circadian rhythm of urinary steroid metabolites

Original Article

Circadian rhythm of urinary steroid metabolites

Walid K Jerjes1, Anthony J Cleare2, Timothy J Peters1 and Norman F Taylor1

Abstract

Addresses1Department of Clinical Biochemistry, Guy’s,King’s College London School of Medicine,Bessemer Road, London SE5 9RS and2Section of Neurobiology of Mood Disor-ders, Division of Psychological Medicine,The Institute of Psychiatry and Guy’s,King’s College London School of Medicine,De Crespigny Park, London SE5 8AF, UK

Correspondence:Mr Walid K JerjesEmail: [email protected]

Background Samples submitted for urinary steroid profile analysis are oftenuntimed, but influence of collection time on interpretation is unknown. We reportcircadian rhythms of the major steroid metabolites and derived ratios in urinecollected at 3-h intervals over 24 h, after first establishing that disturbance of sleepassociated with collection does not alter rhythms on the succeeding day.

Methods Assay of steroid metabolites (gas chromatography) and creatinine inurine collections made by 10 men and 10 women every 3 h starting at 2100 for24 h. Data were subjected to cosinor analysis.

Results Summed cortisol and androgen metabolites exhibited significantcircadian rhythms, as expected, but with a surprisingly long time-lag (maxima ataround 1400). Amplitudes were different, so that the ratio cortisol/androgenmetabolites also showed a significant rhythm. The ratios 5a/5b tetrahydrocortisoland 20-hydroxy/20-oxo cortisol metabolites showed significant rhythms which werenot in phase with total cortisol metabolites, while 11-hydroxy/11-oxo cortisolmetabolites showed no rhythm. There were no gender differences in time ofmaxima. Previously established gender differences in metabolite levels wereconfirmed. Creatinine levels showed no circadian rhythm.

Conclusion Circadian variation should be considered when interpreting resultsfrom urine steroid analysis. Calculation of steroid/steroid or steroid/creatinine ratiosis not informative in untimed collections.

Ann Clin Biochem 2006; 43: 287–294

Introduction

Urinary steroid pro¢ling by gas chromatography (GC)or gas chromatography--mass spectrometry is usefulfor investigation of disordered steroid production andmetabolism and, when 24-h collections are made, fordetermination of steroid production rates. During theprovision of a referral service, we have noted a substan-tial clinical demand for analysis of untimed collections,most commonly taken from childrenwith possible pre-cocious adrenarche and a general belief that relatingsteroid levels to creatinine would o¡set any disadvan-tage of a short collection. The circadian rhythm ofthese metabolites remains unreported. This study setout to examine the variability of steroid metabolitesand creatinine obtained in sequential 3-h urine collec-tions over one day in adult volunteers under carefullycontrolled conditions of collection, since the circadianrhythm of cortisol secretion is entrained by the light-

dark1,2 and sleep-wake cycles3, while age, weight andthe timing of meals may also be in£uential. Since vo-lunteers had to wake up during the night during 24 hsampling we checked whether this disturbed therhythm the following day by comparing results withthose of 3 h sampling during the daytime only.

Urine steroid pro¢ling in sequential collections mayhave a role in the investigation of disorders inwhich cir-cadian variation of cortisol is altered, such as Cushing’ssyndrome4 and depression.5 Single serum cortisol deter-minations are accepted as unreliable for evaluation ofadrenocortical function6 and stress associated withsequential blood sampling may also make it di⁄cult toestablish the cortisol rhythm unless an indwelling can-nula is used; even then, di⁄culties often arise withblockages, pain and the need for resiting cannulae.

Urinary free cortisol shows a similar rhythm to plas-ma cortisol, but with a delay of about 3 h.7 However,less than 5% of secreted cortisol is cleared as free corti-

r 2006 The Association for Clinical Biochemistry 287

sol; the rest is excreted as cortisol metabolites.8 Levelsof total cortisol metabolites (TCM) provide a sensitivemeans of detecting changes in rates of cortisol secre-tion, as in asthmatics treated with inhaled glucocorti-coids.9 Quanti¢cation of individual metabolites permitsexamination of changes in cortisol metabolism such ascortisol--cortisone interconversion. Changes in thisequilibrium are associated with some disorders suchas obesity10 and hirsutism,11,12 but whether there arecircadian changes in normal or pathological situationshas never been explored. We also examined possiblerelationships of age and body mass index with circa-dian rhythm.

