Clinical Endocrinology (1992) 36, 59-68

A novel variant of growth hormone (GH) insufficiency following low dose cranial irradiation

E. C. Crowne, C. Moore, W. H. 8. Wallace, A. L. Ogilvy- Stuart, G. M. Addison, P. H. Morris-Jones and S. M. Shalet Christie Hospital and the Royal Manchester Children’s Hospital, Manchester, UK

(Received 4 February 1991; returned for revision 21 May 1991; finally revised 18 September 1991; accepted 23 September l99lJ

Summary

OBJECTIVE We aimed to investigate the effect of low dose (1800 cGy) prophylactic cranial irradiation on physlologl- cal growth hormone secretion. DESIGN We performed an analysis of 24-hour serum GH profiles using 20-minute sampling. PATIENTS Forty-four children were studied, of whom 21 were long-term survivors of acute lymphoblastlc leukae- mia and 23 were normal children. They were further subdivided into prepubertal, pubertal and post-pubertal groups. MEASUREMENTS GH profiles were analysed by autocor- relation, Fourier transformation and spectral analysis of stationarlzed data, and peak detection using the Pulsar peak detection program. RESULTS In the normal children, there was a Significant increase in the median (range) area under the curve (AUC) of the GH profile between the prepubertal and pubertal groups (62 (11-124) and 137 (142-158) IU/l/h respectively, (P<O.Ol)). There was also a change in the spectral analysis through puberty. The dominant frequencies were spread widely In the prepubertal and post-pubertal groups but sharply focused in the pubertal group. In the cranially irradiated children there was no significant increase in AUC between the prepubertal (62(13-110) IU/l/h) and pubertal groups (92 (14-163) IUlllh). The wide range of dominant frequencies persisted In the pubertal cranially irradiated group due to the presence of additional high frequency pulses. The impression of a disturbance of the periodicity of OH secretion in the cranially Irradiated pubertal group was further supported by the finding that the autocorrelation function in this group alone was not slgnlflcantly different from that which would arise from random data.

Correspondence: S. M. Shalet, Christie Hospital, Wilrnslow Road, Withington, Manchester M20 9BX, UK.

CONCLUSIONS A novel form of OH lnsutfklency has been observed after low dose irradiation In childhood In which an abnormality of periodicity and a quantitative reduction In GH secretion appears restricted to puberty.

Cranial irradiation is now the commonest cause of organic growth hormone deficiency, the prevalance of which is dependent on the dose and fractionation schedule of the irradiation administered to the hypothalamic-pituitary axis (Shalet et al., 1976, 1979; Rappaport & Brauner, 1989) as well as the time elapsed since irradiation (Clayton & Shalet, 1991). A dose of 24-2500 cGy has been used as prophylactic cranial irradiation to protect against central nervous system (CNS) relapse in children with acute lymphoblastic leukae- mia (ALL) and resulted in a significant improvement in survival at the expense of long-term side-effects such as neuropsychological problems and GH deficiency. Abnor- malities in GH secretion reported following prophylactic cranial irradiation with this dose (24-2500 cGy) include both blunted GH responses to pharmacological stimuli in prepu- bertal and pubertal children (Kirk el al., 1987; Shalet et al., 1979) and a normal GH response to a pharmacological stimulus but subnormal spontaneous GH secretion (Blatt et al., 1984). In addition, reduced spontaneous G H secretion during 24-hour profiles has been described in both prepuber- tal and pubertal girls following cranial irradiation for ALL (2400 cGy) with a failure of the expected increase in GH secretion at puberty (Moell et af., 1989) associated with attenuated pubertal growth (Moell et al., 1987).

Since 1980 the dose of irradiation employed in prophylac- tic cranial irradiation for ALL has been reduced to 1800 cGy in an attempt to minimize neuropsychological disturbance and abnormalities of GH secretion (Nesbit et al., 1981). There are no detailed data on the effect of this irradiation dose on GH secretion. The aim of this study is therefore to compare spontaneous GH secretion using 24-hour GH profiles in a group of normal children and a group who have received low dose prophylactic cranial irradiation (1800 cGY).

