ARTICLE IN PRESS
Prostaglandins, Leukotrienes and Essential Fatty Acids 71 (2004) 13–18
In humans, the seasonal variation in poly-unsaturated fatty acids is
related to the seasonal variation in violent suicide and serotonergic
markers of violent suicide
S.R. De Vriesea, A.B. Christophea, M. Maesb,*
b
a
Department of Internal Medicine, Division of Nutrition, Ghent University Hospital, Belgium
Department of Psychiatry, Psychiatric Hospital Vijverdal, 1st Floor, Suite S108, P.O. Box 88, NL-6200 NB Maastricht, The Netherlands
Received 1 December 2003; accepted 3 December 2003
Abstract
Background: Depression is accompanied by a depletion of n-3 poly-unsaturated fatty acids (PUFAs). There is also a negative
correlation between suicide and fish-oil intake (rich in n-3 PUFAs) across different countries. Both depression and suicide show a
seasonal variation and are related to disorders in the serotonergic system.
Aims: The present study was carried out to determine if there is a seasonal variation in the PUFA fractions in serum
phospholipids and whether there are significant relationships between lowered n-3 PUFA status and the seasonal variation in the
number of suicide deaths and serotonergic markers of suicide.
Methods: We took monthly blood samples during 1 calendar year from 23 healthy volunteers and analyzed the PUFA
composition in serum phospholipids and related those data to the annual variation in the mean weekly number of suicides for
Belgium and the Bmax [3H]-paroxetine binding to platelets in the same 23 subjects.
Results: Significant annual rhythms were detected in the long-chain PUFAs only, i.e. arachidonic acid (C20: 4n-6; AA),
eicosapentaenoic acid (C20: 5n-3; EPA), and docosahexaenoic acid (C22: 6n-3; DHA). There was a significant correlation between
the changes over the last 2 weeks in AA and EPA and the mean weekly number of violent, but not nonviolent, suicide deaths in
Belgium. There was a significant correlation between the PUFAs, AA and DHA, and the Bmax [3H]-paroxetine binding to platelets.
Conclusions: Our results show that there is a true seasonality in long-chain PUFAs, such as AA, EPA and DHA. The results
suggest that the seasonality in PUFAs may be related to the incidence of violent suicide and the expression of the serotonin
transporter complex.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: n-3 PUFAs; Seasons; EPA; AA; EPA; Suicide; Serotonin
1. Introduction
There are two families of essential poly-unsaturated
fatty acids (PUFA), i.e. the n-6 and n-3 families. These
fatty acids are essential because they are required for
optimal functioning of the organism but cannot be
synthesized de novo by humans. Linoleic acid (C18: 2n6, LA) and 8-linolenic acid (C18: 3n-3, 8-LNA) are the
parent EFAs. They are desaturated and elongated to
form long-chain PUFAs. For example, arachidonic acid
*Corresponding author. Tel.: +31-43-387-5444; fax: +31-43-3877443.
E-mail address: [email protected] (M. Maes).
URL: http://www.ediver.be.
0952-3278/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.plefa.2003.12.002
(C20: 4n-6; AA), eicosapentaenoic acid (C20: 5n-3;
EPA), and docosahexaenoic acid (C22: 6n-3; DHA) are
important structural components of cell membranes
[1,2] and neural tissues [3,4].
Major depression is associated with lowered n-3
PUFA levels [5–9]. Thus, patients with major depression
have lowered EPA and DHA levels in serum phospholipids and cholesteryl esters or in their red blood cell
membranes [7–9]. The severity of depression is
negatively correlated with the ratio of EPA to AA
in serum phospholipids and red blood cell membranes
[10]. There is a significant negative correlation
between the incidence of suicide deaths across different
countries and the intake in fish oil, which is rich in EPA
and DHA [5].
ARTICLE IN PRESS
14
S.R. De Vriese et al. / Prostaglandins, Leukotrienes and Essential Fatty Acids 71 (2004) 13–18
There is an annual variation in the severity of
depression in depressed patients [11] and in the number
of suicide deaths [12]. We found a seasonal variation in
violent suicide only, but not in nonviolent suicide, with
more deaths in spring and summer than in winter and
fall [12]. There is also a seasonality in total serum
cholesterol which is highly significantly synchronized
with the annual rhythms in violent suicide deaths [13].
