British Journal of Haematology, 1999, 106, 368370

SHORT REPORT

Circadian variation of granulocyte colony

stimulating factor levels in man
B. J I L MA , 1 N. H E RGOVIC H , 1 P. S TO HLAW ETZ , 2 H.-G. E I C H L E R , 1 P. B AU ER 3 A N D O. F. WAG N E R 4
Departments of Clinical Pharmacology, TARGET, 2 Transfusion Medicine, 3 Medical Statistics,
and 4 Medical and Chemical Laboratory Diagnostics, University Hospital of Vienna, Austria
1

Received 26 April 1999; accepted for publication 30 April 1999

Summary. Glucocorticoids dose-dependently increase

plasma levels of granulocyte colony stimulating factor
(G-CSF). Based on the marked circadian rhythm of cortisol
levels, we hypothesized that plasma levels of G-CSF may also
show a diurnal rhythm. A prospective study was conducted
in 12 healthy young volunteers. Blood samples were
obtained every 2 h over 24 h. G-CSF levels averaged
180 ng/l (CI 131229) at 8.00 am, increased continuously and reached peak values at 10.00 p.m. Individual

harmonic regression analysis showed a clear circadian

rhythm. The individual differences between nadir and peak
levels averaged 54% (CI 4365%). This pronounced diurnal
rhythm of G-CSF levels may help understand the circadian
changes in circulating stem cells, bone marrow DNA
synthesis, or bone marrow toxicity induced by some
chemotherapeutic agents.

Glucocorticoids alter the function and number of neutrophils (Mishler & Emerson, 1977; Liles et al, 1997;
Jendiroba et al, 1998). Recently it has been shown that
glucocorticoids enhance the release of colony-stimulating
activity in vitro (Nissen et al, 1983), and increase plasma
levels of granulocyte colony stimulating factor (G-CSF) in a
dose-dependent fashion (Jilma et al, 1998).
Endogenous cortisol levels exhibit a marked circadian
rhythm (Pietrowsky et al, 1994). As exogenous glucocorticoids enhance G-CSF levels, we hypothesized that the
diurnal changes in endogenous cortisol may correlate with
circadian changes in G-CSF levels. Therefore we investigated
the diurnal changes in G-CSF levels.

determine health status. Fasting healthy subjects (m/f 6/6)

reported at the research ward after an overnight sleep at
7.45 a.m. A 18-gauge catheter was inserted into a forearm
vein, and blood samples were obtained at 2 h intervals
(Fig 1). Standardized meals were supplied after blood
samplings at 8.00 a.m., 12.00 a.m. and 6.00 p.m. All
subjects were allowed to move around the research ward,
but had to rest for at least 10 min before blood samplings.
Subjects were requested to sleep after lights had been
turned off at 11.00 p.m.; lights were turned on again at
7.00 a.m.
Venous EDTA plasma was centrifuged at 2000 g for
15 min. All samples were determined as duplicates and
samples from individual subjects were run in the same
assay. Cortisol levels were determined by enzyme immunoassay (AutoDela Cortisol EG&G, Wallac, Finland). G-CSF
levels were analysed with a high-sensitivity EIA (R&D
Systems; Oxon., U.K.; Jilma et al, 1998). G-CSF receptor
expression on neutrophils was quantied by staining of
whole-blood samples at 4 h intervals with a phycoerythrinlabelled anti-CD114 antibody (Serotech, Kidlington, U.K.)
and subsequent owcytometric analysis on a FACSscan
(Becton Dickinson, San Jose, Calif.). Differential blood
counts were measured with a counter (SYSMEX 2000,
Japan). Data is expressed as the mean and 95%
condence intervals. A harmonic regression analysis was
performed.

METHODS
After approval by the Ethics Committee, informed consent
was obtained. The study design was prospective and analyst
blinded. Inclusion criteria were absence of detectable disease,
1940 years of age and a normal body mass index. A battery
of standard laboratory and clinical tests were performed to
Correspondence: Dr Bernd Jilma, The Adhesion Research Group
Elaborating Therapeutics, Department of Clinical Pharmacology,
TARGET, Vienna University Hospital School of Medicine, Wahringer
Gurtel 1820, A-1090 Wien, Austria. e-mail: bernd.jilma@
univie.ac.at.