Patients and methods

Forty-four children (aged 34-18.9 years) were admitted for 24-hour (0800-0800 h) serum GH profiles with a 20-minute sampling interval.

59

60 E. C. Crowne eta/. Clinical Endocrinology (1992) 36

Table 1 Median (range) age and pubertal stage of 44 children undergoing serum GH profiles

Endocrine study Cranial irradiation

Time Age Pubertal Age elapsed

n (years) stage (years) (years)

Group 2 Prepubertal I 1 8.9 (2.4)

Pubertal 7 13.1 (1.2)

Post-pubertal 5 17.0 (2.2)

Group 1 Prepubertal 6 9.3 (1.8)

Pubertal 8 12.7 (2.0)

Post-pubertal 7 14.9 (2.0)

4F 7M 1F:B3

1 F:B5 4M:TV 25 ml

6M:TV 4-12 ml

6M

3FB3-B4 5M:TV 5-12 ml 4F:B5 3M:TV 25 ml

-

-

-

3.5 (2.6-8.6)

6.5

8 4 (3’1-10.5)

(5.8-14.4)

-

-

-

4.8 (3.8-7.2)

5.9

6.4 (4.0-7.9)

(3.8-7’4)

B, Stage of breast development 1-5. TV, Testicular volume.

Group 1 consisted of 21 children who had undergone low- dose cranial irradiation (1800 Gy) as part of their treatment for ALL. They had all received chemotherapy according to standard UK Medical Research Council protocol (UKALL VIII) and were in first remission, a median time of 5.9 (3.8- 7.9) years from cranial irradiation.

Group 2 consisted of 23 normal children, 15 of whom were normal siblings of patients in group 1 and eight of whom were children with genetic short stature. The latter were growing with a normal height velocity (i.e. height velocity above the 25th centile), had a normal GH response to pharmacological stimulation tests, but had a standing height below the third centile and a mid-parental height below the 10th centile.

Groups 1 and 2 were further subdivided according to pubertal stage (Table 1). All the studies were approved by the local ethical committee and informed consent was obtained from the children and their parents.

Pubertal stage was assessed conventionally (Tanner & Whitehouse, 1976). Each child was admitted on the evening before investigation to acclimatize to the ward. During sampling, the subjects were ambulant and ate normal meals at approproate times. At night, extension tubing was attached to the cannula so that collection could continue without disturbing the subjects’ sleep patterns.

Samples were stored at 4°C overnight, spun down and separated the following day, and then serum samples stored

at -40°C until assayed. Serum GH profiles were analysed in one batch within a month of sampling using a standard diagnostic IRMA kit (Netria, St Bartholomew’s Hospital, London) and international reference standards.

Ana I ysis

The regularly sampled GH profiles form classical time series. Spectral analysis was devised to analyse such sequences, where evidence of regular pulsatile behaviour is masked by trends and random features. In this study we examined GH data for any changes in secretory pattern in the irradiated group using Fourier transformation, spectral analysis and autocorrelation of stationarized data.

The data series were also analysed using the Pulsar peak detection program (Merriam & Wachter, 1982).

Spectral analysis is of course not a new technique; rather, the availability of computers has made the approach more amenable. The analysis of time series using the priodogram was first introduced by Schuster in 1898. The approach utilizes the synthesis of time series using sine and cosine waves of different frequencies. Essentially, a trigonometric least-squares fit of the data is performed. This is Fourier modelling where the amplitudes of components at harmonics of a fundamental frequency are computed. The summed squares of amplitudes for a given frequency are termed the intensity. A plot of intensities at all frequencies is the

Clinical Endocrinology (1992) 36 GH deficiency after cranial irradiation 61

periodogram. Originally the periodogram was used to detect and estimate the amplitude of a sinusoidal component buried in random background data.

The sample spectrum can be thought of as a development of the periodogram, for use where frequencies are not harmonically related to the length of the time series.