However, no data are available on the seasonal
variation in the PUFAs and its relation to the annual
variation in suicide.
The neurotransmitter serotonin (5-HT) plays a role in
the pathophysiology of major depression and suicide.
There is now also evidence that the seasonal variation in
serotonergic variables, such as the 5-HT transporter (5HTT) complex, as measured by the Bmax [3H]-paroxetine
binding to platelets is related to the seasonal variation in
violent suicide [13]. The PUFA status in the neuronal
membrane determines in part the expression of serotonergic receptors, such as the 5-HTT, and that of 5-HT
release and reuptake, tryptophan hydroxylase activity
(the rate limiting enzyme in 5-HT synthesis) and the
brain concentrations of 5-HT and 5-hydroxyindolacetic
acid, the major metabolite of 5-HT [14]. No data are
available, however, on the relationships between the
PUFA status and the expression of serotonergic markers
of depression and suicide, e.g. the Bmax [3H]-paroxetine
binding to platelets.
This study was conducted in order to examine
whether there is (a) a seasonal variation in the PUFA
fractions in phospholipids in the serum of healthy
volunteers; and whether (b) this seasonal variation is
related to that in the Bmax [3H]-paroxetine binding to
platelets and the suicide rate in Belgium.
2. Subjects and methods
2.1. Study population
Twenty-three healthy Caucasian volunteers (12 men
and 11 women, mean age 38.5 years, range 23–69 years)
were selected to participate in this study. Inclusion and
exclusion criteria for subjects are described somewhere
else [14]. The geographical coordinates for this study are
51.2EN and 4.5EE around the city of Antwerp,
Belgium. The subjects gave oral informed consent to
participate in the study in accordance with the ethical
standards of the Ethical Committee of the University of
Antwerp. The study period extended from December 11,
1991 until December 25, 1992. Seasons were defined by
their respective solstices and equinoxes, i.e. winter:
December 21–March 20; spring: March 21–June 20;
summer: June 21–September 20; and fall: September
21–December 20.
2.2. Methods
Blood collections were performed under standardized
conditions to minimize sources of preanalytical variation [14,15]. Blood samples were taken after an overnight fast at 8:00 am (30 min). Each subject had 12
consecutive monthly blood samples carried out by the
same investigator. Blood samples in men and postmenopausal women were evenly spaced at monthly
intervals. Blood samples in premenopausal females were
always carried out 5–10 days after the first day of the
menstrual cycle. Serum was stored in plastic tubes under
nitrogen at 80EC until thawed for fatty acid analysis.
All serum samples from one subject are analyzed
simultaneously using the same batch of solvents and
the same capillary GC column [15]. Lipids were
extracted from 1 ml serum according to a modified
Folch extraction with methanol:chloroform (1:2) [16].
The lipids were separated by thin layer chromatography
on rhodamine-impregnated silica gel plates using
petroleum ether (bp 60–80EC; Merck Belgolab, Overijse, Belgium)/acetone 85:15 as mobile phase [17]. The
phospholipid fraction of serum lipids was scraped off
and the fatty acids converted into methyl esters by
transesterification with 2 ml of a mixture of methanol:benzene:HCl (aqueous, 12 N) (80:20:5) [18]. After cooling and adding 2 ml of water, fatty acid methyl esters
were extracted with petroleum ether (bp 40–60EC),
evaporated to dryness under a nitrogen flow at a
temperature not exceeding 40EC, and analyzed by
temperature programmed capillary gas chromatography
(Varian Model 3500, Walnut Creek, CA, USA) on a
25 m 250 Fm (L ID) 0.2 Fm df Silar 10C column
[18]. The injection and detection temperatures were set
at 285EC. The starting temperature of the column was
150EC, which was increased to 240EC after 3 min at a
rate of 2EC/min. The carrier gas was nitrogen with a
flow of 25 cm/s. Peak identification was performed by
spiking with authentic standards (Sigma-Aldrich, Bornem, Belgium). Peak integration and calculation of the
percent composition was performed electronically with a
Varian Model 4290 integrator. The coefficient of
variation of intra-assay samples of the entire method
of fatty acid analysis for peaks bigger than 1 wt% is less
than 5% and for peaks smaller than 1 wt% is less than
10%. The results are expressed as weight percent of total
fatty acids.
