Circadian variation in human cerebrospinal fluid production measured by magnetic resonance imaging CHRISTER NILSSON, FREDDY OLE HENRIKSEN, MARGARETE

STAHLBERG, HERNING,

CARSTEN THOMSEN, AND CHRISTER OWMAN

Department of Medical Cell Research, Section of Neurobiology and Department of Radiation Physics, University of Lund, S-223 62 Lund, Sweden; and Danish Research Center for Magnetic Resonance, University of Copenhagen, Hvidovre Hospital, DK-2650 Hvidovre, D&mark Nilsson, Christer, Freddy Sthhlberg, Carsten Thomsen, Ole Henriksen, Margarete Herning, and Christer Owman. Circadian variation in human cerebrospinal fluid production measuredby magnetic resonanceimaging. Am. J. Physiol. 262 (Regulatory Integrative Comp. Physiol. 31): R20R24, 1992.-Recent advances in magnetic resonanceimaging have madeit possibleto visualize and quantify flow of cerebrospinal fluid (CSF) in the brain. The net flow of CSF through the cerebral aqueduct was usedto measureCSF production in six normal volunteers at different times during a 24-h period. CSF production varied greatly both intra- and interindividually. The averageCSF.production in each time interval showed a clear tendency to circadian variation, with a minimum production 30% of maximum values (12 * 7 ml/h) -1800 h and a nightly peak production -0200 h of 42 t 2 ml/h. The total CSF production during the whole 24-h period, calculated as an averageof all measurements,was 650 ml for the whole group and 630 ml for repeatedmeasurementsin each time interval in one of the volunteers. flow; quantification FLUID (CSF) is produced by the choroid plexus in the four brain ventricles, flowing out into the cranial and spinal subarachnoid spaces from where it is absorbed into the blood stream. The main roles of the CSF are to protect the brain mechanically and to provide a suitable chemical environment for the central nervous system (6). In addition, the CSF provides a route of distribution for nutrients and hormones within the brain (25, 31). Although the cellular basis of CSF secretion is fairly well understood (32), little is known about the physiological regulation of total CSF production, especially in humans. Lack of noninvasive methodology has until now restricted investigations to experimental animals. By magnetic resonance imaging (MRI) it is possible to produce high-resolution images of the body, as well as measure fluid flow (3, 8, 24). Accordingly, several investigators have described a pulsatile flow of CSF through the cerebral aqueduct between the third and fourth ventricles (1, 2, 10, 16). Recently, a method for quantifying CSF production by calculating the net flow of CSF through the cerebral aqueduct using phase imaging has been described (26, 28). The linearity of the MR phase signal versus velocity relation in this type of experiment has been theoretically postulated (18, 23) and has been experimentally verified for a large range of flow velocities (11, 26). We have used this method to investigate the possibility of circadian variation in CSF production, considering that several hormones, such as vasopressin, melatonin and cortisol, which have been shown to influCEREBROSPI'NAL

R20

0363-6119/92

$2.00

Copyright

ence CSF production (7, 9, E), have circadian in blood and CSF (22, 29, 30).

rhythms

METHOD

A recently developed MRI method (26, 28) was used to calculate CSF velocities, flow, and production. This method utilizes the phase information obtained after subtracting two MR imageswith different phaseversusvelocity sensitivity. The pulse sequenceusedwas a gradient echo sequence(echo time 15 ms) with additional pulsed gradients for flow encoding in the slice selective direction. Ordinary (modulus) as well as velocity-sensitive (phase)imageswere collected. The sequence was electrocardiogram (ECG) triggered, and it was executed with two different encoding efficiencies running interleaved. The calibration constant between phase angle and velocity dependson the pulse sequencegradient schedule,and when a subtraction phasemappingtechnique is usedthe net calibration constant is obtained as the difference between the velocity sensitivities between the two basic sequences.Given a detailed knowledge of the magnetic field gradient timing table, the calibration constant can be theoretically determined from known expressions(23), but our experience is that an experimental determination will provide a better accuracy for the systemactually used(26, 28). The ten subtracted phaseimages showed the spatial velocity distribution of flowing matter through the imaging plane, as shown in Fig. 1. The temporal resolution between eachimageis l/10 of a heart cycle. For each discretepoint in time, the averageaqueductCSF velocity (mm/ s) was obtained in the following way. First, to calibrate the sequence,a rotating wheel phantom (20) was used.The wheel provides a range of velocities from -50 to +50 mm/s, and an axial subtracted phase image of the wheel obtained with the pulsesequenceof interest will automatically yield a phasesignal versusvelocity calibration curve. Second,to obtain the average CSF velocity in the aqueduct, a small region of interest was placed over the cerebral aqueduct. From the average phase value, a linear velocity wascalculated usingthe abovediscussed calibration constant. Finally, to obtain volume flow, the area of the aqueduct as obtained from a spin-echoimage with 256 phase encoding stepswas multiplied by the average aqueduct velocity. During every second heart beat, ten acquisitions equally spacedin the cardiac cycle were obtained. With 128 spatial phaseencoding steps,the total acquisition time was 128 x 2 x 2 x (60/HR) s, where HR is the heart rate. Assuming HR = 60 beats/min, the acquisition time will be ~8 min. With the above-describedmethod, six healthy volunteers (5 males,1 female, 25-32 yr) were examined three to five times within 24 h. The measurementswere spreadthroughout the 24-h interval as evenly as clinical laboratory circumstancespermitted. The method has been approved by the local Ethics Committee for Researchfor useon healthy volunteers.