Stationarized time series of the kind considered in this paper exhibit random changes of frequency, amplitude and phase. Hence, sample spectrum intensities fluctuate widely and are difficult to interpret. Given many experimentally determined time series for a given patient, it would be possibly to compute many sample spectra and intensities. The average of all the sample spectra would tend to a limiting ‘power spectrum’ commonly referred to as the ‘spectrum’. This is not normally a practical proposition. A single time series data set can, however, be analysed in a modified way to produce a smooth estimate of the spectrum. This involves the use of windowing, a subject that has been extensively investigated and reported in the literature. The analysis in this paper utilizes the Parzen window with a lag of40, for GH sequencies of 72 points. The estimated power spectra appear with ordinates labelled as ‘intensity’ (Priestley, 1988). Math- ematically, the power spectrum is the Fourier transform of the autocovariance function. A closely related function, the spectral density function, is the Fourier transform of the autocorrelation function.

It is clear that spectral analysis is well suited to the identification of patterns in pulsatile sequences. The pulse sequences are effectively modelled using sinusoidal basis functions and any dominant behaviour is revealed in the estimated power spectra. A lack of periodic behaviour can, however, also be assessed given a knowledge of purely random processes, called ‘white nose’. A time series is purely random if it consists of a sequence of uncorrelated random variables. The autocorrelation function of such a sequence is zero for all non-zero lags, with the same variance I / N for all lags. A purely random process has a power spectrum which has the same value at all frequencies; that is, it is flat. Quite clearly this feature can be used to test how closely a time series approaches randomness. (NB: the origin of the term ‘white noise’ is similar to that of ‘white light’, a spectrum of component colours; each colour is an electromagnetic wave of characteristic frequency and all colours are present).

A stationary time series can be usefully described by its mean, variance and autocorrelation function or equivalently by its mean, variance and spectrum. The autocorrelation function and the spectrum are mathematically equivalent and their relative advantages depend on their representatio- nal value. Autocorrelation functions contain redundant information where there is a dominant periodicity. The function will be peaked at multiples of the dominant period

so a 2-hour periodicity may result in peaks at 2,4,6, etc, hour lags. Where a range of periodicities exists in the parent time series, the result can be an autocorrelation function that is difficult to interpret.

The Pulsar peak detection program essentially identifies the baseline and then uses defined criteria to recognize peaks in the data array. Variations in hormone levels are associated with variations in the assay triples, fluctuations in the baseline, and variations in the pulse profiles themselves. The method screens out long-term changes and then uses a moving average to calculate the baseline. The peaks are identified by the calculation of yardsticks, C(n), where n is the number of measurements which might form the peak, that is, the width. The G(n) are calculated using statistical methods. The non-peak values, or ‘noise’ are taken as being random. A probability distribution is taken which fits this random series and Bonferroni confidence intervals used to identify the upper bounds of the noise.

Statistics

Differences between groups were analysed using the Mann- Whitney U test.

Estimated spectra were analysed separately and pooled by group. The autocorrelation functions were also pooled by group using the 2-transform technique. This results in single functions with larger effective sample sizes. These procedures more clearly reveal any randomization effects due to irradia- tion.

Autocorrelation functions were examined in two ways. The first approach looked for the presence of significant peaks, while the second tested for evidence of no periodic behaviour, using a randomization test (Box & Jenkins, 1976). This involves testing the hypothesis that the entire, true autocorrelation function is identical to zero; that is, is the measured autocorrelation function actually significant or is its behaviour random or ‘noise-like’ (see analytical section above). The autocorrelation function for random noise has a standard error

a,n-l/Z

for all autocorrelation function values (n=71 for an indi- vidual aurocorrelation function after differencing for statio- narization). Therefore, to identify implied periodic beha- viour, peaks should exist in the autocorrelation function with amplitudes between a and 2a (Box & Jenkins, 1976) and those greater than 2a are significant (P<O.O5).

