Clinical Endocrinology (1991), 34, 215-221 ADONIS 0300066491000431

Episodic ACTH and cortisol secretion in normal children W. H. B. Wallace, E. C. Crowne, S. M. Shaiet, C. Moore, S. Gibson, M. D. Llttley and A. White. Departments of Endocrinology and Physics, Christie Hospital, Manchester. Department of Clinical Biochemistry, University of Manchester, UK

(Received 20 February 1990; returned for revision 19 April 1990; finally revised 24 October 1990; accepted 20 November 1990)

Summary

The aim of thls study was to determine the normal relationship between ACTH and cortisol secretlon in chlldren. Fourteen children (nlne male, five female; medlan age 11.3 years) were hospitalized and blood samples were taken every 20 mln for 24 h. A circadian rhythm was observed with median 0900 h and midnight ACTH values of 1.80 and < 0.97 pmoili, and for cortisol296 and 62 nmolll respectively. The median (range) areas under the curve for ACTH and cortisol were 29.7 (9.0-53.8) pmollllh and 5114 (3562-8630) nmollilh respectively. There were no significant differences detected for ACTH and cortisol secretion between males and females, or between prepubertal (n = 9) and pubertal subjects (n = 5). Using a novel form of time series analysls we have shown that both ACTH and cortisol are secreted with a dominant perlodlclty of 0.7-1 .O h, representing 24-34 secretory eplsodes of ACTH and cortlsol In 24 h. For cortisol, but not ACTH, there is a significant secondary periodicity of 2-32 h. To look for shared perlodicltles we have used the technlque of coherency. This reveals that for SIX of the chlldren ACTH and cortisol are secreted with a significant shared perlodiclty of 0.8-1.0 h, and for a further five chlldren a slmllar secondary shared perlodlclty Is present. Therefore In normal chlldren ACTH and cortisol secretion are interdependent and episodic but are not influenced by either pubertal status or gender.

There have been no studies of the episodic nature of ACTH and cortisol secretion in children and there are few in adults (Krieger et al., 1971; Gallagher et al., 1973; Linkowski et al., 1985; Mortola et al., 1987; Liu et al., 1987; Horrocks et al., 1990; Veldhuis et al., 1990). The aim of this study is to describe ACTH and cortisol secretion in normal children so that we can increase our understanding of the pathophysio-

Correspondence: Dr S. M. Shalet. Department of Endocrinology, Christie Hospital, Wilmslow Road, Withington, Manchester M20 9BX. UK.

logy of conditions in which there may be disruption to the hypothalamic-pituitary-adrenal (HPA) axis.

While ACTH radioimmunoassays have been available since the mid- 1960s they require relatively large volumes of blood, are technically difficult to perform and have poor precision at low concentrations. The availability of a specific IRMA for ACTH (White et al., 1987) has enabled profile studies of the pattern of ACTH secretion in adults under normal (Horrocks et al., 1990; Veldhuis et al., 1990) and in different pathophysiological conditions (Gibson et al., 1988) to be performed. Recently Horrocks el al. (1990) have reported striking differences for ACTH, but not cortisol, secretion between males and females, with more pulses of higher amplitude in males. The cause of the observed sex difference in ACTH secretion is unclear. Veldhuis et al. (1990) using a deconvolution model have investigated 24-h ACTH and cortisol secretion in normal adult men to delineate more precisely the physiological basis of the observed circadian rhythm of cortisol. In this study, using a sensitive and specific IRMA assay for ACTH, we have analysed ACTH and cortisol secretion in normal children and looked for an influence of pubertal status and gender on ACTH secretion.

Subjects and methods

We studied 14 normal children (nine male, five female: nine prepubertal, five pubertal) with a median age of 11.3 years (range 3-15-5 years). These children were part of a control group in a study into the effects of cranial irradiation on hypothalamic-pituitary function. Six subjects had familial short stature (mid-parental height more than 1.3 SD below the mean) with a normal growth hormone (GH) response to standard provocative stimuli and a height velocity standard deviation score (SDS) between 0 and -0.8. The remaining eight subjects were normal siblings of children who had been treated for acute fymphoblastic leukaemia. A full history and clinical examination, including pubertal staging according to the method of Tanner (1962), was performed for each subject to exclude undiagnosed disease. None of the subjects was taking any medication. Ethical permission for the study was granted by the local ethical committee and written informed consent was given by the parents of each subject.