0 1992 the American

Physiological

Society

Downloaded from www.physiology.org/journal/ajpregu by ${individualUser.givenNames} ${individualUser.surname} (134.225.001.226) on August 19, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

CIRCADIAN

VARIATION

IN

HIJMAN

CSF

Fig. 1. Cardiac-gated modulus (A) and phase (B) magnetic resonance imaging (MRI) of head, obtained projection, -600 ms after R peak. Cerebral aqueduct (arrows) is seen in both images. In phase image, to velocity and cerebrospinal fluid (CSF) velocity can be calculated from total average phase signal.

RESULTS

In Fig. 2 the calibration curve obtained with the rotating wheel is shown. It is seen that the calibration constant was 23 radians-m-’ .s, obtained with a linear regression coefficient of 0.999. Measurement of CSF production at different times during the day showed y-0.0229x+0.419 r-0.999,n-17

1.5 'si 5

1.0

B 5 2

03

!z 4: 2 0.

0.0

40

-la

-20 Vebcity

0

20

40

60

(mm/s)

Fig. 2. CSF calibration curve obtained with rotating wheel phantom showing phase angle in radians vs. velocity in mm/s. Calibration constant can be determined to 23.3 radians. m-’ . s, obtained with linear regression coefficient of 0.999.

R21

I’ROI)UCTION

in transversal signal is linear

markedly similar velocity curves throughout the cardiac cycle, while the curve offset varied (Fig. 3A ), resulting in pronounced variations in CSF production over a 24-h period (Fig. 3B). The velocity versus time curve was nearly sinusoidal with a period equal to the R-R interval in the ECG. The peak inflow was observed after 10% of the R-R interval, and the peak outflow was observed after 40-50% of the R-R interval. All six volunteers had a minimum production in the afternoon and a peak production after midnight. The average CSF production in each time interval showed a statistically significant variation with a peak production rate -0200 h estimated to be 42 f 2 ml/h (range 35-48 ml/h) (Fig. 4). A production minimum, 30% of maximum production, was seen around 1800 h and calculated to 12 rt 7 ml/h (Fig. 4). Covariance analysis of CSF production, measured as log (ml/h) and adjusted for the individual average CSF production during 24 h, measured as log (ml/24 h), showed that CSF production is not homogenous as a function of time of day (P = 0.009). Also one-way analysis of variance showed a statistically significant variation in CSF production over the 24-h period. To examine the intraindividual variation in each time interval, including the influence of possible methodological errors (26, 28), the CSF production of volunteer 1

Downloaded from www.physiology.org/journal/ajpregu by ${individualUser.givenNames} ${individualUser.surname} (134.225.001.226) on August 19, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

R22

CIRCADIAN

VARIATION

IN HUMAN

CSF PRODUCTION

period, calculated as an average of the production values in each ‘time interval, was 650 t 60 ml for the whole group (n = 6) and 630 ml in the second experiment on volunteer 1. DISCUSSION

40

60

60

100

Percent of cardiac cycle

20

24

Time of day

Fig. 3. Variation in velocity (mm/s) of CSF in cerebral aqueduct of volunteer I (A) determined by MR phase imaging at different times of day. Caudal flow is designated as positive, whereas cranial flow is designated as negative. Net flow (ml/h) at each measurement time (B), equivalent to CSF production volume, was calculated from velocity curves by time averaging of sampled velocity values, multiplied with aqueduct area.