The statistic

62 E. C. Crowne et al. Clinical Endocrinology (1992) 36

2.5 5.0 7.5 10.0 12-5 15.0 17.5 20.0 22.5

Sample

Fbi)

2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5

time ( h )

Fig. 1 Examples of GH profiles for n, normal children and i, children who have undergone low dose cranial irradiation. a, Prepubertal; b, pubertal; and c, post-pubertal. Note the more regular spacing of peaks in the normal groups.

up to a lag k much less than n was used to test the hypothesis that the true autocorrelation function could really be zero, indicating the randomness of the estimated autocorrelation function. Probabilities greater than 0.50 confirmed this hypothesis.

Results

There was no significant difference in the mean time elapsed since cranial irradiation between the prepubertal and puber- tal children of group 1 (4.8 (34-7.2) and 5.9 (4.0-7.9) years respectively) as the pubertal group were significantly older when treated (3.5 (26-8.6) and 6.5 (3.1-10.5) years respect- ively, P=0.04).

Sample GH profiles from the three normal and three irradiated groups are shown in Fig. 1.

Area under the curve (AUC) of the GH profile

Figure 2 shows the results of the mean AUC of the GH profile, or total amount of GH secreted by the different groups during the study. There was a significant increase in

- 100 r $ 00 - 60 2 40

20

Fig. 2 Mean (SEM) area under the curve (AUC) of 24-hour GH profiles of a, normal children and b, children who have undergone low dose cranial irradiation (1800 cGy) sub-divided according to pubertal stage. Areas were computed using the Gill Miller quadrature method. m, Prepubertal; 0, pubertal; pubertal.

the total amount of GH secreted between the prepubertal and pubertal groups of the normal children (mean (range); 66 (1 1-124) and 118 (42-158) IU/l/h respectively, P<O.O1), but there was no such significant increase in AUC between the prepubertal and pubertal groups of those who had had low dose cranial irradiation (62 (13-1 10) and 94 (14-163) IU/l/h respectively).

Clinical Endocrinology (1992) 36 GH deficiency after cranial irradiation 63

r E - c .-

1 1 I I I I I t J

I00 2 5 10‘ Period ( h I

FIg. 3 Pooled estimated power spectra for the GH profiles of a, prepubertal; b, pubertal; and c, the post-pubertal groups. Results for both 0, normal children and A, those treated with low dose cranial irradiation are depicted. Note the spectral flattening for both pubertal and post-pubertal irradiated groups. In particular, the pubertal irradiated group exhibits increased spectral intensities for all short periods, indicating a randomization of GH pulsing.

64 €. C. Crowne et a / . Clinical Endocrinology (1992) 36

I I I I 1

0.2 0'3 F\ (

0'3Kc 0.2 '

Time lag ( h

Fig. 4 Pooled autocorrelation functions from the GH profiles of normal a, prepubertal; b, pubertal; and c, post- pubertal groups. Note the regularity of peaks in all cases. The u and 20 lines for a random series of the same effective (pooled) number of samples are presented as dashed lines. Peaks above the u and below 2u suggest periodic behaviour (see text and Box &Jenkins, 1976). Individual normal patient autocorrelation functions all presented significant peaks.

Fourier transforms showed a spectrum markedly different from that of the

Figure 3 shows the pooled estimated power spectra for the six study groups. The prepubertal normal group had a wide peak encompassing periods between 1.25 and 2.5 hours; in the pubertal normal group the results were focussed about a dominant period of 2.5 hours, whilst the post-pubertal normal group once again had a wide peak covering periods between 1.25 and 2.5 hours.

The prepubertal group who had had cranial irradiation produced a pooled spectrum similar to that of the normal group, with a wide range of periods, although spectral

corresponding normal group. Spectral intensities at the shorter periods actually increase well beyond those found for the normal group. This acts to flatten the spectrum and strongly suggests a randomization of GH pulsatility.