Materials and methods

Volunteer subjectsAll volunteers were healthy and recruited from amonghospital sta¡ and students. One group (10m,10f) madesequential 3-hurine collections from 0600 to 2100 (DTgroup). Men and women were matched for age (32712and 34711 years) and body mass index (BMI) (2474and 2475 kg/m2), respectively (all mean7SD). A sec-ond group (10m,10f,17 were also in the day time [DT]group) made sequential 3-h urine collections, startingat 2100, for 24 h (DTþnight time [NT] group). Theywere also matched for age (31712 and 33712 years)and BMI (2474, and 2375 kg/m2), respectively. Sub-jects were all studied during winter, between October2001and March 2002. No female subjects were takingthe oral contraceptive. Menstrual cycle was not re-corded, since there are no changes of the studiedsteroid metabolites over the menstrual cycle (NFT un-published observation) and no relationship of menstru-al cycle and cortisol circadian rhythm has been foundusing saliva sampling.13 All subjects had been medica-tion-free for at least two weeks, did not smoke, wereasked to limit their intake of ca¡eine and alcohol dur-ing the collection period and had no history of hyper-sensitivity to corticosteroids. All had normal dietaryhabits, taking breakfast, lunch and dinner at about thesame time. All subjects habitually went to bed between2300 and 0100 and got up between 0600 and 0800. Allsubjects provided written informed consent before par-ticipation. King’s Research Ethics Committee Approvalfor this study was obtained.

QuestionnairesAll groups had a good sleep quality. Total PittsburghSleep Quality Index (PSQI) scores were (men, DT)2.471.3; (women, DT) 2.071.3 and (men, DTþNT)3.372.3; (women, DTþNT) 2.872.0, respectively (allmean7SD). ‘Good’ is de¢ned as a score of 7 or less.14

They did not have clinically signi¢cant levels of anxietyor depression. Scores for the Hospital Anxiety and

Depression scale (HADS) were 2.871.9, 3.473.3 and3.373.1, 3.773.1, respectively, for anxiety and3.272.3; 2.773.3 and 2.172.4; 2.572.8, respectively,for depression. Signi¢cant levels of depression and an-xiety are usually de¢ned as10 or more.15

Steroid metabolite assayUrinary steroid pro¢le analysis used high-resolutiongas chromatography of methyloxime-trimethylsilylether (MO-TMS) derivatives, as previously described.16

The intra- and interassay coe⁄cients of variation werebetween 7.1--21.1% and 11.2--21.9%, respectively, fordi¡erent metabolites. Each metabolite was calculatedas mg/3 h period. Derived sums were as previouslyreported.17 Urinary androgen metabolites (AM) weredetermined from the sum of androsterone and aetio-cholanolone.TCMwere the sum of tetrahydrocortisone(THE), tetrahydrocortisol (THF), allo-tetrahydrocortisol(5aTHF), a-cortolone, b-cortolone, a-cortol, and b-cor-tol. 20-hydroxy metabolites of cortisol (20OH) were thesum of a-cortolone, b-cortolone, a-cortol, and b-cortol,and 20-oxo metabolites of cortisol (20OXO) were thesum of THE, THF, and 5aTHF. The ratio of THFþ5aTHF/THE (11OH/11OXO) was calculated as an indexof total net 11b-hydroxysteroid dehydrogenase (11b-HSD) activity. The ratio 5a/5b THF was calculated asan index of 5a versus 5b reduction and 20OH/20OXOas an index of net 20-HSD activity.

Creatinine assayUrinary creatinine was measured by a Bayer Advia1650. Values were calculated as nmol/3 h in order toexamine steroid/creatinine ratios.

Circadian rhythm analysisIndividual and population mean cosinor analysis todetermine circadian rhythm parameters of each vari-able was performed using TSA-Seriel Cosinor software(Expert Soft Technologie, Laboratoire d’InformatiqueBioMeŁ dicale, France) for analysis of biological time ser-ies by least-squares estimation. Population mean cosi-nor analysis is based on the means of parameterestimates obtained from individuals in the study sam-ple to derive the following parameters: (1) the goodnessof ¢t of a cosinor curve ¢tted to the data; (2) midlineestimate statistic of rhythm (MESOR), de¢ned as therhythm-adjusted mean; (3) amplitude, de¢ned as halfthe extent of rhythmic change in a cycle approximatedby a ¢tted curve (di¡erence between nadir and peak);(4) acrophase, de¢ned as the time of peak in the cosinorcurve ¢tted to the data. The acrophase is expressed asa phase angle in degrees, so the formula (Value indegrees/3601)�24 h) can be used to establish the clocktime of the peak.