t

100 i

4-

Timoofdy

Fig. 4. CSF production rate relative to each individual’s maximum production in 5 different time intervals during the day-night cycle (A

= 1200-1600, B = 1600-2000,c = 2000-2400,D= 2400-0400,E = 0600-1000 h), calculated from CSF flow velocities in cerebral aqueduct in 6 healthy volunteers aged 25-32 yr. Each bar represents average CSF production @maximum) in each time interval k SE. Number within parentheses in each bar represents number of measurements in each time interval (n).

was measured five times repeatedly in each time interval during a 24-h period. The variation in each time interval was relatively large (-30%), but the circadian variation in CSF production showed reasonable correlation with the average production rates for the whole group. The total CSF volume produced during the whole 24-h time

Because the CSF is produced in the brain ventricles and absorbed in the arachnoid granulations, there is a net outflow of CSF through the cerebral aqueduct. The net flow is overlayed by a pulsative flow as a result of the brain motion associated with the cardiac cycle (10). By measuring the average flow velocities in the cranial and caudal direction during the heart cycle it is thereby possible to calculate the net outflow of CSF through the cerebral aqueduct (26, 28). Regarding methodological errors, even though the calibration of the MR scanner in terms of velocity units as seen in Fig. 2 is associated with a very low uncertainty, distinct sources of error that influence relative production measurements (26,28) (i.e., measurements adjusted for individual averages), as well as absolute measurements, are miscalculation of time delay among succeeding frames, phase noise, heart rate fluctuation, and patient motion. However, a minor angulation of the imaging slice will not affect volume flow measurements due to the fact that a decrease in linear velocity that can be assumed from this phenomenon will be equaled out by an increase in aqueduct cross-sectional area. Regarding absolute measurements especially, the estimation of the aqueduct area may introduce a further source of error. A total error of -30% in absolute production measurements, added up from an assumption of errors in time delay miscalculation (Xi%), phase noise (lo%), and heart rate fluctuation (5%), has been assumed earlier (28), and gives an error limit that may explain single negative production values obtained in this study. Previous information on human CSF production rates is scarce, owing to the invasive nature of hitherto available techniques. Ventriculolumbar perfusions using radioactive tracers have demonstrated mean production rates averaging 21-22 ml/h for periods up to 6 h, which would equal 500 ml during 24 h (4, 21). Indirect measurements using the modified Masserman technique, based on the time for CSF pressure to return to normal after removal of a specified volume of CSF, have shown average CSF production values in the range 19-25 ml/h (14, 17). It should be noted that measuring net outflow of CSF in the cerebral aqueduct only registers CSF produced in the lateral and third ventricles. Because the fourth ventricle might contribute with up to 40% of the CSF production (5), the total volume of CSF production in our study could be as high as 1.0 l/24 h, a value twice that previously reported (4, 14, 17, 21). The reason for this discrepancy is unclear but may involve experimental errors influencing absolute flow and production measurements as described above. Furthermore, even though the time of day during which the perfusions were performed was not reported in these previous studies, assuming that it was during daytime, our results show a production similar to that previously reported (4, 14, 17, 21). Extrachoroidal sources of CSF production exist as

Downloaded from www.physiology.org/journal/ajpregu by ${individualUser.givenNames} ${individualUser.surname} (134.225.001.226) on August 19, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