The post-pubertal group who had had cranial irradiation also showed strong spectral changes that tended to flatten the spectral estimate. The intensities at all periods were reduced, although a clear peak between periods of 1.25 and 2.5-3.0 hours remained. This suggests randomization similar to that found for the pubertal group, but on a smaller scale.

intensities at the shorter periods were reduced. Spectral peaking was still clearly in evidence indicating strong pulsatile behaviour.

The pubertal group who had had cranial irradiation

Autocorre'ation

The individual autocorrelation functions for the prepubertal groups, normal and irradiated, showed significant peaks for

Clinical Endocrinology (1992) 36 GH deficiency after cranial irradiation 65

0. I

0.0

-0.1

- 0.2

Flg. 5 Pooled autocorrelation functions from the GH profiles of irradiated a, prepubertal; b, pubertal; and c, post- pubertal groups. The u and 2a lines for a random series of the same effective number of samples are represented as dashed lines. For both pubertal and post-pubertal groups there is not even the suggestion of common periodic behaviour, with essentially all peaks below the u line. The prepubertal group displays results similar to the normal group autocorrelation.

I I I 1 1 2 4 6 8 10

Time lag ( h )

Table 2 Mean probabilities that the autocorrelation functions in the six study groups could have arisen from a random time series

Table 3 Mean amplitude and number of pulses in GH profiles from irradiated and control groups (mean f SEM)

Prepubertal Pubertal Post-pubertal

Study group 2 1 2 1 2 1 Probability 0.44 0.45 0.47 0.63 0.34 0.42

Mean pulse amplitude (mU/l) No. of peaks/24 h

many periods (lags). Figure 4 shows the results of pooling the functions in the normal groups, revealing peaking at mul- tiples of the approximately 2-hour period identified from spectral analysis. Figure 5 shows the pooled autocorrelation function for the irradiated groups. Peaks spaced by approxi- mately 2 hours are again evident. The significance of peaks can be assessed from the cr and 20 (see analysis section).

Group 2 Prepubertal I I 8.5 f 1.5 7.4f2.6 Pubertal I 10.8 f 3.4 %Of 1.8 Post-pubertal 5 7.6 f 2.8 5 4 f 1.7

Group I Prepubertal 6 10.0 f 3.9 5.5 f 2.4 Pubertal 8 8.4 f 2.4 7.8f 1.4 Post-pubertal 7 10.6 f 2.1 6.4 f 2.5

66 E. C. Crowne et al. Clinical Endocrinology (1992) 36

In the pubertal normal group the individual autocorrela- tion functions display significant peaks mainly below periods of 7 hours. This feature is preserved on pooling (Fig. 4). In the irradiated pubertal group the individual autocorrelation functions had peaks only at long periods (high lags), suggesting a randomization effect of cranial irradiation on GH secretion. The pooled autocorrelation substantiates this conclusion (Fig. 5). From the pooled function it is quite clear that not only are there no significant peaks but also there is no suggestion of common periodicity. All the autocorrela- tion values, except one at 8.5 hours, are below the Q value for random data.

In the post-pubertal group similar behaviour was found for the autocorrelation functions, both individual and pooled. The pooled data for normal and irradiated post- pubertal are shown in Figs 4 and 5 respectively. The irradiated group demonstrate no significant peaks, with all autocorrelation values below the cr line. This is strong evidence for randomization.

Table 2 shows the mean probabilities that the autocorrela- tion functions of the six groups were actually zero, that is, how closely they approximate the autocorrelation function of a random time series. The pubertal irradiated group has a mean probability of 0.63 & 0.1 1, which is greater than 0.5, so further confirming strong randomization effects in this irradiated group. It is clear from these results that the effect is not so strong in the post-pubertal group, but nevertheless still evident from the pooled spectra and autocorrelation results.

Pulsar peak detection

Results from the Pulsar analysis are shown in Table 3. There were large variances for pulse width, amplitude and inter- pulse interval within individual subjects. No significant differences in mean number of pulses or mean pulse ampli- tudes were detected between the irradiated and control groups.