288 Jerjes et al.

Statistical analysisComparisons between results for the entire period inmen and women were made by the repeated measuresanalyses of variance (ANOVA). Additionally, we per-formed post hoc t-tests on values for each of the 3-hblocks over 24 h to determine if there were any di¡er-ences in cortisol levels at a particular time of the day.The coe⁄cient of correlation between circadianrhythm parameters, age and BMI was calculated bythe general linear regression method.

Results

Comparison of TCM excretion between 0600 and 2100(DT) and the excretion for the same period within theDTþNTseries (Table1) showed there was a signi¢cantcircadian rhythm in both men and women which didnot di¡er between the DT and DTþNT conditions.Thus, the following analyses were based on DTþNTcollections only.

Gender differencesLevels of TCM were signi¢cantly higher in men(P¼0.008) and at all time points: Po0.005 for1200--1500, and Po0.05 for the rest of the periods(Figure1a). Levels of AM were also signi¢cantly higherin men (Po0.0005) and at all time points exceptfor 0600--0900 and 1800--2100: Po0.005 for0900--1800, and Po0.05 for 2100--0600 (Figure1b).

The ratio of 11OH/11OXO was signi¢cantly higher inmen (Po0.05), but no di¡erences were observed in posthoc t-tests between men and women at any 3-h period(Figure 2a). The ratio of 5a/5b THF was signi¢cantlyhigher in men (F (1,18)¼9.8, P¼0.006) and at all timepoints (2100--1200, Po0.05; 1200--2100, Po0.005)(Figure 2b).The ratio of 20OH/20OXOwas signi¢cantlylower in men (Po0.05), but no di¡erences were ob-served for any 3-h period over 24 h (Figure 2c).

Table 1 Effect of prior night-time collections on circadian rhythm parameters of TCM between 0600 and 2100: comparison ofcollections made only in this period (DT) with collections made as part of a 24 h series (DT+NT)

Males Females

DT DT+NT t-test DT DT+NT t-test

% Rhythm(goodness of fit)

77% Po0.0001 72% po0.001 N/A 71% Po0.0001 72% Po0.001 N/A

MESOR (mg/3 h) 1025 (657–1325) 1031 (670–1330) 0.6 467 (291–561) 470 (284–570) 0.8Amplitude (mg/3 h) 513 (429–635) 542 (337–627) 0.7 222 (220–270) 235 (214–296) 0.6Acrophase �206 (�235 to �177) �210 (�240 to �189) 0.8 �201 (�221 to �184) �212 (�256 to �169) 0.9

Values are expressed as means (95% confidence intervals), Acrophase is presented as phase angle in degrees where 3601=24 hDT, day time; NT, night time

Cortisol metabolites (TCM)

0200400600800

100012001400160018002000

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

WomenMen

*

** *

*

**

*

*

Androgen metabolites (AM)

0

100

200

300

400

500

600

�g

/3 h

�g

/3 h

* **

**

**

**

TCM/AM

0.0

1.0

2.0

3.0

4.0

5.0

Rat

io

Clock time

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

Clock time

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

Clock time

(a)

(b)

(c)

WomenMen

WomenMen

Figure 1 Mean values and standard errors of (a) total urinarycortisol metabolite excretion; (b) urinary androgen metabolitesand (c) cortisol metabolites/androgen metabolites at 3-hintervals over 24 h. *Po0.05, **Po0.005


Urinary steroid metabolite circadian rhythm 289

Circadian rhythmCosinor analysis derived population mean circadianparameter estimates by gender are detailed in Table 2.Both males and females showed a signi¢cant rhythmover 24 h of TCM and AM. The amplitude of the TCMrhythmwas lower than that of the AM rhythm, so thatTCM/AM also showed a circadian rhythm (Figure 1c).Creatinine level showed no circadian rhythm, so thatAM/creatinine and TCM/creatinine showed circadianrhythms with similar characteristics to AM and TCMalone (Figure 3 a and b).

The amplitude and MESOR of AM were higher inmen compared to women, with no changes in theacrophase. The acrophase of TCM was not statisticallydi¡erent from AM, althoughTCM lagged behind AM by56 and 44min for men and women, respectively.

There were signi¢cant rhythms for 5a/5b THF and20OH/20OXO, but no signi¢cant rhythm of the 11OH/11OXO ratio in either men or women. The MESORs forTCM, 11OH/11OXO ratio and 5a/5b THF ratio werehigher in men than in women, while the 20OH/20OXO ratio MESOR was lower. For amplitude, onlyTCM di¡ered between groups, being signi¢cantly high-er in men than women. The acrophase of TCMoccurred signi¢cantly earlier than those for the 5a/5bTHF and 20OH/20OXO ratios in both men andwomen. Calculated acrophases are also shown for the11OH/11OXO ratio in Table 2, but since the circadianrhythmwas not signi¢cant these would not be valid.