The children were admitted to hospital the night before sampling was started to acclimatize them to the hospital environment. They were encouraged to remain ambulant throughout the day and eat normal meals at appropriate times. Sampling was commenced in the morning, after insertion of an indwelling cannula into a forearm vein, and

21 5

216 W. H. B. Wallace e t a l .

was continued at 20-min intervals for 24 h. During the night, extension tubing was attached to the cannula to facilitate the collection of samples without disturbing the subjects’ sleep pattern. The total volume of blood collected over 24 h was limited to less than 10% of the subject’s estimated blood volume. Samples were collected into heparinized containers, centrifuged and separated every hour. The samples were stored at - 70°C until analysis.

Assays

ACTH was measured in unextracted plasma using a two-site IRMA (White et al., 1987) with a detection limit of 0.97 pmol/l. The assay employs two monoclonal antibodies, radiolabelled antibody 1A12 is bound to ACTHISIS and antibody 2A3, which is coupled to Sephacryl binds to ACTH24-39. The assay detects intact ACTH but not ACTH fragments.

Cortisol was measured in unextracted plasma, using incubation at low pH to eliminate binding by corticosteroid binding plasma proteins. The cortisol antiserum was prein- cubated with donkey anti-sheep antiserum and normal sheep serum (all supplied by the Scottish Antibody Production Unit) diluted in 0.13 M sodium chloride/phosphate buffer, pH 4.0. This reagent was incubated with cortisol-3-carboxy- methyloxime 1251 histamine and standard or sample for 90 min at 37°C before centrifugation, separation of the super- natant and counting of the precipitate in a multigamma counter. All samples from one individual were measured in duplicate in the same assay.

Data analysis

Both ACTH and cortisol were regularly sampled for 24 h in each patient so the available data form classical time series. Such series are open to both temporal and frequency domain analysis. Where underlying patterns of secretion are sus- pected, temporal analysis (by, for example, autocorrelation) is of value, and this has its counterpart, the estimation of univariate spectra, in the frequency domain. Most com- monly the temporal and frequency domains are related by Fourier transformation. This describes temporal data as a trigonometric least-squares fit. Quite simply, the sinusoidal functions required to fit the observed data are determined. Thus, if a given time series exhibits some form of regulated pulsatile behaviour, then the corresponding spectrum is likely to contain distinct peaks at particular frequencies or periodicities.

In addition to the extraction of features peculiar to cortisol or ACTH individually, we required an analytical extension that would indicate significant shared periodic features in paired cortisol and ACTH assays. The encapsulation of

information in the auto and cross-covariance data was a primary goal. Therefore, computation of the coherency spectrum was chosen as the most appropriate form of data analysis since it represents correlation between the spectral coefficients o f assays at given frequencies, w (i.e. periodici- ties). We calculated squared coherency, involving the smoothed estimate of the cross-spectrum, fca(w), for a cortisol-ACTH assay pair, and the smoothed univariate spectral estimates, f c (o) , fa(@) for cortisol and ACTH respectively. Bothfc(w) and fa(w)spectra are computed using the corresponding covariance data. The squared coherency, C (w), at frequency w is between zero and one and is given by

Scientific software libraries incorporate the necessary mathe- matical subroutines required to generate coherency spectra. We have utilized the National Algorithms Group Fortran library, section G13. This is available on the Christie Hospital and Paterson Institute’s DEC VAX computer LAN, forming part of the Manchester University Network. The analytical steps for our particular form of time series analysis involved stationarization of data and autoregressive modelling before actual computation of coherency results. The computational steps are outlined below. More detailed guidance for this type of analysis can be found in Priestley’s text on spectral analysis (Priestley, 198 1).

Stationarization of cortisol and ACTH data by differentiation Inspection of the functional form of the assay plots indicates the degree of differencing required. For example, where there are only linear trends in the data, first-order differencing will normally suffice. If quadratic behaviour is discernible, second-order processing can be applied.