Discussion

GH deficiency, as defined by a subnormal response to a pharmacological stimulus, is a very common occurrence after the higher doses of cranial irradiation used in the treatment of children with brain tumours, i.e. in excess of 3000 cGy to the hypothalamic-pituitary axis (Clayton & Shalet, 1991; Darendilier et al., 1990), the abnormal GH response reflecting an underlying reduction in spontaneous GH secretion (Ahmed & Shalet 1985; Lannering & Alberts- son-Wiklund, 1987).

Cranial irradiation doses in the region of 2400-2500 cGy

were used prophylactically in children with ALL, and have been associated with similar abnormalities in GH secretion, although less frequently (Clayton & Shalet, 1991; Rappaport & Brauner, 1989; Shalet et al., 1979). In addition, the phenomenon of GH neurosecretory dysfunction, consisting of reduced spontaneous GH secretion in the presence of normal GH responses to pharmacological stimuli (Spiliotis et al., 1984), has also been described after this intermediate radiation dose (Blatt et al., 1984) although the exact prevalance has not yet been defined. For these reasons, when we wished to examine the impact of the low dose cranial irradiation (1 800 cGy) currently used prophylactically in children with ALL, we chose to study physiological GH secretion over 24 hours.

A further critical factor in the onset of GH deficiency is the time interval between irradiation and study. It has been shown that the incidence of GH deficiency, defined by pharmacological tests, is influenced markedly by the post- irradiation time interval (Clayton & Shalet, 1991). In the present study, however, the latter was not significantly different between the prepubertal and pubertal ALL groups.

Age at irradiation is another, but more controversial, variable (Brauner et al., 1986; Clayton & Shalet, 1991) and it may be that the youngest at the time of irradiation are more vulnerable to irradiation-induced damage to the hypothala- mic-pituitary axis. In the present study it is the pubertal ALL group alone who show abnormalities of GH secretion despite the fact that they were significantly older than the prepubertal group at irradiation.

Moell et al. (1989) studied spontaneous GH secretion in both prepubertal and pubertal girls who received prophylac- tic cranial irradiation (2400 cGy) for ALL. GH secretion was greatly reduced in both prepubertal girls and pubertal girls, failing to show the normal and expected amplitude-modu- lated rise in response to the increase in circulating sex-steroid levels associated with puberty (Mauras et al., 1987, Martha et al., 1989).

The GH observations by Moell et al. (1989) are more remarkable when seen in the context of their growth data (Moell et al., 1987). Despite subnormal GH secretion, the prepubertal girls grew normally, but the pubertal girls exhibited an attenuated growth spurt and lost one standard deviation in standing height over the duration of puberty (Moell et al., 1987).

In our study, the results differ from those obtained by Moell et al. (1989) in that spontaneous GH secretion was normal in the prepubertal group following low dose cranial irradiation (1800 cGy). Although sufficient growth data are not available, our results imply that there will be very few prepubertal children treated by. the low dose radiation schedule for whom G H replacement therapy will be required.

Clinical Endocrinology (1992) 36 GH deficiency after cranial irradiation 67

Our pubertal group did, however, show abnormalities of spontaneous G H secretion despite the fact that each under- went puberty spontaneously and showed normal pubertal progression. Furthermore, other studies have shown normal sex steroid levels a t the completion of puberty in individuals similarly treated (Wallace et al., 1990). The absence of the expected increase in G H secretion during puberty therefore presumably reflects the failure of G H control mechanisms in the hypothalamus to respond appropriately to the rise in endogenous sex steroids.

Apart from the reduction in the total amount of G H secreted, there was also a significant disturbance of the periodicity of G H secretion in the irradiated pubertal children. The pulsatile nature of G H secretion, found in all species so far studied, underlines its physiological signifi- cance, although its quantitative relationship to growth has not been defined. Animal model studies have shown increased production of IGF-I mRNA in both rib growth- plates and skeletal muscle after pulsatile compared with continuous G H administration (Jansson et al., 1989), and the different growth patterns between male and female rats may be explained by different patterns of G H secretion, females showing a higher baseline and lower G H pulses than the males (Jansson ei al., 1985).

