C-reactive protein is associated with negative symptoms of thought disorder
and positive psychotic symptoms in patients with schizophrenia
Victor Vis (11353503)
Begeleider: Bodyl Brand
29-05-2020
Keywords: Formal thought disorder, FTD, C-reactive protein, CRP, schizophrenia, inflammation hypothesis
Abstract
Inflammation may play an important role in the onset and development of schizophrenia. Although several studies have found associations between symptom severity and C-reactive protein (CRP), a biomarker for inflammation, findings are often inconsistent in regard to certain psychotic symptoms and cognitive deficits. A large meta-analysis however, showed a significant correlation between CRP and positive psychotic symptoms. Formal thought disorder (FTD) is a core schizophrenia symptom and has been shown to be an indicator of disease course. The association between CRP and FTD however remains understudied. This association was therefore studied in the current paper. The association between CRP and psychotic symptomatology was also studied to provide more evidence on their relation. FTD was measured using the Thought And Language Disorder (TALD) scale and its positive and negative subscales. Psychotic schizophrenia symptomatology was measured using the Positive And Negative Syndrome Scale (PANSS) and its positive, negative and general subscales. CRP levels were obtained from circulating venous blood in 30 patients (83% male) diagnosed with schizophrenia. BMI and smoking behaviour were corrected for in the associations with CRP. A significant association was found between CRP and negative TALD score (p=0.030, R2=0.165) but not with total or positive TALD score. According to the PANSS, CRP
was significantly associated with positive symptoms (p=0.015, R2= 0.153) but not total, negative or
general symptomatology. CRP appears to be positively associated with the negative symptoms of FTD and positive psychotic symptomatology. These results suggests an association between FTD and CRP and are supplemental evidence for the inflammation hypothesis in schizophrenia.
Introduction
Schizophrenia is a debilitating disorder with a wide array of symptoms. The global prevalence is just under 1% but great diversity exists between geographical regions (Kahn et al., 2015). The symptoms have a common division into positive and negative symptoms (Kahn et al., 2015). Positive symptoms add
something to the reality of the patient, such as hallucinations, delusions or disorganisation, while negative symptoms are more characterised by the absence of thoughts or behaviour, such as anhedonia, lethargy or blunted affect. This division however, does not encompass all symptoms present in schizophrenia. An overarching core symptom of schizophrenia is formal thought disorder (FTD). FTD is a collection of multiple dysfunctions characterised by a defective thought and language production system, resulting in non-fluent and disorganised flow of speech and thought (Bora, Yalincetin, Akdede, & Alptekin, 2019). Similarly to the symptomatology of schizophrenia, FTD dysfunctions can be divided into positive and negative subgroups. Positive FTD is characterised by symptoms such as tangentiality, where patients completely digress from the original topic, or lack of content, where patients talk with an adequate amount of words but with barely any substance. Negative FTD is characterised by symptoms such as speech poverty or such as slowed thinking, where patients take long pauses between and/or before answers.