Reports of all deaths by suicide in Belgium from
January 1, 1979 to December 31, 1987, were obtained
from the National Institute of Statistics in Brussels. In
Belgium, physicians make reports of all unnatural
deaths and complete a standardized questionnaire that
is sent to the National Institute of Statistics, where it is
included in the official death statistics. We made a
distinction between violent and nonviolent suicides [12].
We computed the weekly number of suicides, violent
ARTICLE IN PRESS
S.R. De Vriese et al. / Prostaglandins, Leukotrienes and Essential Fatty Acids 71 (2004) 13–18
and nonviolent suicides for each study year from
January 1 through December 31. The mean weekly
number of deaths per week was entered in subsequent
statistical analysis. The assay of the Bmax [3H]-paroxetine binding to platelets (expressed as fmoles/mg
protein) has been described in detail in our previous
publications [15].
2.3. Statistical analysis
A comprehensive review of the statistical procedures
used to measure seasonality can be found in previous
papers [14,19]. Seasonal variation has been ascertained
by means of analysis of variance (ANOVA; followed by
Fisher’s least significance difference to test multiple post
hoc differences between the four seasons), cosinor
analysis, and least-squares cosine spectral analysis
[12,19]. Spectral analysis identifies all significant
rhythms by means of a significant signal-to-noise ratio
in the spectral peaks. F-statistics are generated as
indicators of this signal-to-noise ratio and are listed in
a periodogram or F-spectrum. Spectral analyses were
performed on the pooled time series of the 23 healthy
subjects after normalization of the fatty acid data and
the Bmax [3H]-paroxetine binding relative to the yearly
mean of the monthly measurements in each of the
subjects. This normalization eliminates the interindividual variability in the data. Relationships between the
PUFAs and suicide and the Bmax [3H]-paroxetine
binding to platelets were examined by regression
analyses, which were pooled over the time series of the
23 healthy volunteers (to eliminate interindividual
variability) and corrected for auto-correlation
3. Results
3.1. Seasonal variation in PUFAs
Table 1 shows the mean PUFA values in the seasons.
AA was significantly lower in winter than in the other
seasons. EPA was significantly lower in winter and
15
spring than in summer. DHA was significantly lower in
winter than in the other seasons. Total n-3 was
significantly lower in winter than in autumn and
summer. There were no significant seasonal differences
in any of the other PUFA fractions, including C18: 2n-6
and C18: 3n-3.
Fig. 1 shows the results of spectral analyses performed
on the most important PUFA fractions. Significant
periodicities in the F-spectra were found for AA, EPA,
DHA and total n-3, but none of the other PUFAs (data
not shown). The F spectra showed that only the annual
rhythms and no other periodicities were significant.
Table 2 shows the results of cosinor analyses performed
on the above variables which were significant in the
spectral analyses. The acrophase of the annual rhythm
in the four variables occurred somewhere between
August and September.
3.2. Relationships between PUFAs and suicide deaths and
3
the Bmax [ H]-paroxetine binding
In a previous study [12], we have established the
chronograms of the weekly number of deaths due to
violent and nonviolent suicide for Belgium for the
period 1979–1987. In the present study, we examined the
relationships between the PUFAs, including AA, EPA,
DPAn-6 and DHA, and the occurrence of violent or
nonviolent suicide. Simple regression analyses showed
no significant correlations either between any of the
PUFA parameters and nonviolent suicide or between
any of the above PUFAs, including EPA (r ¼ 0:15;
P ¼ 0:3) and DHA (r ¼ 0:21; P ¼ 0:1), and the occurrence of violent suicide. However, the delta AA (delta