GH secretion is under the control of two peptide factors; growth hormone releasing factor (GRF) and somatostatin (SMS). G H pulses are believed to occur as a consequence of an increase in G R F and decrease in SMS secretion. The subsequent increase in SMS secretion is important in that it ‘ends’ the pulses of GH secretion and maintains the trough periods when G H secretion is undetectable. This prevents pituitary densensitization and allows the production of the next G H pulse (Tannenbaum, 1989). Bercu e t al. (1986) reported raised basal levels of G H with low amplitude G H pulses in a child following cranial irradiation for a brain tumour. Studies in rats have shown that lesions that deplete SMS in the median eminence result in a raised G H baseline and more frequent bursts of G H secretion (Willoughby & Martin, 1978; Willoughby et al., 1977). A disturbance in SMS secretion might explain the randomized bursts of G H secretion seen in our pubertal irradiated group.

Clearly this is an area that requires further study. In particular, we need to establish the growth pattern of these children during puberty. In addition we must determine if these abnormalities of G H secretion affect all or just a minority of children following low dose cranial irradiation. Our results d o not allow us to make this distinction. Nevertheless the observations are novel in that we have described for the first time an abnormality of periodicity and quantitatively reduced G H secretion in childhood restricted to puberty.

Acknowledgements

We are grateful for the support of the Leukaemia Research Fund and Eli Lilly during this study, and t o Dr D. A. Price for allowing us to study children under his care.

References

Ahmed, S.R. & Shalet, S.M. (1986) The effects ofcranial irradiation on growth hormone secretion. Acia Paediairica Scandinavica, 75, 255-260.

Bercu, B.B., Schulman, D., Root, A.W. & Spiliotis, B.E. (1986) Growth hormone (GH) provocative testing fequently does not reflect endogenous GH secretion. Journal of Clinical Endocrino- logy and Meiabolism, 63, 709-716.

Blatt, J., Bercu, B.B., Gillin, J.C., Mendelson, W.B. & Poplack, D.G. (1984) Reduced pulsatile growth hormone secretion in children after therapy for acute lymphoblastic leukaemia. Journal of Pediatrics, 104, 182 -186.

Box, G.E.P. &Jenkins, G.M. (1976) Time Series Analysis: forecasi- ing and conirol. Holden-Day, San Francisco.

Brauner, R., Czernichow, P. & Rappaport, R. (1986) Greater susceptibility to hypothalamopituitary irradiation in younger children with acute lymphoblastic leukaemia. Journal of Pedia- irics, 108, 332.

Clayton, P.E. & Shalet, S.M. (1991) Radiation-induced growth hormone deficiency: speed of onset is dose dependent. Journal of Pediairics, 118, 226-228.

Darendilier, F., Livesey, E.A., Hindmarsh, P.C. & Brook, C.G.D. (1 990) Growth and growth hormone secretion in children follow- ing treatment of brain tumours with radiotherapy. Acia Paedia- trica Scandinavica, 79, 950-956.

Jansson, J-O., Eden, S. & Isaksson, O.G.P. (1985) Sexual dimor- phism in the control of growth hormone secretion. Endocrine Reviews, 6, 128- 150.

Jansson, J-O., Isaksson, O.G.P., Eden, S., Isgaard, J., Carlsson, L. & Ekberg S. (1989) Effects of plasma GH pattern on growth factors and body growth. In Hormonal Regulation of Growth (eds H. Frisch & M. 0. Thorner), pp. 185-199. Serono Publications no. 58. Raven Press, New York.

Kirk, J.A., Ragnupathy, P., Stevens, M.M., Cowell, C.T., Menser, M.A.. Bergin. M., Tink, A,, Vines, R.H. & Silink, M. (1987) Growth failure and growth-hormone deficiency after treatment for acute lymphoblastic leukaemia. Lancet, i, 190- 193.