CIRCADIAN

VARIATION

IN HUMAN

well (5). Because the route through which this extrachoroidal CSF contributes to CSF flow is not known (i.e., rostrally or caudally of the aqueduct), the influence on our measurements is difficult to determine. Possible circadian variation in CSF production has been speculated on (12) but has been difficult to investigate. Interestingly, support of the present findings may be found in continuous recordings of intracranial pressure in children with acute hydrocephalus that show daily cyclic variations with nighttime peaks, absent in children with chronic hydrocephalus (13). The mechanisms involved in the fluctuations in CSF production described here are presently unclear and could involve a number of substances shown to influence CSF hydrodynamics (7, 9,15). CSF production might also vary with other factors. It was recently reported that CSF production was 50% lower in elderly subjects compared with young (17). Measurements of CSF volume by MRI demonstrated an increase in CSF volume premenstrually in women (27), indicating that hormones might influence CSF production. The functional implications of circadian variations in CSF production remain to be investigated but could involve the role of the CSF as a pathway of metabolites from brain to blood or as carrier of chemical signals within the brain (19,25,31). The clinical implications of large fluctuations in CSF production may be important because of the frequent use of CSF samples to obtain an index of cerebral chemical activity. The importance of standardized times for CSF sampling should be apparent considering the possibility of variable dilution of substances released from the brain parenchyma into the CSF at different times of day. Another important aspect is the study of patients with abnormal CSF hydrodynamics, such as normal pressure hydrocephalus, brain atrophy with secondary dilatation of the ventricles, and benign intracranial hypertension. Possible circadian variation of CSF flow and production has to be accounted for before CSF flow measurements can be used in the clinical evaluation of these patients. This study was supported by the Swedish Medical Research Council Grants B90-39X-07307-04 and 14X-732, the Swedish Society for Medical Research, and the Danish Medical Research Council Grant 1296356. Address for reprint requests: C. Nilsson, Dept. of Medical Cell Research, Section of Neurobiology, University of Lund, Biskopsgatan 5, S-223 62 Lund, Sweden. Received 21 February 1991; accepted in final form 23 August 1991.

Formation and absorption of cerebrospinal fluid in man. Bruin 91: 707-720,1968. 5. Davson, H., Pathophysiology

K. Welch, and M. B. Segal. of the Cerebrospinol Fluid.

The Physiology

and

London: Churchill

&

Livingstone, 1987, p. 189-203. 6. Davson, H., Pathophysiology

K. Welch,

and

M.

of the Cerebrospinal

B. Segal. Fluid.

The

Physiology

and

London: Churchill

&

Livingstone, 1987, p. 375-445. 7. Decker,

Stimulatory effects of melatonin of choroid plexuses in golden hamsters.

J. F., and W. B. Quay.

on ependymal epithelium

J. Neural Transm. 55: 53-67, 1982. 8. Dijk, P. V. Direct cardiac NMR imaging of heart wall flow velocity. J. Comput. Assisted Tomogr. 8: 429-436,1984. 9. Faraci, F. M., W. G. Mayhan, and D. D. Heistad.

vasopressin on production vasopressin (&)-receptors.

of cerebrospinal

and blood

Effect of fluid: possible role of

Am. J. Physiol. 258 (Regulatory Integrative Comp. Physiol. 27): R94-R98, 1990. 10. Feinberg, D. A., and A. S. Mark. Human brain motion and

cerebrospinal fluid circulation imaging. Radiology 163: 793-799, 11. Firmin, nore.

D. N.,

G. L. Nayler,

The application

ment. Magn. Reson. 12. Flanagan, M. F.

demonstrated

with MR velocity

1987. P. J. Kilner,

and

D. B. Long-

of phase shifts in NMR for flow measure-

Med.

14: 230-241,199O.

Relationship between CSF and fluid dynamics in the neural canal. J. Manipulative Physiol. Ther. 11: 489-492,

1988. 13. Hayden,

P. W., D. B. Shurtleff, and E. L. Foltz. Ventricular fluid pressure recordings in hydrocephalic patients. Arch. Neurol.

23: 147-154, 1970. 14. Katzman, R., and

F. Hussey. A simple constant-infusion manometric test for measurement of CSF absorption. Neurology 20:

534-544,197o. 15. Lindvall-Axelsson,

M., P. Hedner, and C. Owman. Corticosteroid action on choroid plexus: reduction in Na’-K+-ATPase activity, choline transport capacity, and rate of CSF formation.

Exp. Brain Res. 77: 605-610, 1989. 16. Mascalchi, M. Cardiac-gated phase MR imaging of aqueductal CSF flow. J. Comput. Assisted Tomogr. 12: 923-926,1988. 17. May, C., J. A. Kaye, 3. R. Atack, M. B. Schapiro, R. P. Friedland, and S. I. Rapoport. Cerebrospinal fluid production is reduced in healthy aging. Neurology 40: 500-503, 1990. 18. Moran, P. R. A flow velocity zeugmatographic interlace for NMR imaging in humans. Magn. Reson. Imaging 1: 197-203,1982. 19. Nilsson, C., M. Lindvall-Axelsson, and C. Owman. Role of

