥170.220.38

Shift change (from 1994 or initiation of employment until clinic examination in 2001), categories were based on a median split. Units for AUCI and AUCG are nmol/L-minutes. Abbreviations: AUCG – Area Under the Curve (Ground); AUCI – Area Under the Curve (Increase). Adjustments: Waking AUCG adjusted for rank; Waking AUCi adjusted for gender; Diurnal AUCG adjusted for education and age; Diurnal slope adjusted for gender and age group.