Lannering, B. & Albertsson-Wiklund, K. (1987) Growth hormone release in children after cranial irradiation. Hormone Research, 27,

Martha, P.M., Rogol, A.D., Veldhuis, J.D., Kerrigan, J.R., Good- man, D.W. & Blizzard, R.M. (1989) Alterations in the pulsatile properties of circulation growth hormone concentrations during puberty in boys. Journal of Clinical Endocrinology and Meiab- o h m , 69,563-570.

Mauras, N., Blizzard, R.M., Link, K., Johnson, M.L., Rogol, A.D. & Veldhuis, J.D. (1987) Augmentation of growth hormone secretion during puberty: evidence for a pulse amplitude-modu- lated phenomenon. Journal of Clinical Endocrinology and Meiab- olism, 64, 596-601.

Merriam, G.R. & Wachter, K.W. (1982) Algorithms for the study of episodic hormone secretion. American Journal of Physiology, 243,

13-22.

E3 10-E3 18.

68 E . C. Crowne et al. Clinical Endocrinology (1992) 36

Moell, C., Ganvicz, S., Westgren, U. & Wiebe, T. (1987) Disturbed pubertal growth in girls treated for acute lymphoblastic leukae- mia. Paediatric Hematology and Oncology, 4, 1-5.

Moell, C.. Garwicz, S., Westgren, U., Wiebe, T. & Albertsson- Wiklund, K. (1989) Suppressed spontaneous secretion of growth hormone in girls after treatment for acute lymphoblastic leukae- mia. Archives of Disease in Childhood, 64,252-258.

Nesbit, M.E. Jr, Sather, H.N., Robison, L.L., Ortega, J., Littman, P.S., D’Angio, G.J. & Denman Hammond, G. (1981) Presympto- matic central nervous system therapy in previously untreated childhood acute lymphoblastic leukaemia: comparison of 1800 and 2400 rad. Lancet, i, 461-466.

Priestley, M.B. (1988) Spectral Analysis and Time Series. Academic Press.

Rappaport, R. & Brauner, R. (1989) Growth and endocrine disorders secondary to cranial irradiation. Pediatric Research, 25,

Shalet, S.M., Beardwell, C.G., Pearson, D. & Morns-Jones, P.H. (1976) The effects of varying doses of cerebral irradiation on growth hormone production in childhood. Clinical Endocrino-

Shalet, S.M., Price, D.A., Beardwell, C.G., Morris-Jones, P.H. & Pearson, D. (1979) Normal growth despite abnormalities of growth hormone secretion in children treated for acute leukaemia. Journal of Pediatrics 94, 719-722.

561 -567.

logy, 5, 287-290.

Spiliotis, B.E., August, G.P., Hung, W., Sonis, W., Mendelson, W. & Bercu, B.B. (1984) Growth hormone neurosecretory dysfunc- tion a treatable cause of short stature. Journal of the American Medical Association, 251,2223-2230.

Tannenbaum, G.S. (1989) Interrelation of growth hormone-releas- ing factor and somatostatin in the regulation of growth hormone secretion. In Hormonal Regulation of Growth (eds. H. Frische & M. 0. Thorner), pp. 1-17. Serono Publications no. 58. Raven Press, New York.

Tanner, J.M. & Whitehouse, R.H. (1976) Clinical longitudinal standards for height, weight, height velocity, weight velocity and stages in puberty. Archives of Disease in Childhood, 51, 170-179.

Wallace, W.H.B., Shalet, S.M., Lendon, M. & Morris-Jones, P.H. (1990) Male fertility in long term survivors of acute lymphoblastic leukaemia (ALL) in childhood. Hormone Research, 33 (Suppl. 3), 4 (abstr.).

Willoughby, J.O., Terry, L.C., Brazeau, P. & Martin, J.B. (1977) Pulsatile growth hormone, prolactin, and thyrotropin secretion in rats with hypothalamicdeafferentation. Brain Research, 127,137- 152.

Willoughby, J.O. & Martin, J.B. (1978) Pulsatile growth hormone secretion: inhibitory role of medial preoptic area. Brain Research, 148,240-244.