There were no signi¢cant di¡erences between menand women in age, BMI and acrophase of TCM.Whenthe data were combined, there was a signi¢cant nega-tive correlation between age and acrophase of TCM(r¼�0.93, Po0.001), whereas a non-signi¢cant nega-tive correlation was seen between BMI and acrophaseof TCM (r¼�0.64, P¼0.07).

Discussion

Effect of sleep disruptionThis study established that waking every 3 h to passurine does not have a signi¢cant e¡ect on subsequentsteroid levels, so our 24-h data are likely to provide avalid indicator of the normal rhythmof steroid metabo-lite excretion.This accords with studies of the e¡ects ofsleep disruption on HPA axis activity, reviewed by Stei-ger,18 which indicate that the cortisol circadian rhythmin blood is relatively robust. Night shift work does notresult in any change in the acrophase in the ¢rst days,although there is a decrease of rhythm amplitude.19

After East--West air travel, adjustment of the cortisolrhythm requires several days.20 Experimental regimesof sleep deprivation have shown relatively small short-term changes.21 Anticipation of awakening induces amarked increase of ACTH,22 so that, paradoxically, astudy in which an experimenter wakes the subjectmay be less disruptive than our protocol, in which thevolunteers planned their own awakenings. However,no e¡ect on this of our collection regime was evident.

Circadian rhythm of TCM and AMTo our knowledge, this is the ¢rst study to examine thecircadian rhythm of urinary steroid metabolites inhealthy volunteers. Men and women show a similarphase of the rhythm. The circadian rhythm of AMshows a lower amplitude than TCM. This is to be ex-pected because androgens arise from the gonads aswell as from the adrenals.23 Since the ratio TCM/AMshowed a circadian rhythm, we consider that itwould be unwise to assess androgen production bycalculation of this ratio in untimed samples. Although

Women

Men

Women

Men

Women

Men

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

Clock time

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

Clock time

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

Clock time

11OH/11OXO

0.00.20.40.60.81.01.21.4

Rat

io

5�/5� THF

0.0

0.5

1.0

1.5

2.0

Rat

ioR

atio

* *

**

******

*

20OH/20OXO

0.00

0.10

0.20

0.30

0.40

0.50

(a)

(b)

(c)

Figure 2 Mean values and standard errors of (a) 11OH/11OXO ratio; (b) 5a/5b THF ratio and (c) 20OH/20OXO ratio at3-h intervals over 24 h. *Po0.05, **Po0.005


290 Jerjes et al.

our data were obtained in adults, this conclusion canprobably be applied to the assessment of androgenproduction in children with signs of early puberty oradrenarche. It should be noted, however, that untimedcollection may otherwise be very informative inchildren. Higher levels of TCM and AM excretion inmen, shown by ANOVA and comparison of MESORand Amplitude, agree with our previous ¢ndings basedon single 24-h urine sample collections.8,24

Comparing the phase of the cortisol rhythm by var-ious measures, serum cortisol shows a peak between0700 and 1100,6 while urinary free cortisol peaks ataround 1200 midday7 and we have similar ¢ndings inour volunteers (Jerjes et al., unpublished data). Thus,there is a lag time between serum cortisol and urinefree cortisol of 2--3 h and a further delay of about 2 hfor the appearance of cortisol metabolites. This delaybetween serum change and the appearance of urinarymetabolites should be taken into account in dynamicstudies where urine analysis would be informative,such as ACTH stimulation tests in investigation ofcongenital adrenal hyperplasia.

Cortisol metabolite ratiosValues for cortisol metabolite ratios over 24 h are inagreement with our previous reports based on single

Table 2 Comparison of circadian rhythm parameters of steroids in healthy men and women over 24 h

Parameters % Rhythm(goodness of fit)

MESOR Amplitude Acrophase Equivalenttime (h)

MenTCM 81%, Po0.0001 884 (590–1180)** 406 (437–465)*** �213 (�255 to �190) 141211-OH/11-Oxo 22%, P=0.1 0.92 (0.79–1.05)* 0.086 (0.05–0.16) �271 (�288 to �253) 18045a/5b THF 68%, Po0.005 1.31 (1.06–1.55)** 0.29 (0.18–0.41) �334 (�336 to �302)a 221620OH/20OXO 51%, Po0.05 0.25 (0.20–0.30)* 0.059 (0.037–0.087) �340 (�240 to �360)b 2240AM 51%, P=0.04 420 (356–484)*** 89 (40–166)** �199 (�212 to �176) 1316TCM/AM 63%, P=0.02 3.08 (2.17–40) 1.06 (1.01–1.18) �234 (�277 to �191) 1536AM/creatinine 50%, P=0.03 191 (132–209)* 38 (25–51)* �177 (�219 to �135) 1148TCM/creatinine 62%, P=0.001 490 (362–602)* 214 (192–255)*** �211 (�242 to �180) 1404