AA=AA, 2 weeks earlier minus the actual AA)
(r ¼ 0:63; P ¼ 0:00001), delta EPA (r ¼ 0:64;
P ¼ 0:00001) and delta DHA (r ¼ 0:58; P ¼ 0:00005)
were significantly and negatively correlated to violent,
but not nonviolent, suicide rate. Multiple regression
analysis showed that delta EPA was the single best
variable correlating to violent suicide. Fig. 2 shows the
correlation between violent suicide rate in Belgium and
the delta EPA values. Negative correlations were found
Table 1
Seasonal differences in serum phospholipid PUFA fractions in 23 healthy volunteers who had monthly blood samplings during 1 calendar year
C18:2n-6
C18:3n-3
C20:4n-6
C20:5n-3
C22:5n-6
C22:6n-3
total n-6
total n-3
(LA)
(LNA)
(AA)
(EPA)
(DPAn-6)
(DHA)
Fall
Winter
Spring
Summer
F
df
P
17.77 (3.95)
0.68 (0.69)
5.05 (3.79)
0.46 (0.42)
0.52 (0.67)
1.78 (1.33)
25.5 (7.5)
3.75 (1.84)
18.11 (4.15)
0.64 (0.51)
4.55 (3.15)
0.39 (0.33)
0.57 (0.77)
1.52 (0.96)
25.6 (6.9)
3.30 (1.29)
18.58 (4.27)
0.71 (0.55)
4.81 (3.45)
0.44 (0.43)
0.51 (0.64)
1.81 (1.29)
26.1 (6.7)
3.65 (1.64)
17.56 (4.09)
0.73 (0.61)
4.81 (3.74)
0.55 (0.62)
0.54 (0.64)
1.87 (1.37)
25.2 (7.5)
3.81 (1.71)
1.8
0.5
4.9
3.1
0.6
8.1
0.6
3.3
3/171
3/171
3/171
3/171
3/171
3/169
3/171
3/171
0.2
0.7
0.003
0.02
0.6
0.0001
0.6
0.02
All results are expressed as mean (8SD). All results of ANOVAs with seasons and subjects as factors.
ARTICLE IN PRESS
S.R. De Vriese et al. / Prostaglandins, Leukotrienes and Essential Fatty Acids 71 (2004) 13–18
16
Fig. 1. F spectrum obtained by least-squares cosine spectral analysis performed on the fatty acid data. Shown are the F values obtained at 100
different periods over a period of 1 year. The spectral analyses were performed on the pooled time series of the 23 healthy subjects after normalization
of the fatty acid data relative to the yearly mean of the monthly measurements. AA, EPA, DHA and total n-3 PUFAs, but not the other fractions,
showed significant periodicities. The peak significances of those periodicities were detected at 365 days (annual variations).
Table 2
Results of cosinor analyses with the PUFA fractions as dependent
variables and an annual rhythm as explanatory variable
Dependent
variables
F
df
P
Amplitude Acrophase
(wt%)
(deg)
20: 4n-6 (AA)
20: 5n-3 (EPA)
22: 6n-3 (DHA)
Sum n-3
4.9
4.9
10.8
7
2/250
2/248
2/258
2/260
0.008
0.008
0.0001
0.001
0.21
0.099
0.18
0.32
269
261
245
257
The cosinor analyses are performed on the pooled time series of the
PUFA data which were normalized relative to their yearly means.
between violent suicide rate and the AA and EPA, but
not other PUFA values, established some weeks earlier,
e.g. a time lag of 8 weeks showed for AA a correlation
coefficient of r ¼ 0:41 (P ¼ 0:003) and for EPA r ¼
0:45 (P ¼ 0:001).
Previously, we determined the seasonal variation in
the Bmax [3H]-paroxetine binding to platelets in normal
volunteers [15]. In the present study, we examined the
relationship between the PUFA fractions and the Bmax
[3H]-paroxetine binding to platelets. We found significant negative correlations between the Bmax [3H]paroxetine binding to platelets and AA (r ¼ 0:23;
P ¼ 0:0008) and DHA (r ¼ 0:16; P ¼ 0:01), but not
DPAn-6 (r ¼ 0:06; NS) or EPA (r ¼ 0:09; NS). There
was also a significant negative correlation between Bmax
[3H]-paroxetine binding to platelets and total n-3
(r ¼ 0:20; P ¼ 0:003), but not total n-6. Multiple
regression analysis showed that 12.5% of the variance
in the Bmax [3H]-paroxetine binding to platelets could be
explained by AA (F ¼ 13:9; P ¼ 0:0005; negatively
loaded), the AA levels 1 week earlier (F ¼ 5:8; P ¼
0:02; positively loaded) and the DHA values 2 weeks
earlier (F ¼ 6:0; P ¼ 0:01; negatively loaded). This
suggests that the expression of the 5-HTT is related to
decreases in AA over the last week and lowered DHA
levels 2 weeks earlier.