Autoregressive-integrated moving average ( A RIMA) modell- ing of both cortisol and ACTH assays This is an iterative procedure, where the minimum number of model parameters is required for a satisfactory fit to assay data. Test criteria often operate by computing the residuals between the data and the model itself, followed by generation of a statistic measuring remnant correlation in the residuals. A good model fit leaves little correlation in the residuals. The NAG libraries provide for Box-Ljung testing of model adequacy (Ljung & Box, 1978).

Prewhitening of cortisol and ACTH data One of the ARIMA models is used to filter or ‘whiten’ the two data sets before cross-spectral computation. It may be appropriate to align the two modified series by shifting or lagging. The need for this can be determined by inspection of the cross-correla- tions. If any alignment ’ is required then the remaining

AC JH and cortisol secretion in children 217

ARIMA model is useful in predicting any necessary addi- tional data values.

Computation of smoothed cross and auto-spectra from modi- fied data The determination of these spectra is necessarily performed using the same computational parameters. Con- sideration must be given to the form of smoothing window utilized. We chose Parzen windowing. In addition, the window cut-off lag is important. Once again, the cut-off lag is best determined by inspecting results produced by varying the window size. In our case we chose a lag of 40, where the maximum window available was 72 (Priestley, 1981).

Computation of coherency spectrum and confidence limits Estimates of the squared coherency have values between zero and one. In the NAG implementation, the hypothesis that the cortisol and ACTH assays are unrelated, at any fre- quency w, may be rejected at the 5% level, if the squared coherency is greater than a returned level of coherency. For a

cut-off lag of 40 this is 0.7. To avoid mis-matching of infrequent sampling intervals to the predominant period of pulsatility, which is called aliasing, the minimum sampling rate for adequate representation of a continuous profile is formally specified in the ‘sampling theorem’ (Dainty & Shaw, 1974)..For 20-min sampling aliasing is not a problem for periods longer than 0.66 h or 1.5 cycles per hour.

Rather than use simple quadrature technique such as the trapezoidal method, the areas under the curves (AUC) forming the assay profiles were calculated by the Gill Miller Quadrature. The Mann-Whitney (I-test was used to com- pare unpaired data. Values in the text are given as median (range).

Results

The 24-h profiles for ACTH and cortisol from all 14 subjects are shown in Figs 1 and 2. The range of cortisol (Fig. 2) secretion is very similar to that described for adults

5 1

10

L 5300 I500 2100 0300 0900

0900 1500 2100 0300 0900 1500 2100 0300 0900 1500 2100 0300 0900 1500 2100 0300 0300

Time ( h 1

Fig. 1. The 24-h profiles for ACTH are displayed for all 14 normal children. The profiles are labelled with the subject’s number.

218 W . H. 13. Wallace et a / .

1000 1

600 i

2 3 4

I

t

. , ,0900 I500 2100 0300 0 9

14 0

Time ( h )

Fig. 2. The 24-h profiles for cortisol are displayed for all 14 normal children. The profiles are labelled with the subject’s number.

( < 50-600 nmol/l), with an early morning peak and progres- sive decline throughout the afternoon followed by a quies- cent period in the early evening which lasted up to 5 h. ACTH secretion, on the other hand, is below the lower limit of detection of the assay (0.97 pmol/l) for a large part of the profile. Secretion is however clearly episodic with several peaks in the 4.5-6.8 pmol/l range.

The normal circadian rhythm of ACTH and cortisol secretion was preserved. The median 0900 h and midnight ACTH values were 1.8 ( < 1.1-4.5 pmol/l) and < 0.97 pmol/l respectively. For ACTH, the midnight value was below the level of detection of the assay in all 14 subjects. For cortisol, the median 0900 h and midnight values were 296 (1 36-720) nmol/l and 62 ( < 48-33 1) nmol/l respectively.

In Table I we have displayed the median AUC for ACTH and cortisol for all subjects. It is evident that there are no significant differences for ACTH and for cortisol secretion between males (n=9) and females (n=5) and between prepubertal (n= 9) and pubertal (n = 5) subjects in our series.