WomenTCM 85%, Po0.0001 393 (316–470) 183 (173–241) �214 (�253 to �175) 141611-OH/11-Oxo 30%, P=0.2 0.75 (0.61–0.90) 0.063 (0.08–0.19) �265 (�355 to �240) 17405a/5b THF 78%, Po0.001 0.73 (0.52–0.94) 0.17 (0.068–0.27) �340 (�380 to �305)c 224020OH/20OXO 51%, Po0.05 0.33 (0.27–0.39) 0.046 (0.012–0.079) �360 (�389 to �330)c 2400AM 52%, P=0.04 191 (139–243) 40 (26–82) �203 (�205 to �186) 1332TCM/AM 66%, P=0.001 2.78 (1.71–3.81) 0.7 (0.65–1.02) �236 (�271 to �202) 1544AM/creatinine 49%, P=0.04 145 (126–196) 21 (13–39) �201 (�172 to �224) 1324TCM/creatinine 65%, P=0.006 327 (233–422) 135 (139–176) �217 (�255 to �180) 1428

Values are expressed as means (95% confidence intervals), Acrophase is presented as phase angle in degrees where 3601=24 h. For men versuswomen: *Po0.05, **P o0.005, ***Po0.0005. Within the group: acrophase TCM versus metabolite ratios: aPo0.05, bPo0.005 and cPo0.0005.Equivalent times are given for convenience. MESOR and amplitude are expressed as mg/3 h for steroids and their ratios, and mg/nmol for AM andTCM over creatinine ratios

TCM/creatinine

0100200300400500600700800900

1000

�g

/nm

ol

�g

/nm

ol

AM/creatinine

050

100150200250300350400 Men

Women

MenWomen

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

2100–2400

2400–0300

0300–0600

0600–0900

0900–1200

1200–1500

1500–1800

1800–2100

Clock time

Clock time

(a)

(b)

Figure 3 Mean values and standard errors of (a) total urinarycortisol metabolite excretion/creatinine and (b) urinary andro-gen metabolites/creatinine at 3-h intervals over 24 h



24 h urine sample collections.8,25 The 11OH/11OXOratio represents an index of 11b-HSD enzyme activity:type 1 is predominant in the liver and acts as a reduc-tase (cortisone to cortisol),26 while type 2 is predomi-nantly renal and acts as a dehydrogenase (cortisol tocortisone).27 Decrease of this ratio is associated withincrease of cortisol metabolic clearance rate, as inapparent cortisone reductase de¢ciency,11 polycysticovary syndrome,12 growth hormone excess28 andobesity.10,29 These states are probably all due to de-creased11b-HSD type1activity.30 Increase in this ratiois seen in patients with apparent mineralocorticoidexcess,31 in which 11b-HSD type 2 is de¢cient, and inCushing’s syndrome.32 The latter may be due to anacute increase of cortisol exceeding the capacity of11b-HSD type 2, since it is also seen after ACTH stimu-lation.33 It might thus have been expected that the cir-cadian peak in cortisol secretion would be associatedwith signi¢cant daily increase in the11OH/11OXO ratiodue to amass e¡ect of cortisol, but this is clearly not thecase. Indeed, values were highest at 2100--0300, whenurinary cortisol metabolite secretion was at a nadir.Thus, untimed urine samples should permit an accu-rate assessment of11b-HSD status.

In contrast, the ratio 5a/5b THF shows a strikingdaily rhythm, with an acrophase that lags behind thatof TCM by 8.1h (males) and 8.4 h (females). This ratiore£ects relative activities of the 5a- and 5b-reductasesin the liver and probably other tissues, including fat. Itis increased in obese subjects, suggesting an enhance-ment of 5a-reductase activity, since fat contains 5a-reductase activity, but not 5b-reductase activity.34 It islikewise decreased in anorexia nervosa.35 Comparisonof a variety of clinical disorders indicates that 5a/b re-duction and 11-HSD activity are independently modu-lated,32 but changes do occur secondarily to markedchanges in cortisol 11-oxoreduction. The ratio is in-creased in de¢ciency of 11b-HSD type 2 (apparentmineralocorticoid excess,36) and decreased in apparentde¢ciencyof11b-HSD type1 (apparent cortisone reduc-tase de¢ciency,37). Acute cortisol increase after ACTHstimulation does not result in changes,33 althoughCushing’s syndrome is associated with decrease of thisratio.38 Again, there is no evidence for a mass e¡ectwithin our data.