Two meta-analyses indicated FTD as a marker for psychosis severity (Roche, Creed, Macmahon, Brennan, & Clarke, 2015) and linked FTD to certain cognitive deficits (Kerns & Berenbaum, 2002). Furthermore, certain symptoms of FTD have been shown to be good predictors of poor clinical outcome in patients with schizophrenia (Wilcox, Winokur, & Tsuang, 2012). Having more severe FTD during first psychosis onset predicted more severe symptomatology 10-20 years later (Wilcox et al., 2012). Early treatment of FTD in patients might therefore enhance the course of illness. Lastly, general functioning and neurocognitive profile is worse in patients with moderate/severe FTD compared to patients with mild FTD (Comparelli et al., 2020). The moderate/severe FTD group scored significantly lower on reasoning and problem solving, speed of processing and social cognition. Symptoms of FTD interfere with many aspects of ‘normal’ functioning in a personal or professional environment, especially when combined with the other symptoms of schizophrenia. Indeed patients suffering from schizophrenia have been shown to have significantly lower quality of life compared to healthy controls (Bechdolf et al., 2005; Sidlova et al., 2011). Not only patients themselves, however, suffer from this disorder, as wider society bares a substantial financial burden. Annually, up to US$102 billion is spent on patients with schizophrenia (Chong et al., 2016). Indirect costs appeared to contribute 50%–85% of the total immense expenditure. This
encompasses mortality costs due to premature death and productivity losses due to symptoms, such as sick leave and unemployment. There is a need to decrease symptom severity in patients with
schizophrenia to increase their quality of life and to reduce associated costs. In order to reduce symptomatology of schizophrenia, it is important to understand the underlying pathophysiology. There is a great body of evidence supporting a link between immunological factors and schizophrenia. A large review on the association between major psychiatric disorders and the immune system showed that a dysfunctional immune system may be involved in the etiology of schizophrenia (Gibney & Drexhage, 2013). Additionally, a dysregulated monocyte activation is found in patients with schizophrenia (Müller et al., 2012). This would cause pathogens to remain within the body for longer, which could in turn lead to chronic inflammation. Furthermore, a large study on the association between autoimmune diseases and
schizophrenia was comprised of data of 7,704 patients (Eaton et al., 2006). It found that five autoimmune disorders (thyrotoxicosis, acquired hemolytic anemia, intestinal malabsorption, interstitital cystitis, and Sjögren’s syndrome) had a higher incidence in patients prior to schizophrenia onset as well as in patients’ parents. Another large, more recent study included 6,479 healthy individuals (Begemann et al., 2019). This study showed that the presence of atopic disorders such as asthma, eczema and allergic rhinitis, significantly increased psychotic experiences, especially hallucinations. Lastly, a systematic review showed that certain prenatal infections may be associated with development of schizophrenia in adulthood (Khandaker, Zimbron, Lewis, & Jones, 2013). This body of research supports the inflammatory hypothesis which links chronic inflammation to schizophrenia (Feigenson, Kusnecov, & Silverstein, 2014).
An acute inflammatory protein secreted by the liver called C-reactive protein (CRP), is a key biomarker of inflammation in research (Lopresti, Maker, Hood, & Drummond, 2014; Sproston & Ashworth, 2018). CRP appears to play a regulatory role in the inflammation response (Sproston & Ashworth, 2018), indicating it as a possible treatment option according to the inflammation hypothesis. Indeed, elevated CRP levels have repeatedly been found in patients with schizophrenia (Fernandes et al., 2016) and proven to be a biomarker for onset risk and symptom severity (Fond, Lançon, Auquier, & Boyer, 2018). Research on the exact relation between CRP and schizophrenia however remains indecisive. One large meta-analysis including a total of 85,000 participants showed moderately elevated levels of CRP in patients with schizophrenia regardless of antipsychotics use (Fernandes et al., 2016). They also showed a correlation between the elevation of CRP and positive symptom severity but not negative symptom severity. In contrast, other studies found an association between elevated CRP levels and negative symptoms but not positive symptoms (Boozalis, Teixeira, Cho, & Okusaga, 2018; Joseph et al., 2015). Yet another study, including 413 patients, found an association between CRP and cognitive impairments but not any psychotic symptoms (Dickerson, Stallings, Origoni, Boronow, & Yolken, 2007). These studies show that while the exact relation between schizophrenia and CRP remains elusive, elevated levels are noticed, as well as an association with symptomatology.
Presently, the relation between CRP and FTD remains understudied. Chang et al. (2019) found a significant relation between CRP and FTD as measured with the Thought And Language Disorder scale (TALD, Kircher et al., 2014) and specifically with the objective positive subscale of the TALD (Chang et al., 2019). However, this study had a limited sample size of 33 participants and it did not correct for smoking, which has a proven effect on CRP levels (Hage & Szalai, 2007; Kushner, Rzewnicki, & Samols, 2006). Therefore, there is a paucity in evidence for a relation between CRP and FTD.