368

Keywords: circadian, diurnal, G-CSF, cortisol, plasma.

q 1999 Blackwell Science Ltd

Short Report
RESULTS
Results from one subject were retrospectively excluded
from analysis because he exhibited abnormalities in the
differential blood count on the study day. His G-CSF plasma
levels exceeded 60 ng/l, probably indicating a subclinical
infection.
Basal cortisol levels averaged 472 nmol/l (CI 382562)
and showed the well-known circadian variation (Fig 1).
G-CSF levels averaged 180 ng/l (CI 131229) at
8.00 a.m. All individuals except one exhibited a signicant
harmonic variation in G-CSF levels over the 24 h period. In
ve subjects the test of the parameter for the cosine term
showed a P-value as small as < 0001. The average
parameter estimate for the cosine term was 27 6 22
(SD) (t-test against zero mean: P 0002). The average
amplitude, calculated from the 11 individual amplitudes,
was 634 ng/l (range 607 to 667 ng/l), so that the
estimated daily variation was 35% of the daily mean
(194 ng/l). The mean individual min.max. differences
averaged 54% (CI 4365%) of the minimum. The average
phase of the harmonic oscillation was close to zero (the time
scale starting at 8.00 a.m.), indicating that the nadir values
of our sample were measured early in the morning. However,
the phase of the harmonic oscillation varied markedly
between individuals, suggesting that the circadian rhythm
is shifted between individuals.
Neutrophil counts averaged 28 109/l (CI 2433) at

Fig 1. Diurnal changes (D%) of cortisol, granulocyte colony

stimulating factor (G-CSF) plasma levels and neutrophil counts
(mean 6SEM from 11 healthy subjects).

369

8.00 a.m. and demonstrated a circadian rhythm (Fig 1). On

average, 99% of neutrophils were CD114 and the mean
uorescence intensity (MFI) was 63 (CI 4085) at 8.00 a.m.,
both of which did not show diurnal changes (max. MFI
change 2%; CI 9 to 4%; data not shown).
DISCUSSION
This is the rst systematic investigation showing that
G-CSF levels exhibit a diurnal rhythm (Fig 1). The
individual difference between the nadir and the acrophase was 54%. Therefore our data are in good agreement
with a previous report of approx. 40% higher G-CSF levels
at 9.00 p.m. than at 7.00 a.m. in nine subjects (Takeuchi
et al, 1996). However, these authors measured G-CSF
levels only at three times of the day, and did not study
whether basal levels were reached again on the following
morning.
Recently we found a dose-dependent increase in G-CSF
levels 24 h after infusion of dexamethasone 004 mg/kg
(corresponding to 90 mg cortisol equivalent) or 1 mg/kg
b.i.d.: G-CSF levels increased by 240% and by 870%,
respectively (Jilma et al, 1998). Thus, the magnitude of the
observed diurnal G-CSF change appeared to be proportional
to the circadian changes of endogenous cortisol production.
However, any relation of circadian G-CSF changes to diurnal
cortisol changes must be considered as a descriptive pattern
and not proof of a causal relation. Yet our trial represents a
sound rationale for further studies on the regulation of
circadian changes in G-CSF levels.
Clearance of G-CSF is determined in large part by the
absolute neutrophil counts (Cebon et al, 1994), because
G-CSF is cleared by binding to its receptor on leucocytes.
In our study neutrophil counts increased and G-CSF receptor
expression on neutrophils was unchanged. Therefore it is
unlikely that decreased clearance contributed to the diurnal
increase in G-CSF levels.
As the biological activity of G-CSF occurs at picomolar
concentrations (Khwaja et al, 1993), it is possible that
circadian changes of G-CSF levels bear biological relevance.
Lasky et al (1983) found that circulating uncommitted
pluripotential precursors (CFU-GEMM) formed more colonies
late in the day. Other studies have shown a circadian
variation in human bone marrow DNA synthesis which
peaks at noon (Smaaland, 1997). It is therefore possible that
these proliferative changes are mediated by the changes in
G-CSF levels. This may help explain why bone marrow
toxicity induced by chemotherapeutic agents may markedly
differ depending on the time of administration (Smaaland,
1997), and is important because glucocorticoids are
frequently part of a chemotherapeutic regimen, and may
alter toxicity (Smaaland, 1997), conceivably by virtue of
G-CSF release. Finally, a recent study in mice suggests that
the pharmacokinetics and pharmacodynamics of G-CSF
may underlie considerable diurnal variations (Ohdo et al,
1998).
Further studies are warranted to delineate the biological
consequences of the marked circadian changes in G-CSF
levels.