The acrophase for the ratio 20OH/20OXO lags be-hind that of TCM by 8.5 h (males) and 9.7 h (females).This ratio does not change in many clinical states, butit is elevated in alcoholic liver disease,39 which may bedue to increased liver activity of 20-reductases ordecreased A-ring reduction.

The signi¢cant changes of 5a/5b THF and 20OH/20OXO ratios over 24 h are not co-dependent nor dothey appear to be due to changed mass e¡ect of cortisolover the day. They are unlikely to result from changesin enzyme level, since the duration is short. Alternative

causes include changes in substrate, inhibitor, cofactoror hormone concentrations, perhaps related to mealintake, but no candidate compounds can currently besuggested. Whether they represent passive e¡ects orare evidence for active modulation of steroid hormoneaction, perhaps related to circadian entrainment ofphysiological functions, is currently unclear. Our ¢nd-ings do indicate that interpretation of cortisol metabo-lite ratios in untimed urine collections would bemisleading if reference limits did not take the circadian£uctuation into account.

Correlates of circadian rhythm parametersThe ¢nding of a signi¢cant negative correlation be-tween age and acrophase is in line with previousreports of an earlier acrophase of plasma cortisol inolder subjects.40 This might be due to a changed sleep-wake cycle, since older subjects tend to show a forwardshift in sleeping times.41 The negative trend betweenBMI and acrophase of urinary cortisol metabolitesmight suggest that obese subjects have an earlier peakof cortisol, although none of our volunteers were clini-cally obese.This could re£ect a faster cortisol metabolicclearance rate or it might be due to the confoundinge¡ects of age, since subjects tend to get fatter with age.

Food intake stimulates cortisol release. Carefulstudies by Follenius et al.,42 using frequent plasmasampling, showed a marked e¡ect of a meal at noonand less consistent e¡ects at other times, relatively in-dependent of customary eating patterns. In our study,no e¡ect is apparent. This may be because the urinecollections were made at relatively wide intervals, andtiming of meals was not controlled experimentally.

Investigation of the serum cortisol rhythm in disor-ders such as depression and Cushing’s syndrome is notonly of interest in the circadian physiology in those dis-eases but also for their diagnostic utility. Approxi-mately 50% of depressed patients fail to suppresscortisol in response to dexamethasone administra-tion43 and this attenuated feedback probably related toelevated circulating cortisol levels and hypersecretionof corticotrophin-releasing hormone and ACTH.44,45

Dahl et al.5 reported that patients with depressionshowed a preserved plasma cortisol circadian rhythmbut with lower amplitude and an earlier acrophase.The disappearance of the cortisol circadian rhythm inCushing’s disease is well known.46 Patients show dete-rioration of serum cortisol rhythmicity as the diseaseprogresses.4 Abnormal circadian rhythms may be part ofa cascade of dysfunction in the HPAaxis, inwhich exces-sive cortisol production damages glucocorticoid recep-tors, which impairs feedback inhibition of CRH, leadingto further excessive secretion of ACTHand cortisol.47

We conclude that 3-h urinary cortisol collectionsover 24 h are a valid way to measure circadian £uctua-


292 Jerjes et al.

tions in the excretion of androgen and cortisol and itsmetabolites.We suggest that such £uctuations need tobe taken into account when interpreting results ofurinary steroid analyses. Calculation of steroid/steroidor steroid/creatinine ratios is not informative inuntimed collections.With improved understanding ofthe circadian rhythm of steroid metabolites in normalsubjects, such alterations in excretion patternsmay have a future place in diagnosing and monitoringdisease.

AcknowledgementsWe thank the volunteers who took part in this studyand Mr J Keating and department sta¡ for carrying outthe creatinine analyses.

References

1 Czeisler CA, Allan JS, Strogatz SH, et al. Bright light resets thehuman circadian pacemaker independent of the timing of thesleep-wake cycle. Science 1986; 233: 667–71

2 Shanahan TL, Czeisler CA. Light exposure induces equivalentphase shifts of the endogenous circadian rhythms of circulatingplasma melatonin and core body temperature in men. J ClinEndocrinol Metab 1991; 73: 227–35

3 Bouhuys AL, Flentge F, Van den Hoofdakker RH. Effectsof total sleep deprivation on urinary cortisol, self-rated arousal,and mood in depressed patients. Psychiatry Res 1990; 34: 149–62