This study aimed to investigate the relation between CRP and FTD and the relation between CRP and psychotic symptoms. The correlation between the elevation of CRP levels and severity of FTD, as well as the severity of psychotic symptoms was looked at in patients diagnosed with schizophrenia. It is
hypothesized that elevation of CRP levels coincide with more severe FTD and more severe positive psychotic symptoms. CRP levels were measured in the blood, FTD was tested using the TALD scale and psychotic symptoms were tested using the Positive And Negative Syndrome Scale (PANSS). Based on previous research (Chang et al., 2019; Fernandes et al., 2016), higher levels of circulating CRP are expected to be associated with increased total TALD scores and particularly the positive subscale of the TALD. Furthermore, CRP is expected to be associated with positive PANSS scores.
Methods
Study populationData used in the current study was obtained as part of a larger study known as Raloxifene Augmentation in Patients with a Schizophrenia spectrum Disorder (RAPSODI, UMCU, 2016). This study was approved by the medical ethical commission of University Medical Centre (UMC) Utrecht and followed the declaration of Helsinki guidelines. Rapsodi is a randomised, placebo controlled, clinical trial to test the effect of raloxifene augmentation. Only baseline measurements were used in the current study to avoid any effects of the given medication. In- and outpatients throughout the Netherlands with a schizophrenia disorder were included, selected by doctors and researchers connected to Rapsodi. Diagnosis was dependant on the Mini-International Neuropsychiatric Interview-Plus version 5.0.0. (M.I.N.I.-plus, Sheehan et al., 1998), based on the Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV, American Psychiatric Association, 2000). Patients had to be on a stable dose of antipsychotic medication for at least two weeks prior to study enrolment. All participants were over eighteen years of age and capable of understanding the purpose and details of this study in order to provide written informed consent. Exclusion criteria
Patients with any history of (breast) cancer or cardiovascular diseases were excluded, as well as patients with possible kidney or liver failure (failure, as defined by over three times the upper normal limit). Glomerular filtration rate (abnormal defined by <30ml/min) and creatine for kidney function and abnormal serum bilirubin, alkalic phosphatase, gamma-glutamyl transpeptidase (γ-GT), aspartate aminotransferase (ASAT) or alanine aminotransferase (ALAT) levels for liver function. Also patients with hypertriglyceridemia (triglycerides > 6 mmol/L) and patients receiving oestrogen or androgens as hormonal therapy were excluded.
Measurements
All measurements took place on the same day. Demographics
Patients were asked about their smoking behaviour (cigarettes or self-rolled shag per week), medication and disease complaints beside schizophrenia. Length and weight were used to calculate body mass index (BMI) and blood pressure was measured. In the MINI-plus we enquired about the date of the first psychotic complaints which was used to determine duration of illness.
CRP levels
CRP levels were measured in circulating venous blood. Blood was collected at the blood clinic and analysed in the lab at UMC Utrecht.
Formal thought disorder
Formal thought disorder was measured by using a Dutch translation of the 30-item thought and language disorder scale (Kircher et al., 2014 translated in Dutch by S.B.J. Oude Ophuis & O. van de Goor) consisting of objective positive, objective negative, subjective positive and subjective negative items. For each item there is a score of zero, when the symptom is absent, to four, when the symptom is present in extreme
form, leading to a minimum score of zero and a maximum score of 120. Scoring was done by trained researchers. Objective items (1-21) were scored based on use of language during the whole visit. The first eighteen items (1-18) are positive FTD items and the last three (19-21) are negative FTD items. Subjective items (22-30) were scored using a standard list of nine statements with the question to what extent patients found themselves in those statements and whether it bothered them. The first seven (22-28) statements are negative FTD items and the last two (29 and 30) are positive FTD items.
