Hyperprolactinemia in antipsychotic-naive patients
with first-episode psychosis
A. Riecher-Rössler
1
*, J. K. Rybakowski
2
,M.O.Pflueger
1
, R. Beyrau
3
, R. S. Kahn
4
, P. Malik
5
,
W. W. Fleischhacker
5
; and the EUFEST Study Group†
1
University of Basel Psychiatric Clinics, Basel, Switzerland
2
Department of Adult Psychiatry, Poznan University of Medical Sciences, Poznan, Poland
3
Department of Clinical Chemistry, University Hospital Basel, Switzerland
4
Department of Psychiatry, Rudolf Magnus Institute of Neuroscience, University Medical Centre Utrecht, The Netherlands
5
Department of Biological Psychiatry, Medical University Innsbruck, Austria
Background. Hyperprolactinemia is frequent in patients with schizophrenic psychoses. It is usually regarded as
an adverse effect of antipsychotics but has recently also been shown in patients without antipsychotic medication.
Our objective was to test whether hyperprolactinemia occurs in antipsychotic-naive first-episode patients (FEPs).
Method. In the framework of the European First Episode Schizophrenia Trial (EUFEST), 249 out of 498 FEPs were
eligible for this study, of whom 74 were antipsychotic naive. All patients were investigated regarding their serum
prolactin levels with immunoassays standardized against the 3rd International Reference Standard 84/500.
Results. Twenty-nine (39%) of the 74 antipsychotic-naive patients showed hyperprolactinemia not explained by any
other reason, 11 (50%) of 22 women and 18 (35%) of 52 men.
Conclusions. Hyperprolactinemia may be present in patients with schizophrenic psychoses independent of antipsychotic
medication. It might be stress induced. As enhanced prolactin can increase dopamine release through a feedback
mechanism, this could contribute to explaining how stress can trigger the outbreak of psychosis.
Received 9 August 2012; Revised 17 January 2013; Accepted 24 January 2013; First published online 18 April 2013
Key words: Antipsychotic, dopamine, first-episode psychosis, prolactin, schizophrenia, stress.
Introduction
Hyperprolactinemia is frequently found in patients
with schizophrenic psychoses and is usually considered
to be an adverse effect of antipsychotic medication.
It occurs with most conventional antipsychotics
but also during treatment with some new-generation
antipsychotics and is mainly associated with D
2
re-
ceptor affinity (Bushe et al. 2008; Fitzgerald & Dinan,
2008; Riecher-Rössler et al. 2009). However, there
have also been recent reports on hyperprolactinemia
in antipsychotic-free patients with first-episode psy-
chosis (Kahn et al. 2008; Aston et al. 2010; Garcia-Rizo
et al. 2012) and in patients with an at-risk mental state
(ARMS) for psychosis (Aston et al. 2010).
Prolactin is a peptide hormone secreted mainly
by prolactin-producing cells of the anterior pituitary.
Its main function is to induce lactation. The control
of prolactin production and secretion is complex
(Fig. 1). The main regulatory mechanism is inhibition
by dopamine, which is synthesized in neurons of
the hypothalamus and then secreted through portal
blood into the anterior pituitary, where it exerts
its actions on prolactin-producing cells through D
2
receptors. Dopamine is thus the main prolactin-
inhibiting factor (PIF) (Fitzgerald & Dinan, 2008;
Low, 2008; Prabhakar & Davis, 2008). Dopamine toni-
cally inhibits prolactin release. A negative feedback
control of prolactin secretion is mediated by a short
loop mechanism to the hypothalamus. Prolactin acti-
vates prolactin receptors leading to increased dopa-
mine synthesis and release (Fitzgerald & Dinan, 2008;
Low, 2008).
Prolactin secretion is stimulated by sucking but
also by stress (Low, 2008; Prabhakar & Davis, 2008).
Prolactin-releasing factors (PRFs) probably include
thyrotropin-releasing hormone (TRH), vasoactive
intestinal polypeptide (VIP) and oxytocin. Estrogens
also stimulate prolactin synthesis through a positive
feedback loop, and indirectly through their anti-
dopaminergic properties (Low, 2008).
* Address for correspondence: Prof. Dr med. A. Riecher-Rössler,
Center for Gender Research and Early Detection, University of Basel
Psychiatric Clinics, c/o University Hospital Basel, Petersgraben 4,
CH-4031 Basel, Switzerland.
(Email: anita.riecher@upkbs.ch)
† Members of the EUFEST Study Group are listed in the Appendix.