In Table 2 the dominant periodicities present after Fourier transformation of the autocovariance data for ACTH and cortisol are shown. Cortisol shows two significant periodici- ties, at a period of 0.7-1.0 h (all 14 children) and at a longer periodicity of 2-3.2 h (in eight children). The shorter periodicity is equivalent to approximately 24-34 cortisol secretory episodes, while the longer periodicity is equivalent to about 8-12 secretoryepisodes in 24 h. For ACTH there is a clearer dominant periodicity for 13 of the 14 children at

ACTH and cortisol secretion in children 219

Table 1. Comparison of ACTH and cortisol secretion in normal children

ACTH (AUC) Cortisol (AUC) @mol/ I/h) (=ol/l/h)

Number of subjects Median (range) Median (range)

Total 14 29.7 (9-54) 5114 (3562-8631)

Males 9 30 (9-54) 4833 (3562-5766) Females 5 29 (1547) 5658 (3902-863 1)

NS NS

Prepubertal 9 29 (9-54) 5223 (4627-863 1) Pubertal 5 30 (1 247) 4290 (35624833)

NS NS

NS, Not significant.

Table 2. Fourier transform of autocovariance data for ACTH and cortisol

Subject

1 2 3 4 5 6 7 8 9

10 1 1 12 13 14

Cortisol dominant

periodicities

2.2, *1.1, 0.8 1.5, 3.0, *0.9 1.1, *3.2, 0.8 4.4, 1.4, *2.0. *0.9 6.9, *1.0, *2.4 *1.2, 6.0, '0.8 '0.9, 1.4 4.0, 1 .O, 1 .5 1.4, 4.0, 0.8 1.0, 2.7, 0.8 *0.9, 2.5, 0.7 5.3, 1.5, *1.0, *0.8 1.9, 4.4, 0.8 4.8, *1.0, 2.4

ACTH Coherency

periodicities periodicities dominant significant shared

*1.0, 0.7 ll.0, 3.2 *0.9, 1.3 *0.8 0.7 *3.0 *0.8, *2.0 Nil sig

*0.9, * 1.0, *0.7 Nil sig 1.7, *1.0 *O%, 2.6 0.9 Nil sig 0.8, 1.2 2.1, 3.2 Indeterminate Nil sig *1.0, 0.7 *1.0, 1.9, 2.5 *0.9, 2-4 *04, *1.3, 2.4 0.7, 0.8, 1.5 1.7 *0.9, 3.0 Nil sig

'1.0 * 1 '0, *2.4-2' I

Values are in hours. * Related periodicities.

0.7-1.0 h, equivalent to 2434 ACTH secretory bursts per 24 h. For five of the children there is an indication of possibly significant periodicities at or less than 0.66 h, equivalent to 36 secretory episodes in 24 h. However, because of aliasing, no conclusions can be drawn for periods of 0.66 h or less (Dainty & Shaw, 1974) with a sampling interval of 20 min.

To look for significant shared periodicities between ACTH and cortisol, we have used the technique ofcoherency (see data analysis). In Table 2 the significant shared periodi- cities are shown that have achieved a 95% confidence level.

In Fig. 3 the coherency spectra for each child is shown, and the periodicities that have achieved significance are indi- cated. For six of the children there is a significant shared periodicity of 0-8-1 -0 h, representing 24-30 related secretory episodes in 24 h for ACTH and cortisol. In a further five children there is a similar shared periodicity which has not achieved a 95% confidence level, and for the remaining three children there were no shared periodicities apparent.

An inspection of the cross-correlation results forming part of the input for coherency computation confirms a close relationship between ACTH and cortisol, with no discernible lag present for seven children and a lag of 20 min or less for the remainder. The sampling interval of 20 min precludes further definition of this relationship.

Dlscusslon

Despite the advent of ACTH radioimmunoassays over 20 years ago there have been few studies of the episodic secretion of ACTH and cortisol secretion in normal adults, and to our knowledge there have been no reports in children. In this study we have shown that the normal circadian rhythm of ACTH and cortisol secretion is preserved in children; for all the children the midnight ACTH value was below the limit of detection of the assay (0.97 pmol/l).