Psychotic symptomatology
In order to measure the overall symptom severity, the standard 30-item Positive And Negative Syndrome Scale (PANSS) was used (Kay, Fiszbein, & Opler, 1987). The PANSS is divided into seven positive, seven negative and sixteen general items. The scoring is based on a semi-structured interview and each item is scored on a basis of one to seven. One is absent and seven is the most extreme form of the symptom. The PANSS interviews and scoring were always done by two raters to minimise interrater variability.
Data analysis
Due to the abundant presence of patients with low CRP levels with small, less meaningful concentration differences between them, a cut-off value of 3 mg/L (Rifai & Ridker, 2003) was used. CRP values were log transformed to counter the right skewness and better comply to the assumption of linearity. Missing data of covariates was replaced by the grand mean of the study sample. All diseases that potentially cause inflammation were checked and no consistent relation was found between these diseases and elevated CRP levels. Also patients who had administered drugs with anti-inflammatory effects were checked. Again no consistent relation was found between the anti-inflammatory drugs and CRP levels.
Statistical analysis was performed in R-studio 1.0.153 (R Core Team, 2017), using the “car” package for the regression analysis (Fox & Weisberg, 2011). All tests were two-sided with a significance level of α=0.05. Assumptions for linear regression were checked. Linearity was visually checked, multivariate normality was checked using a Shapiro-Wilk test over the model residuals, homoscedasticity was tested using a Breusch-Pagan test and multicollinearity was checked by measuring the variance inflation factor (VIF). Means and standard deviations (SD) were calculated for age, BMI, blood pressure, smoking behaviour and duration of illness as well as all test scores and subscales. Seven univariate and seven multivariate
regression analyses were performed. The association between CRP and total TALD score, as well as the association between CRP and the total positive and the total negative subscale scores were evaluated. Also the association between CRP and the total PANSS, the positive PANSS, the negative PANSS and the general PANSS scores was tested. Relevant covariates that showed an association with CRP levels were incorporated into each of the models.
Results
A total of 30 schizophrenia patients were included in the present study, of which 83% were male. 45 patients were excluded from analysis due to low CRP levels. Demographic data on age, BMI, blood pressure, smoking behaviour, CRP levels and duration of illness can be seen in Table 1. Mean test scores on the total TALD and its total positive and total negative sub scales, as well as the total PANSS and its sub
scales, can be found in Table 1. Age, duration of illness and blood pressure were not associated with CRP and were therefore excluded from further analysis.
Table 1
Demographic data and test scores of participating patients.
Variable Mean SD
Age 39.4 ±10.16
BMI 28.02 ±4.75
Systolic blood pressure 126.26 ±13.10
Smoking behaviour (per week) 14.80 ±13.12
CRP levels (mg/L) 6.76 ±3.82
Log CRP levels 1.78 ±0.49
Duration of illness (years) 14.41 ±8.47
Total TALD score 18.27 ±8.69
- Total positive TALD score 8.33 ±5.03 - Total negative TALD score 9.93 ±5.81
Total PANSS score 57.07 ±15.08
- Positive PANSS score 14.80 ±4.12 - Negative PANSS score 12.80 ±5.19
- General PANSS score 29.47 ±8.28
Note. Demographic data and test scores of participants that were included in the analysis are shown. The grand sample
mean is in the middle column and the standard deviation in the third column.
SD= standard deviation, BMI = Body Mass Index, CRP = C-Reactive Protein, TALD = Thought And Language Disorder, PANSS = Positive And Negative Syndrome Scale. Smoking behaviour is expressed in amount of cigarettes per week. CRP levels are in mg/L. Age and duration of illness are in years.
CRP and TALD test scores
CRP significantly predicted negative TALD score in a univariate regression model
(F(1,28)=4.567,p=0.041) with an R2 of 0.140 and a beta-coefficient of 4.402 (t(28)=2.137,p=0.041) (Table
2). After correcting for BMI and smoking behaviour in a multivariate regression model, CRP still significantly predicted the negative TALD score (F(1,26)=5.259,p=0.030) with an R2 of 0.165 and a
beta-coefficient of 5.116 (t(26)=2.293,p=0.30). Taken together however, CRP, BMI and smoking did not significantly predict total TALD score (F(3,26)=1.985,p=0.141) with an R2 of 0.186 (Table 2).