q 1999 Blackwell Science Ltd, British Journal of Haematology 106: 368370

370

Short Report

ACKNOWLEDGMENTS
This study was supported by the Austrian National Bank.
We are indebted to Dr T. Lang and Marika Fritz for their
assistance.
REFERENCES
Cebon, J., Layton, J.E., Maher, D. & Morstyn, G. (1994) Endogenous
haemopoietic growth factors in neutropenia and infection. British
Journal of Haematology, 86, 265274.
Jendiroba, D.B., Lichtiger, B., Anaissie, E., Reddy, V., O'Brien, S.,
Kantarjian, H. & Freireich, E.J. (1998) Evaluation and comparison
of three mobilization methods for the collection of granulocytes.
Transfusion, 38, 722728.
Jilma, B., Stohlawetz, P., Pernerstorfer, T., Eichler, H.G., Mullner, C. &
Kapiotis, S. (1998) Glucocorticoids dose dependently increase
plasma levels of granulocyte colony stimulating factor in man.
Journal of Clinical Endocrinology and Metabolism, 83, 10371040.
Khwaja, A., Carver, J., Jones, H.M., Paterson, D. & Linch, D.C. (1993)
Expression and dynamic modulation of the human granulocyte
colony-stimulating factor receptor in immature and differentiated
myeloid cells. British Journal of Haematology, 85, 254259.
Lasky, L.C., Ascensao, J., McCullough, J. & Zanjani, E.D. (1983)
Steroid modulation of naturally occurring diurnal variation in
circulating pluripotential haematopoietic cells (CFU-GEMM).
British Journal of Haematology, 55, 615622.

Liles, W.C., Huang, J.E., Llewellyn, C., SenGupta, D., Price, T.H. &
Dale, D.C. (1997) A comparative trial of granulocyte-colonystimulating factor and dexamethasone, separately and in combination, for the mobilization of neutrophils in the peripheral
blood of normal volunteers. Transfusion, 37, 182187.
Mishler, J.M. & Emerson, P.M. (1977) Development of neutrophilia by
serially increasing doses of dexamethasone. British Journal of
Haematology, 36, 249257.
Nissen, C., Moser, Y., Speck, B., Burgin, M. & Bendy, H. (1983)
Dexamethasone enhances `CSA' release and depresses `BPA'
release. British Journal of Haematology, 53, 301310.
Ohdo, S., Furukubo, T., Arata, N., Yukawa, E., Higuchi, S.,
Nakano, S. & Ogawa, N. (1998) Inuence of dosing time on
pharmacological action of G-CSF in mice. Life Science, 62,
PL163PL168.
Pietrowsky, R., Meyrer, R., Kern, W., Born, J. & Fehm, H.L. (1994)
Effects of diurnal sleep on secretion of cortisol, luteinizing
hormone, and growth hormone in man. Journal of Clinical
Endocrinology and Metabolism, 78, 683687.
Smaaland, R. (1997) Chronobiology and chronopharmacology of
the hematopoietic system. Physiology and Pharmacology of
Biological Rhythms (ed. by P. H. Redfern and B. Lemmer),
pp. 487555. Springer, Berlin.
Takeuchi, M., Tanimoto, Y., Tada, A., Soda, R. & Takahashi, K.
(1996) Circadian rhythm of plasma levels of endogenous
granulocyte colony stimulating factor in healthy volunteers.
(Letter). International Journal of Hematology, 64, 293295.

q 1999 Blackwell Science Ltd, British Journal of Haematology 106: 368370