Using a novel form of time series analysis we have demonstrated that ACTH and cortisol secretion are inter- dependent, with a shared dominant periodicity of 098-1.0 h representing 24-30 related secretory bursts of ACTH and cortisol. In children neither pubertal status nor gender appears to influence ACTH or cortisol secretion. Chipman et al. (1982) measured 24-h cortisol secretory patterns in 13 children with juvenile rheumatoid arthritis. The mean 24-h ACTH concentrations were measured but were often less than the sensitivity of the radioimmunoassay (4.5 pmol/l).

Liu et al. (1987) and Mortola et al. (1987) have character- ized pulsatile ACTH and cortisol secretion in normal women using an unextracted ACTH radioimmunoassay. They reported between 6 and 13 pulses of ACTH in 24 h and found a mean ACTH concentration of 2.0+ 1.0 pmol/l. Link- nowski et al. (1985) studied ACTH and cortisol secretion in normal men using a similar ACTH assay. They found a mean ACTH level of 6.8 k4.3 pmol/l, considerably greater than our own findings for children, with between 14 and 20 pulses in 24 h. More recently Horrocks et al. (1990) have studied pulsatile ACTH and cortisol secretion over 24 h in 10 normal adults (five male: age range, 21-32 years) using the same IRMA for ACTH as in our own study. Assuming that ACTH clearance rates are not influenced significantly by sex, then it would appear that the adult males in their series secrete significantly more ACTH than the adult females

220 W . H, 5. Walface et a/.

I .n.

0.6

0.2

4.8 1.6 0.96 0.68 4.8 1.6 0.96 0.

71 4.8 1.6 0.96 0.68 13

I 4.8 1.6 0.96 0.68 4.8 1.6 0.96 0.68

Periodicity ( h )

Fig. 3. The coherency spectrum for each subject is displayed and significant shared periodicities are indicated with an arrow. Line X, shown on each coherency spectrum, represents the 95% confidence interval above which significance is achieved. The spectra are labelled with the subject's number.

(mean and AUC: P=0-004), but there were no significant differences between males and females for cortisol secretion. The AUC for cortisol secretion in our children is similar to that found for the adult subjects, but the adult males in their series secrete significantly more ACTH than our normal children suggesting that the ACTH level required to main- tain normal cortisol secretion may be lower in children and women than in men. The reasons for the apparent increase in ACTH secretion in adult men are not clear, but it may be due to a different set-point in cortisol feedback, or a fall in the sensitivity of the adrenal cortex to ACTH.

In the study by Horrocks et al. (1990), using the peak detection system of Clayton et al. (1987), the number of ACTH pulses, but not cortisol pulses, was significantly greater in the males than in the females (ACTH: males, 18 (14-22); females, 10 (7-15); cortisol: males, 12 (7-18); females, 15 (11-20)). Using a novel form of time series analysis we have found no sex differences in the periodicity of ACTH or cortisol secretion in children.

A recent study (Veldhuis et al., 1990) has attempted to delineate the mechanisms responsible for the prominent circadian rhythm of the ACTH-adrenal axis. Sampling every

ACTH and cortisol secretion in children 221

10 min for 24 h, eight normal men (aged 30-60 years) were studied. Plasma ACTH was measured using a sensitive (0.23 pmol/l), reproducible and specific two-site IRMA and cortisol was measured by RIA. Analysis of the data was by a deconvolution model designed to resolve quantitively the number, amplitude and duration of ACTH secretory bursts. They concluded that the observed 24-h plasma ACTH profile can be accounted for by a burst-like mode of ACTH secretion (40 secretory bursts/24 h), which by amplitude modulation (but not frequency control) gives rise to a 24-h rhythm in circulating ACTH concentrations. Our own findings for ACTH of a dominant periodicity of 0.7-1.0 h in children, with a trend towards 0.66 h in five children, are not dissimilar. Moreover it is likely that a shorter sampling interval and a more sensitive ACTH assay would improve the detection of small-amplitude ACTH secretory bursts and result in a reduction in the duration of the perceived dominant periodicity. In our view it is neither ethically nor technically possible to shorten the sampling interval to less than 20 min in young children over a period of 24 h.

In summary, we have shown that the normal circadian rhythm of ACTH and cortisol secretion is present in children, and both hormones are secreted with a period of 0-7-1.0 h. This represents 24-34 secretory bursts of ACTH and cortisol in 24 h. In children neither pubertal status nor gender appears to influence ACTH or cortisol secretion and it would appear that the level of ACTH required to maintain normal cortisol secretion may be lower in children than in adult males.