Positive TALD score was in a univariate regression model not significantly predicted by CRP (F(1,28)=0.368,p=0.549) with an R2 of 0.013 and a beta-coefficient of -1.158 (t(28)=-0.606,p =0.549) (Table
3). After correcting for BMI and smoking behaviour, CRP did still not significantly predict positive TALD score (F(1,26)=0.064,p=0.802) with an R2 of 0.002 and a beta-coefficient of 0.477 (t(26)=0.253,p=0.802).
Taken together, CRP, BMI and smoking did not significantly predict positive TALD score (F(3,26)=2.527,p=0.079) with an R2 of 0.226 (Table 3).
Total TALD score was in a univariate regression model also not significantly predicted by CRP (F(1,28)=0.987,p=0.329) with an R2 of 0.034 and a beta-coefficient of 3.244 (t(28)=0.993,p=0.329) (Table
(F(1,26)=2.854,p=0.103) with an R2 of 0.087 and a beta-coefficient of 5.593 (t(26)=1.689,p=0.103). Taken
together, CRP, BMI and smoking did not significantly predict total TALD score (F(3,26)=2.156,p=0.118) with an R2 of 0.199 (Table 4).
Table 2
Negative TALD score regression models.
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for negative TALD score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
TALD = Thought And Language Disorder, CRP = C-Reactive Protein, BMI = Body Mass Index.
Table 3
Positive TALD score regression models.
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for positive TALD score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
TALD = Thought And Language Disorder, CRP = C-Reactive Protein, BMI = Body Mass Index.
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP 4.402 [0.182,8.622] 2.137 0.041* 0.140 4.567 0.041*
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.186 1.985 0.141
CRP 5.116 [0.530,9.702] 2.293 0.030* 0.165 5.259 0.030*
BMI -0.032 [-0.497,0.433] -0.141 0.889 0.010 0.327 0.573
smoking -0.099 [-0.267,0.069] -1.213 0.236 0.012 0.368 0.549
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP -1.158 [-5.072,2.755] -0.606 0.549 0.013 0.368 0.549
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.226 2.527 0.079
CRP 0.477 [-3.395,4.349] 0.253 0.802 0.002 0.064 0.802
BMI -0.205 [-0.597,0.188] -1.073 0.293 0.020 0.657 0.425
Table 4
Total TALD score regression models.
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for total TALD score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
TALD = Thought And Language Disorder, CRP = C-Reactive Protein, BMI = Body Mass Index.
CRP levels and PANSS test scores
After a logarithmic transformation on total PANSS and positive PANSS scores and a reciprocal transformation on negative PANSS score, the assumptions for regression analysis were met for each PANSS model.
Positive PANSS score was significantly predicted by CRP in a univariate regression model (F(1,28)=5.320,p=0.029) with an R2 of 0.160 and a beta-coefficient of 0.232 (t(28)=2.306,p=0.029) (Table
5). After correcting for BMI and smoking behaviour in a multivariate regression model, CRP still significantly predicted positive PANSS score (F(1,26)=6.722,p=0.015) with an R2 of 0.153 and a
beta-coefficient of 0.244 (t(26)=2.593,p=0.015). Taken together, CRP, BMI and smoking also significantly predicted positive PANSS score (F(3,26)=6.012,p=0.003) with an R2 of 0.410 (Table 5).
CRP did not significantly predict negative PANSS score in a univariate regression model
(F(1,28)=1.046,p=0.315) with an R2 of 0.036 and a beta-coefficient of -0.011 (t(28)=-1.023,p=0.315) (Table
6). After correcting for BMI and smoking behaviour, CRP did still not significantly predict negative PANSS score (F(1,26)=0.313,p=0.581) with an R2 of 0.011 and a beta-coefficient of -0.007 (t(26)=-0.559,p=0.581).