4. Discussion
The first major finding of this present study is that
AA, EPA, DHA and total n-3 fractions, show—in a
normal population—a statistically significant annual
rhythm with an acrophase around August–September
and lows in winter. It is interesting to note that the
parent PUFAs, C20: 3n-6 and DPAn-6 showed no
significant seasonal variation. This could suggest that
there is a seasonal variation in the delta-5-desaturase.
The origin of the above seasonal rhythms has remained
elusive, but the seasonal variation in many human
physiological functions is related to genetically determined processes (‘‘endogenous’’ rhythms) which may be
adjusted in time (or ‘‘entrained’’ or ‘‘synchronized’’) by
cycles in light–dark span or ambient temperature.
The second major finding of this study is the
significant correlation between the seasonality in PUFAs, such as EPA and DHA, and the occurrence of
violent suicide. The lower the PUFA levels some weeks
earlier and the higher the increase over the last couple of
weeks, the higher the number of suicide deaths. Since
decreased EPA and DHA may be related to the
pathophysiology of suicide and depression (see Introduction), the above findings may suggest that the
seasonal variation in EPA or DHA could in part
explain the seasonality in violent suicide occurrence.
ARTICLE IN PRESS
S.R. De Vriese et al. / Prostaglandins, Leukotrienes and Essential Fatty Acids 71 (2004) 13–18
17
Fig. 2. Correlation between mean weekly number of violent suicides in Belgium and the delta EPA, i.e. EPA 2 weeks earlier minus the actual EPA
(r ¼ 0:64; P ¼ 0:00001).
Another major finding of this study is the significant
negative correlation between PUFAs, such as AA and
DHA, and the Bmax [3H]-paroxetine binding to platelets.
[3H]-paroxetine is a selective and potent inhibitor of the
5-HT reuptake in serotonergic neurons [20,21]. [3H]Paroxetine is very useful as ligand to examine the 5-HTT
complex [20,21]. There is a seasonal variation in the
Bmax [3H]-paroxetine binding to platelets [15] and in
various other serotonergic variables, such as plasma
tryptophan and cerebrospinal fluid 5-hydroxy-indoleacetic acid [14,22–24]. As described in the Introduction,
the PUFA levels in the brain modulate the metabolism
of the neurotransmitter 5-HT, which is known to be
related to suicide [25]. We have discussed elsewhere that
decreased n-3 levels could be related to the serotonergic
etiology of suicide and depression. The seasonal
variation in violent suicide rate is synchronized to that
in serotonergic variables, including the Bmax [3H]paroxetine binding to platelets [15]. Therefore, we
hypothesize that the annual variation in violent suicide
rate is related to the yearly rhythms in the 5-HTT
complex, which, in turn, are related to those in some
long-chain PUFAs and that these biological variables
may determine a changing susceptibility to commit
violent suicide.
References
[1] J. B!ezard, J.P. Blond, A. Bernard, P. Clouet, The metabolism and
availability of essential fatty acids in animal and human tissues,
Reprod. Nutr. Dev. 34 (1994) 539–568.
[2] S.M. Innis, Essential fatty acids in growth and development,
Prog. Lipid Res. 30 (1991) 39–103.
[3] M.A. Crawford, W. Doyle, P. Drury, A. Lennon, K. Costeloe,
M. Leighfield, N-6 and n-3 fatty acids during early human
development, J. Intern. Med. Suppl. 225 (1989) 159–169.
[4] M.A. Crawford, The role of essential fatty acids in neural
development: implications for perinatal nutrition, Am. J. Clin.
Nutr. 57 (1993) 703S–709S.
[5] J.R. Hibbeln, Fish consumption and major depression, Lancet
351 (1998) 1213.
[6] M. Maes, R.S. Smith, Fatty acids, cytokines, and major
depression, Biol. Psychiatry 43 (1998) 313–314.
[7] M. Peet, B. Murphy, J. Shay, D. Horrobin, Depletion of omega-3
fatty acid levels in red blood cell membranes of depressive
patients, Biol. Psychiatry 43 (1998) 315–319.
[8] M. Maes, R. Smith, A.B. Christophe, P. Cosyns, R. Desnyder,
H.Y. Meltzer, Fatty acid composition in major depression:
decreased omega 3 fractions in cholesteryl esters and increased
c20:4 omega 6/c20:5 omega 3 ratio in cholesteryl esters and
phospholipids, J. Affect. Disord. 38 (1996) 35–46.