Psychological Medicine (2013), 43, 2571–2582. © Cambridge University Press 2013
doi:10.1017/S0033291713000226
ORIGINAL ARTICLE
https://doi.org/10.1017/S0033291713000226
Downloaded from https://www.cambridge.org/core. WWZ Bibliothek, on 12 Feb 2019 at 16:14:55, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Hyperprolactinemia (i.e. prolactin serum values
above the normal range) can be caused by many
physiological processes, by psychological distress ( for
review see Low, 2008; Prabhakar & Davis, 2008), and
also by pathological conditions such as prolactin-
secreting tumors. Thus, hyperprolactinemia in patients
with schizophrenic psychoses may not only be treat-
ment related but also possibly disease related.
Hyperprolactinemia can suppress gonadal func-
tion with a reduced physiological production of
estrogens and testosterone (Dickson et al. 2000;
Prabhakar & Davis, 2008). This can have serious clinical
consequences (e.g. sexual dysfunction, infertility or
osteopenia) in both men and women (for review see
(Haddad & Wieck, 2004; Miller, 2004; Prabhakar &
Davis, 2008; Riecher-Rössler et al. 2009).
We therefore aimed to determine, in the framework
of the European First Episode Schizophrenia Trial
(EUFEST; Fleischhacker et al. 2005; Kahn et al. 2008),
whether antipsychotic-naive first-episode patients
(FEPs) show hyperprolactinemia and, if so, whether
it is associated with factors such as pretreatment, age,
gender or severity of illness. To our knowledge this
is one of the largest samples of antipsychotic-naive
first-episode psychosis patients studied regarding
these issues.
Paraventricular
nucleus
Arcuate
nucleus
Dopamine
neuron
PRF
neurons
Hypothalamus
TRH
Oxytocin
VIP
Dopamine
Estrogen
to CNS
Pituitary
PRL
Breast
Multiple
target organs
PRL receptors
Estrogen receptors
Spinal
afferent
Suckling
stimulus
GABA
+
+
+
+
+
+
+
–
–
–
–
Fig. 1. Control of prolactin (PRL) production and secretion. PRF, prolactin-releasing factor; TRH, thyrotropin-releasing
hormone; VIP, vasoactive intestinal polypeptide; GABA, γ-aminobutyric acid; CNS, central nervous system. Reproduced,
with permission from Elsevier, from W. J. Low (2008). Neuroendocrinology. In Williams Textbook of Endocrinology
(ed. H. M. Kronenberg, S. Melmed, K. S. Polonsky and P. R. Larsen), 9th edn, pp. 85–295. W.B. Saunders Co.:
Philadelphia, PA.
2572 A. Riecher-Rössler et al.
https://doi.org/10.1017/S0033291713000226
Downloaded from https://www.cambridge.org/core. WWZ Bibliothek, on 12 Feb 2019 at 16:14:55, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Method
Data were obtained as part of the EUFEST, the meth-
odology of which has been described in more detail
in previous publications (Fleischhacker et al. 2005;
Kahn et al. 2008). A total of 50 centers participated.
Eligible patients were aged 18–40 years and met
DSM-IV criteria for schizophrenia, schizophreniform
disorder or schizo-affective disorder. Onset of positive
symptoms had to have occurred no more than 2 years
before study entry, and no more than 6 weeks of
lifetime exposure to antipsychotics was allowed.
After complete description of the study to the subjects,
written informed consent was obtained. The trial
complied with the Declaration of Helsinki, and was
approved by the ethics committees of the participating
centers. The Julius Center for Health Sciences and
Primary Care monitored the trial according to Good
Clinical Practice and International Conference on
Harmonisation guidelines.
We included 249 patients from the EUFEST sample
(total 498 patients) in the current study. The remaining
249 patients were excluded for the following reasons:
in 222 it could not be safely confirmed that they had
had their blood samples for prolactin drawn before
having received their first antipsychotic medication,
21 had unknown previous medications apart from
antipsychotics, and six were non-Caucasian. The latter
were excluded as there is some evidence that there
might be ethnic differences regarding the hormonal
response to psychological stressors (Chong et al. 2008).
Baseline data included demographics, diagnosis,
treatment setting, psychopathology [Positive and
Negative Syndrome Scale (PANSS); Kay et al. 1987],
severity of illness [Clinical Global Impressions (CGI)
scale; Guy, 1976], and other parameters described
in the original papers. Pretreatment history was ascer-
tained by questioning patients and significant others,
in addition to reviewing available charts.
Prolactin measurements
In each center, immunoassays were used for prolactin
measurements, which were standardized against
the 3rd International Reference Standard 84/500. The
lower and upper reference limits were calculated as
the 5th and 95th percentiles. Hyperprolactinemia
in this reference is defined as a value above the
95th percentile, that is > 0.38 U/l in men and > 0.53 U/l
in women. To convert values from ng/ml to U/l we
used a conversion factor of 0.0212 (Kahn et al. 2008).