Taken together, CRP, BMI and smoking did not significantly predict negative PANSS score (F(3,26)=0.752,p=0.531) with an R2 of 0.080 (Table 6).
General PANSS score was not significantly predicted by CRP in a univariate regression model (F(1,28)=1.010,p=0.323) with an R2 of 0.035 and a beta-coefficient of 3.127 (t(28)=1.005,p=0.323) (Table
7). After correcting for BMI and smoking behaviour, CRP did still not significantly predict general PANSS score (F(1,26)=0.963,p=0.336) with an R2 of 0.035 and a beta-coefficient of 3.351 (t(26)=0.981,p=0.336).
Taken together, CRP, BMI and smoking did not significantly predict general PANSS score (F(3,26)=0.575,p=0.637) with an R2 of 0.062 (Table 7).
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP 3.244 [-3.446,9.934] 0.993 0.329 0.034 0.987 0.329
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.199 2.156 0.118
CRP 5.593 [-1.212,12.398] 1.689 0.103 0.087 2.854 0.103
BMI -0.237 [-0.926,0.453] -0.706 0.487 0.000 0.006 0.940
Total PANSS score was not significantly predicted by CRP in a univariate regression model (F(1,28)=3.115,p=0.088) with an R2 of 0.100 and a beta-coefficient of 0.159 (t(28)=1.765,p=0.088) (Table
8). After correcting for BMI and smoking behaviour, CRP did still not significantly predict total PANSS score (F(1,26)=2.429,p=0.131) with an R2 of 0.077 and a beta-coefficient of 0.150 (t(26)=1.558,p=0.131). Taken
together, CRP, BMI and smoking did not significantly predict total PANSS score either (F(3,26)=1.816,p=0.169) with an R2 of 0.173 (Table 8).
Table 5
Positive PANSS score regression models.
Table 6
Negative PANSS score regression models.
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP 0.232 [0.026,0.438] 2.306 0.029* 0.160 5.320 0.029*
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.410 6.012 0.003**
CRP 0.244 [0.050,0.437] 2.593 0.015* 0.153 6.722 0.015*
BMI 0.018 [-0.002,0.037] 1.870 0.072 0.186 8.207 0.008**
smoking -0.008 [-0.015,-0.001] -2.406 0.024* 0.071 3.107 0.090
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for positive PANSS score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
PANSS = Positive And Negative Syndrome Scale, CRP = C-Reactive Protein, BMI = Body Mass Index.
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP -0.011 [-0.034,0.011] -1.023 0.315 0.036 1.046 0.315
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.080 0.752 0.531
CRP -0.007 [-0.031,0.018] -0.559 0.581 0.011 0.313 0.581
BMI -0.001 [-0.003,0.002] -0.571 0.573 0.015 0.418 0.524
smoking -0.000 [-0.001,0.000] -1.034 0.311 0.054 1.526 0.228
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for negative PANSS score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
Table 7
General PANSS score regression models.
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for general PANSS score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
PANSS = Positive And Negative Syndrome Scale, CRP = C-Reactive Protein, BMI = Body Mass Index.
Table 8
Total PANSS score regression models.