the cerebrospinal fluid in volume transmission involving the choroid plexus. In: Volume Transmission in the Brain, edited by K. Fuxe and L. F. Agnati. New York: Raven, 1991, chapt. 24, p. 307315. 20. Nordell, B., F. Stahlberg, A. Ericsson, and C. Ranta. A rotating phantom for the study of flow effects in MR imaging. Magn. Reson. Imaging 6: 695-705,1988. R. C., E. S. Henderson, A. K. Ommaya, M. D. 21. Rubin, Walker, and D. P. Rall. The production of cerebrospinal fluid in man and its modification by acetazolamide. J. Neurosurg. 25: 430-436,1966. W. J., and S. M. Reppert. Neural regulation of the 22. Schwartz,

circadian vasopressin rhythm in cerebrospinal fluid: a preeminent role for the suprachiasmatic nuclei. J. Neurosci. 5: 2771-2778, 1985. 23. Singer,

J.

introduction

NMR diffusion and flow measurements and an to spin phase graphing. J. Phys. E Sci. Instrum. 11:

R.

281-291,1978. 24. Singer, J. R.,

REFERENCES 1. Bergstrand,

R23

CSF PRODUCTION

G., M. Bergstrom, B. Nordell, F. Stahlberg, A. Hemmingsson, G. Sperber, K.-A. Thuomas,

Ericsson, B. Jung. Cardiac gated MR imaging of cerebrospinal Comput. Assisted Tomogr. 9: 1003-1006,1985. 2. Bradley, W. G., K. E. Kortman, and B. Burgoyne.

A. and fluid. J.

and L. E. Crooks. Nuclear magnetic resonance blood flow measurements in the human brain. Science Wash. DC

221: 654-656,1983. R. Nucleoside 25. Spector,

and vitamin homeostasis in the mammalian central nervous system. Ann. NY Acad Sci. 481: 221-229,1986.

Flowing cerebrospinal fluid in normal and hydrocephalic states: appearance on MR images. Radiology 159: 611-616, 1986. 3. Bryant, D. J. Measurement of flow with NMR imaging using a gradient pulse and phase difference technique. J. Comput. Assisted

26. Stahlberg, Stubgaard, 0. Henriksen,

Tomogr. 4. Cutler,

27. Teasdale, Lawrence,

8: 588-593,1984. R. W. P., L. Page,

J. Galicih,

and

G. V. Watters.

F., J. Mogelvang, C. Thomsen, B. Nordell, M. A. Ericsson, G. Sperber, D. Greitz, H. Larsson, and B. Persson. A method for quantification of

flow velocities in blood and CSF using interleaved gradient-echo pulse sequences. Magn. Res. Imaging 7: 655-667, 1989. G. M., R. Grant, D. M. Hadlev,

B. Condon, and D. Wvper.

J. Pattersson,

Intracranial

A.

CSF

Downloaded from www.physiology.org/journal/ajpregu by ${individualUser.givenNames} ${individualUser.surname} (134.225.001.226) on August 19, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

R24

CIRCADIAN

VARIATION

IN HUMAN

volumes: natural variations and physiological changes measured by MRI. Acta Neuroschir. Suppl. 42: 230-235, 1988. 28. Thomsen, C., F. Stalberg, M. Stubgaard, B. Nordell, and the Scandinavian flow group. Fourier analysis of CSF flow velocities. A MRI study. Radiology 177: 659-665, 1990. 29. Vaughan, G. M. Human melatonin in physiologic and diseased states: neural control of the rhythm. J. Neural. Transm. Suppl. 21: 199-215,1986.

CSF PRODUCTION

30. Waldenlind, E., S. A. Gustafsson, K. Ekbom, and L. Wetterberg.. Circadian secretion of cortisol and melatonin in cluster headache during active cluster periods and remission. J. Neurol. Neurosurg. Psychiatry 50: 207-213,1987. 31. Wood, J. H. Neuroendocrinology of cerebrospinal fluid: peptides, steroids, and other hormones. Neurosurgery 11: 293-305, 1982. 32. Wright, E. M. Transport processes in the formation of cerebrospinal fluid. Rev. Physiol. Biochem. Pharmacol. 83: 3-34, 1978.

Downloaded from www.physiology.org/journal/ajpregu by ${individualUser.givenNames} ${individualUser.surname} (134.225.001.226) on August 19, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

Circadian variation in human cerebrospinal fluid production measured by magnetic resonance imaging.

Recent advances in magnetic resonance imaging have made it possible to visualize and quantify flow of cerebrospinal fluid (CSF) in the brain. The net ...
1MB Sizes 0 Downloads 0 Views