Statistical methods
The R environment for statistical computation was
used to analyze the EUFEST dataset (R Development
Core Team, 2011). Simple parametric and non-
parametric statistical tests, such as the Mann–Whitney
U test and the Student t test, were performed to analyze
group differences. χ
2
tests and Fisher’s exact tests were
applied wherever count data were analyzed.
Log-transformed prolactin values were modeled
by age, sex and PANSS positive/negative score to
account for the influence of relevant variables. A gen-
eralized linear model (GLM) approach was adopted,
specified by a log link function and gamma family dis-
tribution. As the mean function indicates a squared
relationship between mean log prolactin and its stan-
dard deviation, the assumption of an exponential
residual distribution seems to be justified and hence
the choice of a GLM belonging to the gamma family.
An analysis of deviance was performed to determine
a particular factor’s contribution in explaining pro-
lactin deviance [a concept functional equivalent to
variance in ordinary least squares analysis but specific
to GLMs and other modern approaches in linear
modeling].
Results
Of the 249 FEP patients with reliable records on medi-
cation before prolactin measurements, 74 (52 men
and 22 women) were naive for the intake of any anti-
psychotic medication. Clinical and demographic char-
acteristics of the 74 antipsychotic-naive and the 175
not-naive patients are summarized in Table 1.
Hyperprolactinemia in antipsychotic-naive patients
As expected, a high proportion of the patients
with prior or current intake of antipsychotics showed
hyperprolactinemia (i.e. levels above the upper limit
of the normal range). However, 30 (40.5%) of the 74
antipsychotic-naive patients also showed hyperprolac-
tinemia: 11 (50%) of the 22 women and 19 (36.5%) of
the 52 men.
The mean prolactin values were also strongly
elevated not only in the non-naive but also in the
antipsychotic-naive patients, and the elevation was
especially high in women. Table 2 shows the pro-
portion of patients with hyperprolactinemia in both
groups according to gender, along with the mean pro-
lactin values.
For further analyses, unless stated otherwise, we
focused on the antipsychotic-naive sample.
Influence of other previous medications
Of the 30 antipsychotic-naive patients with hyper-
prolactinemia, 20 had taken some other medication:
benzodiazepines (n = 13), biperiden (n = 3), pridinol
Hyperprolactinemia in antipsychotic-naive first-episode patients 2573
https://doi.org/10.1017/S0033291713000226
Downloaded from https://www.cambridge.org/core. WWZ Bibliothek, on 12 Feb 2019 at 16:14:55, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
(n =2), hydroxyzine (n=1), propanolol (n= 1), meto-
clopramide (n =1), cetirizine (n = 1), omeprazol (n=1)
and sulfacetamide (n =1). None of the antipsychotic-
naive women had taken oral contraceptives. There
was no significant difference in prolactin values be-
tween the 20 patients with previous medication
(mean 1.22 ± 0.86 U/l) and the 10 who were completely
medication naive (mean 0.70± 0.30 U/l, Mann–Whitney
U= 64, p = 0.118). Only one of these patients had taken
drugs known to increase prolactin (metoclopramide
and omeprazol). For further analyses we excluded
this patient. We were thus left with 29 out of 74
(39.2%) FEP patients whose hyperprolactinemia could
not be explained by antipsychotics or any other
medication.
Influence of age and gender
There was no significant age difference between the 29
antipsychotic-naive patients with unexplained hyper-
prolactinemia (mean age 25.1 ± 5.4 years) and the 44
antipsychotic-naive with normal prolactin values
(mean age 26.2 ± 5.7 years) (t = 0.8, df =72, p= 0.412),
but there were significant gender differences. Anti-
psychotic-naive women had higher prolactin values
than men, and the range of values was much larger
in the female patients (Fig. 2).
Influence of symptomatology
We also tested whether hyperprolactinemia was
associated with the severity of psychopathological
Table 1. Sociodemographic and clinical characteristics of antipsychotic-naive versus non-naive first-episode patients (FEPs)
Antipsychotic naive
(n= 74)
Antipsychotic non-naive
(n= 175)
Total
(n= 249) p
Men/women, n (%) 52 (70)/22 (30) 105 (60)/70 (40) 157 (63)/92 (37) 0.151
a
Age (years), mean±S.D. 25.8± 2.8 26.0± 5.5 25.9± 5.5 0.739
b
Years of education, mean ± S.D. 12.8 ± 2.8 12.6 ± 3.1 12.7±3.0 0.376
b
Diagnosis, n
Schizophrenic psychosis 46 84 30
Schizo-affective psychosis 6 15 21 0.106
c
Schizophreniform disorders 22 76 98
S.D., Standard deviation.
a
Fisher’s exact test for count data.
b
Mann–Whitney U test for ranked data.
c
χ
2
test for contingency tables.