4 Glass AR, Zavadil AP, Halberg F, Cornelissen G, Schaaf M.Circadian rhythm of serum cortisol in Cushing’s disease. J ClinEndocrinol Metab 1984; 59: 161–5

5 Dahl RE, Ryan ND, Puig-Antich J, et al. 24-h cortisol measures inadolescents with major depression: a controlled study. BiolPsychiatry 1991; 30: 25–36

6 Nomura S, Fujitaka M, Sakura N, Ueda K. Circadian rhythms inplasma cortisone and cortisol and the cortisone/cortisol ratio. ClinChim Acta 1997; 266: 83–91

7 Morineau G, Boudi A, Barka A, et al. Radioimmunoassay ofcortisone in serum, urine, and saliva to assess the status of thecortisol–cortisone shuttle. Clin Chem 1997; 43: 1397–2407

8 Raven PW, Taylor NF. Sex differences in the human metabolismof cortisol. Endocr Res 1996; 22: 751–5

9 Fink RS, Pierre LN, Daley-Yates PT, Richards DH, Gibson A,Honour JW. Hypothalamic-pituitary-adrenal axis function afterinhaled corticosteroids: unreliability of urinary free cortisolestimation. J Clin Endocrinol Metab 2002; 87: 4541–6

10 Stewart PM, Boulton A, Kumar S, Clark PM, Shackleton CH.Cortisol metabolism in human obesity: impaired cortisone–4cortisol conversion in subjects with central adiposity. J ClinEndocrinol Meta 1999; 84: 1022–7

11 Taylor NF, Barlett WA, Dawson DJ. Cortisone reductasedeficiency: evidence for a new inborn error in metabolism ofadrenal steroids. J Clin Endocrinol 1984; 102: 89

12 Rodin A, Thakkar H, Taylor NF, Clayton R. Hyperandrogenismin polycystic ovary syndrome. Evidence of dysregulation of 11beta-hydroxysteroid dehydrogenase. New Engl J Med 1994; 330:460–5

13 Kudielka BM, Kirschbaum CA. wakening cortisol responses areinfluenced by health status and awakening time but not by

menstrual cycle phase. Psychoneuroendocrinology 2003; 28:35–47

14 Buysse DJ, Reynolds CF, Monk TH, Berman SR, Kupfer DJ.The Pittsburgh Sleep Quality Index: a new instrument forpsychiatric practice and research. Psychiatry Res 1989; 28:193–213

15 Zigmond AS, Snaith RP. The hospital anxiety and depressionscale. Acta Psychiatr Scand 1983; 67: 361–70

16 Raven PW, Checkley SA, Taylor NF. Extra-adrenal effects ofmetyrapone include inhibition of the 11-oxoreductase activity of 11beta-hydroxysteroid dehydrogenase: a model for 11-HSD Ideficiency. Clin Endocrinol 1995; 43: 637–44

17 Trainer PJ, Drake WM, Perry LA, Taylor NF, Besser GM, MonsonJP. Modulation of cortisol metabolism by the growth hormonereceptor antagonist pegvisomant in patients with acromegaly.J Clin Endocrinol Metab 2001; 86: 2989–92

18 Steiger A. Sleep and the hypothalamo-pituitary-adrenocorticalsystem. Sleep Med Rev 2002; 6: 125–38

19 Leese G, Chattington P, Fraser W, Vora J, Edwards R, WilliamsG. Short-term night-shift working mimics the pituitary-adrenocor-tical dysfunction in chronic fatigue syndrome. J Clin EndocrinolMetab 1996; 81: 1867–70

20 Desir D, Van Cauter E, Fang VS, et al. Effects of "jet lag" onhormonal patterns. I. Procedures, variations in total plasmaproteins, and disruption of adrenocorticotropin-cortisol periodicity.J Clin Endocrinol Metab 1981; 52: 628–41

21 Leproult R, Copinschi G, Buxton O, Van Cauter E. Sleep lossresults in an elevation of cortisol levels the next evening. Sleep1997; 20: 865–70

22 Born J, Hansen K, Marshall L, Molle M, Fehm HL. Timing the endof nocturnal sleep. Nature 1999; 397: 29–30

23 Brooks RV, Makin HLJ, eds. Androgen: Physiology and Pathology.In: Biochemistry of Steroid Hormones. 2nd edn. Oxford: BlackwellScientific, 1984; 565–94

24 Raven PW, Taylor NF. Steroid metabolism in healthy men andwomen. J Endocrinol 1995; 147: 100

25 Raven PW, Taylor NF. 11Beta-HSD and 17beta-HSD as biologicalmarkers of depression: sex differences and correlation withsymptom severity. Endocr Res 1998; 24: 659–62