Acknowledgements

We would like to express our gratitude to Professor M. B. Priestley, Department of Mathematics, University of Man- Chester Institute of Science and Technology, for his time, invaluable advice and comments concerning the appropriate use of spectral analysis and interpretation of results.

We would also like to thank Dr P. H. Morris-Jones and Dr D. A. Price for permission to study children under their care.

This study was supported by a grant from the Lmkaemia Research Fund.

References

Chipman, J.J., Boyar, R.M. & Fink, C.W. (1982) Anterior-pituitary adrenal function of gold-treated patients with rheumatoid arthri- tis. Journal of Rheumurology 9 ,6348 .

Clayton, R.N., Royston, J.P., Chapman, J., Wilson, M., Obirai, M., Sawers, R. &Lynch, S.S. (1987) Is changing hypothalamic activity important for control of ovulation? British Medical Journal, 295,

Dainty, J.C. & Shaw, R. (1974) Fourier transforms. In Image Science. Academic Press.

Gallagher, T.F., Yoshida, K., Roffwarg, H.D., Fukushima, D.K.. Weitzman, E.D. & Hellman, L. (1973) ACTH and cortisol secretory patterns in man. Journal of Clinical Endocrinology and Metabolism, 36, 1058-1073.

Gibson, S., Littley, M.D., White, A., Richardson, P. & Shalet, S.M. (1988) Six-hour profiles of ACTH and cortisol in disorders of the hypothalamic-pituitary-axis. Journal of Endocrinology, 117(Suppl), 194.

Horrocks, P.M., Jones, A.F., Ratcliffe, W.A., Holder, G., White, A., Holder, R., Ratcliffe, J.G. & London, D.R. (1990) Patterns of ACTH pulsatility over 24 hours in normal males and females. Clinical Endocrinology, 32, 127-134.

Krieger, D.T., Allen, W., Rizzo, F. & Krieger, H.P. (1971) Characterisation of the normal pattern of plasma corticosteroid levels. Journal of Clinical Endocrinology and Metabolism, 32,266- 284.

Linkowski, P., Mendlewicz, J., Ldercq, R., Brasseur, M., Hubain, P., Golstein, J., Copinschi, G. 8r Van Cauter, E. (1985) The 24- hour profile of adrenocorticotrophin and cortisol in major depressive illness. Journal of Clinical Endocrinology and Metab- olism, 61,429438.

Liu, J.H., Kazer, R.R. & Rasmussen, D.D. (1987) Characterisation of the twenty-four hour secretion patterns of adrenocorticotro- phin and cortisol in normal women and patients with Cushing’s disease. Journal of Clinical Endocrinology and Metabolism, 64,

Ljung, G.M. & Box, G.E.P. (1978) On a measure oflack of fit in time series models. Biometrica, 65,297-303.

Mortola, J.F., Liu, J.H., Gillin, J.C., Rasmussen, D.D. & Yen, S.S.C. (1987) Pulsatile rhythms of adrenocorticotropin (ACTH) and cortisol in women with endogenous depression: evidence for increased ACTH pulse frequency. Journal of Clinical Endocrino- logy and Metabolism, 64,962-968.

Priestley, M.B. (1981) Spectral analysis and time series. Volumes 1 and 2. Academic Press.

Tanner, J.M. (1962) Growth at Adolescence. Blackwell Scientific Publications, Oxford.

Veldhuis, J.D., Iranmesh,A., Johnson, M.L. & Lizarralde, G. (1990) Amplitude, but not frequency, modulation of adrenocortico- tropin secretory bursts gives rise to the nyctohemeral rhythm of the corticotropic axis in mass. Journal of Clinical Endocrinology and Metabolism, 71,452-463.

White, A., Smith, H., Hoadley, M., Dobson, S.H. & Ratcliffe, J.G. (1987) Clinical evaluation of a two site immunoradiometric assay for adrenocorticotrophin in unextracted human plasma using monoclonal antibodies. Clinical Endocrinology, 26,41-52.

7-12.

1027-1035.