Discussion
Results showed that CRP significantly predicted negative TALD score. After correction for BMI and smoking behaviour, which have shown to have an effect on CRP level, CRP still significantly predicted negative TALD score, explaining 16,5% of its variation. Neither total TALD nor positive TALD score were significantly predicted by CRP. Furthermore, CRP did significantly predict positive PANSS score. After
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP 3.127 [-3.247,9.501] 1.005 0.323 0.035 1.010 0.323
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.062 0.575 0.637
CRP 3.351 [-3.668,10.369] 0.981 0.336 0.035 0.963 0.336
BMI 0.142 [-0.569,0.853] 0.411 0.685 0.020 0.567 0.458
smoking -0.087 [-0.344,0.170] -0.693 0.495 0.007 0.194 0.663
Univariate model B [95% CI] t-value p-value R2 F-value p-value
CRP 0.159 [-0.026,0.343] 1.765 0.088 0.100 3.115 0.088
Multivariate model B [95% CI] t-value p-value R2 F-value p-value
Total - - - 0.173 1.816 0.169
CRP 0.150 [-0.048,0.347] 1.558 0.131 0.077 2.429 0.131
BMI 0.011 [-0.009,0.031] 1.164 0.255 0.092 2.890 0.101
smoking -0.003 [-0.010,0.005] -0.774 0.446 0.004 0.130 0.721
Note. Data on the univariate regression with CRP (second row) and multivariate regression with CRP, BMI and smoking
(fourth to last row) for total PANSS score. The beta-coefficients per variate with a 95% confidence interval in the second
column. Its corresponding t- and p-values in the third and fourth column. The R2 in the fifth column with the corresponding
F- and p-values in the sixth and seventh column.
correction for BMI and smoking, CRP still significantly predicted positive PANSS score, explaining 15.3% of its variation. CRP did not significantly predict negative, general or total PANSS score. The hypothesis that elevated levels of CRP indicate more severe FTD is partially supported while the hypothesis that elevated levels of CRP indicate more severe positive psychotic symptomatology is fully supported by the current data. An increase in CRP was associated with positive psychotic symptoms and with negative symptoms of FTD but not with positive symptoms of FTD.
The expectation that CRP would show a stronger association with positive symptoms of FTD was not met. With a highest score of 21 out of a possible 80 and a SD of 5.03, the mean positive TALD score was relatively low in our sample, which means that there was little variation on which there could have been an effect. More variation in TALD scores might have increased chances to detect any associations. An interesting finding is that CRP was positively associated with both negative symptoms of FTD and positive psychotic symptomatology. Although the positive/negative division in FTD symptoms does not necessarily match the positive/negative division in psychotic symptomatology, it is conceivable that patients who show more positive psychotic symptoms, which add something to the reality to the patients, show more positive symptoms of FTD, such as increased speech. On the other hand, it might be expected that patients who show more negative psychotic symptoms, which take something away from the reality of the patients, show more negative symptoms of FTD, including decreased amount of speech or thought. A possible explanation for this discrepancy could be the distinctive scoring mechanisms. For the total negative TALD score, seven out of ten items are scored subjectively, whereas each PANSS subscale is entirely scored objectively by researchers. This would however not be an explanation of the absence of an association between CRP and positive TALD score, where eighteen out of twenty items are scored
objectively, similarly to all positive PANSS items.
Our findings with regard to symptoms of FTD do not correspond with previous work. Chang et al. (2019) found a positive association between CRP level and both total TALD score and objective positive TALD score, while the current results show a positive association between CRP and total negative TALD score but not total TALD nor total positive TALD score. There were some distinctions in methods used between the current study and Chang et al. (2019) warranting possibilities for different results. Whereas Chang et al. (2019) included all 33 patients whose CRP levels (µ=0.405 mg/L) were measured in the analysis, the current study only included 30 patients with elevated levels of CRP (µ=6.76 mg/L) in the analysis.
Furthermore Chang et al. (2019) incorporated gender, atypical antipsychotics use, age and illness duration as covariates into each model, while the current analysis included BMI and smoking behaviour as
covariates. Lastly, Chang et al. (2019) tested each of the four TALD subscales separately whereas the current study disregarded the objective/subjective division in order to create two larger positive and negative subscales. To my knowledge, these two studies are the only two studies to look at the
association between FTD and CRP. Additional research is needed in order to make a more valid claim on the true association between CRP and FTD.