26 Ricketts ML, Shoesmith KJ, Hewison M, Strain A, Eggo MC,Stewart PM. Regulation of 11 beta-hydroxysteroid dehydrogenasetype 1 in primary cultures of rat and human hepatocytes.J Endocrinol 1998; 156: 159–68

27 White PC, Mune T, Agarwal AK. 11beta-Hydroxysteroid dehy-drogenase and the syndrome of apparent mineralocorticoidexcess. Endocr Rev 1997; 18: 135–56

28 Monson JP. Growth hormone and cortisol metabolism. ClinEndocrinol (Oxford) 1998; 49: 281–2

29 Rask E, Walker BR, Soderberg S, et al. Tissue-specific changes inperipheral cortisol metabolism in obese women: increased adipose11beta-hydroxysteroid dehydrogenase type 1 activity. J ClinEndocrinol Metab 2002; 87: 3330–6

30 Tomlinson JW, Draper N, Mackie J, et al. Absence of Cushingoidphenotype in a patient with Cushing’s disease due to defectivecortisone to cortisol conversion. J Clin Endocrinol Metab 2002; 87:57–62

31 Stewart PM, Walker BR, Holder G, O’Halloran D, Shackleton CH.11 beta-Hydroxysteroid dehydrogenase activity in Cushing’ssyndrome: explaining the mineralocorticoid excess state of theectopic adrenocorticotropin syndrome. J Clin Endocrinol Metab1995; 80: 3617–20



32 Raven PW, Taylor NF. Evidence for independent modulation ofhuman 11-HSD and 5 alpha/5 beta reductase activities. EndocrRes 1996; 22: 811–5

33 Taylor NF, Fisher J, Llewelyn DEH, Raven PW. Origin of impaired11-dehydrogenation of corticosteroids in conditions of corticoster-oid excess. J Clin Endocrinol 1994; 143: 56

34 Andrew R, Phillips DI, Walker BR. Obesity and gender influencecortisol secretion and metabolism in man. J Clin Endocrinol Metab1998; 83: 1806–9

35 Vanluchene E, Aertsens W, Vandekerckhove DV. Steroidexcretion in anorexia nervosa patients. Acta Endocrinol (Copenh)1979; 90: 133–8

36 Monder C, Shackleton CH, Bradlow HL, et al. The syndrome ofapparent mineralocorticoid excess: its association with 11 beta-dehydrogenase and 5 beta-reductase deficiency and someconsequences for corticosteroid metabolism. J Clin EndocrinolMetab 1986; 63: 550–7

37 Phillipou G, Palermo M, Shackleton CH. Apparent cortisonereductase deficiency: a unique form of hypercortisolism. J ClinEndocrinol Metab 1996; 81: 3855–60

38 Vanluchene E, Vandekerckhove D, Thiery M, Van Holder R.Changes in cortisol metabolism in various physiological andpathological situations. Ann Endocrinol (Paris) 1981; 42: 284–5

39 Cronholm T, Curstedt T, Schmidt DN, Sjovall J. Steroid profiles inurine and plasma of alcoholics during withdrawal. Alcohol 1985; 2:677–82

40 Touitou Y, Sulon J, Bogdan A, et al. Adrenal circadian system inyoung and elderly human subjects: a comparative study.J Endocrinol 1982; 93: 201–10

41 Sherman B, Wysham C, Pfohl B. Age-related changes in thecircadian rhythm of plasma cortisol in man. J Clin EndocrinolMetab 1985; 61: 439–43

42 Follenius M, Brandenberger G, Hietter B. Diurnal cortisol peaksand their relationships to meals. J Clin Endocrinol Metab 1982; 55:757–61

43 Carroll BJ. The dexamethasone suppression test for melancholia.Br J Psychiatry 1982; 140: 292–304

44 Nemeroff CB, Widerlov E, Bissette G, et al. Elevated concentra-tions of CSF corticotrophin-releasing factor-like immunoreactivityin depressed patients. Science 1984; 226: 1342–4

45 DeSouza EB. Corticotrophin-releasing factor receptors: physiol-ogy, pharmacology, biochemistry and role in central nervoussystem and immune disorders. Psychoneuroendocrinology 1995;20: 789–819

46 Weitzman ED. Circadian rhythms and episodic hormone secretionin man. Ann Rev Med 1976; 27: 225–43

47 Stokes PE. The potential role of excessive cortisol induced byHPA hyperfunction in the pathogenesis of depression. EurNeuropsychopharmacol 1995; 5: 77–82

Accepted for publication 31 March 2006


294 Jerjes et al.

circadian rhythm of urinary steroid metabolites

Documents