[9] M. Maes, A.B. Christophe, J. Delanghe, C. Altamura, H. Neels,
H.Y. Meltzer, Lowered omega3 polyunsaturated fatty acids in
serum phospholipids and cholesteryl esters of depressed patients,
Psychiatr. Res. 85 (1999) 275–291.
[10] P.B. Adams, S. Lawson, A. Sanigorski, A.J. Sinclair, Arachidonic
acid to eicosapentaenoic acid ratio in blood correlates positively
with clinical symptoms of depression, Lipids 31 (suppl) (1996)
S157–S161.
[11] M. Maes, H.Y. Meltzer, E. Suy, F. De-Meyer, Seasonality in
severity of depression: relationships to suicide and homicide
occurrence, Acta Psychiatr. Scand. 88 (1993) 156–161.
[12] M. Maes, P. Cosyns, H.Y. Meltzer, F. De-Meyer, D. Peeters,
Seasonality in violent suicide but not in nonviolent suicide or
homicide, Am. J. Psychiatry 150 (1993) 1380–1385.
[13] M. Maes, S. Scharpe, P. Cosyns, H. Neels, W. Stevens, C. Bridts,
et al., Biochemical, immune and metabolic correlates of seasonal
variation in violent suicide in Belgium, Eur. Psychiatry 43 (1996)
21–33.
[14] M. Maes, S. Scharpe, R. Verkerk, P. D’Hondt, D. Peeters, P.
Cosyns, P. Thompson, F. De Meyer, A. Wauters, H. Neels,
Seasonal variation in plasma l-tryptophan availability in healthy
volunteers. Relationships to violent suicide occurrence, Arch.
Gen. Psychiatry 52 (1995) 937–946.
ARTICLE IN PRESS
18
S.R. De Vriese et al. / Prostaglandins, Leukotrienes and Essential Fatty Acids 71 (2004) 13–18
[15] P. D’Hondt, M. Maes, J.E. Leysen, W. Gommeren, L. Heylen, F.
DeMeyer, et al., Seasonal variation in platelet [3H]-paroxetine
binding in healthy volunteers. Relationship to climatic variables,
Neuropsychopharmacology 15 (1996) 187–198.
[16] J. Folch, M. Lees, G.H.S. Stanley, Asimple method for the
isolation and purification of total lipids from animal tissues, J.
Biol. Chem. 226 (1957) 497–509.
[17] A.B. Christophe, F. Matthijs, New method for the determination
of the fatty acid pattern of serum lipid classes, Clin. Chim. Acta
16 (1967) 39–43.
[18] S.R. De Vriese, M. Dhont, A.B. Christophe, Oxidative stability of
low density lipoproteins and vitamin E levels increase in maternal
blood during normal pregnancy, Lipids 36 (2001) 361–366.
[19] F. De Meyer, P. Vogelaere, Spectral resolution of cardiocirculatory variations in men measured by autorhythmometry
over 2 years, Int. J. Biometeorol. 34 (1990) 105–121.
[20] E.T. Mellerup, P. Plenge, High affinity binding of 3H-paroxetine
and 3H-imipramine to rat neuronal membranes, Psychopharmacology (Berl.) 89 (1986) 436–439.
[21] S.C. Cheetham, J.A. Viggers, N.A. Slater, D.J. Heal, W.R.
Buckett, [3H]paroxetine binding in rat frontal cortex strongly
correlates with [3H]5-HT uptake: effect of administration of
various antidepressant treatments, Neuropharmacology 32 (1993)
737–743.
[22] C. Swade, A. Coppen, Seasonal variations in biochemical
factors related to depressive illness, J. Affect. Disord. 2 (1980)
249–255.
[23] T.D. Brewerton, W.H. Berrettini, J.-I.J. Nurnberger, M. Linnoila,
Analysis of seasonal fluctuations of CSF monoamine metabolites
and neuropeptides in normal controls: findings with 5HIAA and
HVA, Psychiatr. Res. 23 (1988) 257–265.
[24] A. Carlsson, L. Svennerholm, B. Winblad, Seasonal and circadian
monoamine variations in human brains examined post mortem,
Acta Psychiatr. Scand. Suppl. 280 (1980) 75–85.
[25] A. Roy, M. Linnoila, Suicidal behavior, impulsiveness and
serotonin, Acta Psychiatr. Scand. 78 (1988) 529–535.