Table 2. Hyperprolactinemia and prolactin values of antipsychotic-naive versus non-naive patients with first-episode psychosis
Antipsychotic naive Antipsychotic non-naive
Men
(n= 52)
Women
(n= 22)
Total
(n= 74)
Men
(n= 105)
Women
(n= 70)
Total
(n= 175)
Proportion of patients with
hyperprolactinemia
a
, n (%)
19 (36.5) 11 (50.0) 30 (40.5) 75 (71.4) 52 (74.3) 127 (72.6)
Prolactin (U/l) blood levels
Mean±
S.D. 0.37± 0.32 0.98± 0.94 0.55± 0.64 0.83± 0.91 1.60± 1.44 1.14 ± 1.21
Median 0.26 0.58 0.34 0.68 1.16 0.80
Range 0.04–1.50 0.11–3.29 0.04–3.29 0.00–8.58 0.10–6.66 0.00–8.58
Men versus women
Mann–Whitney U test 850.5 4956.0
p value <0.001 <0.001
S.D., Standard deviation.
a
Normal values: men<0.38 U/l; women <0.53 U/l.
2574 A. Riecher-Rössler et al.
https://doi.org/10.1017/S0033291713000226
Downloaded from https://www.cambridge.org/core. WWZ Bibliothek, on 12 Feb 2019 at 16:14:55, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
symptoms and/or diagnostic subtype in antipsychotic-
naive patients. Table 3 shows the different PANSS
and CGI scores and also specific diagnoses for all
antipsychotic-naive patients, comparing the groups
with and without unexplained hyperprolactinemia.
There was no diagnostic subtype disproportionately
more frequently affected by hyperprolactinemia.
Patients with hyperprolactinemia did not differ from
patients without hyperprolactinemia in terms of the
severity of psychopathological symptoms. Further-
more, there was no correlation of log prolactin
values with the total CGI score (Spearman’s ρ = 0.02,
p= 0.850), the PANSS positive score (r = 0.04, t
72
=0.31,
p= 0.756) or the PANSS negative score (r = 0.11, t
72
=
0.95, p = 0.345). Additionally, we tested the correlation
with the anxio-depressive dimension of the PANSS
according to El Yazaji et al. (2002), and this was also
not significant (r = 0.04, t
72
=0.30, p = 0.763).
GLM
In a next step we analyzed the factors potentially
influencing prolactin levels, such as sex, age and
severity of psychopathology, that is PANSS positive
and negative scores, within the 73 antipsychotic-naive
patients. Using an analysis of deviance of a GLM, we
entered age, gender and the PANSS positive and nega-
tive scores into the model.
Age and the PANSS positive score did not contrib-
ute to explaining a significant amount of variance
of the data. However, there were significant main
effects of gender (χ
2
=24.9, df= 1, p < 0.001) and the
PANSS negative score (χ
2
=4.6, df = 1, p =0.032), and
an interaction of gender with the PANSS negative
score (χ
2
=7.9, df =1, p =0.005). The mean prolactin
level in women with an average negative symptom
load (PANSS negative) amounted to 0.98 U/l. By con-
trast, the mean prolactin level in men was only 37%
of the value seen in women. The negative symptoms
(PANSS negative) yielded a differential effect on the
prolactin level depending on gender: whereas an
increasing value of the PANSS negative score was
strongly associated with a decreased prolactin level
in women (–7%/scale unit PANSS negative), a slightly
inverse relationship was found in men (+3%/unit
PANSS negative).
2
Women Men
0
–2
–4
–6
Naive
N = 22
Log prolactin (U/l)
N = 70 N = 51
a
Cut-off: log 0.38 U/l
Cut-off: log 0.53 U/l
N = 105
Not naive Naive
Not naive
Fig. 2. Prolactin values of patients with first-episode psychosis, naive and not naive for antipsychotics, differentiated by
gender. Log prolactin levels depend on gender and neuroleptic treatment. Dotted lines indicate a gender-specific cut-off value.
Values above this line represent hyperprolactinemia. The boxes surrounding some of the values comprise all values that fall
within the first and third quantiles. The horizontal line that is approximately centered within the box corresponds to the
median of the particular subsample.
a
After exclusion of one man (see text).
Hyperprolactinemia in antipsychotic-naive first-episode patients 2575
https://doi.org/10.1017/S0033291713000226
Downloaded from https://www.cambridge.org/core. WWZ Bibliothek, on 12 Feb 2019 at 16:14:55, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
X