One finding of Chang et al. (2019) however, was reproduced in the current study. CRP had a significant positive association with positive PANSS score. This finding is also in accordance with the meta-analysis of Fernandes et al. (2016) who found a correlation between positive schizophrenia symptoms and Bolu et al. (2019) who found an association between CRP and positive schizophrenia symptoms in both patients with first-episode psychosis and patients diagnosed with schizophrenia.
The exact biological mechanism of CRP remains not fully understood. However, a possible explanation for the association between CRP and schizophrenia symptomatology might be the effect of CRP on the blood brain barrier (BBB). CRP has been shown to increase BBB permeability (Hsuchou, Kastin, Mishra, & Pan, 2012; Shcherbakova et al., 1999). This could lead to an increase of cytokines and CRP in the central nervous system, which might induce microglial activation (Adami et al., 2001). Pro-inflammatory cytokine production would be increased, including the production of interleukin 6 (Smith, Das, Ray, & Banik, 2012). This can lead to both neurodegeneration (Smith et al., 2012) and to the metabolism of kynurenine to kynurenic acid (KYNA, Schwieler et al., 2015). KYNA functions as an antagonist on NMDA receptors, comparable with the function of ketamine (Schwarcz, Bruno, Muchowski, & Wu, 2012). Ketamine has been shown to induce schizophrenia-like symptoms in healthy individuals and to increase psychotic symptoms in patients with schizophrenia (Schwarcz et al., 2012). In conclusion, CRP might set in motion a chain of events leading to both neurodegeneration and increased levels of KYNA.
Some aspects of the current study need further clarification. As previously mentioned, the TALD score was divided into two subscales of total positive and total negative TALD score, as opposed to four smaller subscales including an additional separation in objective and subjective subscale. The subjective positive subscale exists of only two items and the objective negative subscale exists of only three items. This would lead to too little possible variation for these subscales. They were therefore combined with their objective positive and subjective negative counterparts respectively to create the two larger total positive and total negative subscales.
Another aspect that might have influenced the current results is sample size. Many patients had low CRP levels with very little variation in between them. A small difference in CRP level in patients with no elevated CRP is less meaningful than a larger difference in CRP level in people with elevated CRP. A cut-off value of three mg/L was used. This is a commonly used cut-off value regarding elevated CRP and roughly encompasses the upper tertile of the population (Rifai & Ridker, 2003). This lead to the exclusion of 45 patients. Even though it considerably reduces the power of the models, it was an inevitable step in order to test our hypothesis.
Lastly, antipsychotic use was not incorporated in the current analysis. Even though some types of anti-psychotic medication have been shown to have certain anti-inflammatory effects (Tourjman et al., 2013), a meta-analysis by Fernandes et al. (2016) showed that CRP level is not altered by antipsychotics.
Furthermore, the effect of various antipsychotic medication on symptomatology differs resulting in an unsure relation between test scores and antipsychotic use (Jibson & Tandon, 1998).
The current results provide evidence that besides psychotic symptoms, also certain symptoms of FTD appear to be influenced by CRP level. This is further evidence for the inflammation hypothesis in
schizophrenia and particularly for a role of CRP. As mentioned before, the precise role of CRP remains not fully understood but CRP does appear to play a regulatory role in inflammation processes (Sproston & Ashworth, 2018) and it might set in motion a chain of events leading to neurodegeneration and increased levels of KYNA. It is therefore a potential target for treatment. A longitudinal double-blind placebo-controlled medication study is proposed with an anti-inflammatory drug. A potential candidate is salsalate, a relatively old non-steroidal anti-inflammatory drug (NSAID) used in the treatment of rheumatoid arthritis. Recently it has been getting attention in research as a potential type 2 diabetes treatment, with its ability to lower insulin resistance and decrease glucose levels through its effect on an
inflammatory pathway (Anderson, Wherle, Park, Nelson, & Nguyen, 2014). Salsalate reduces inflammation and it has been shown to significantly lower CRP levels in non-diabetic obese people (Fleischman, Shoelson, Bernier, & Goldfine, 2008; Koska et al., 2009). It would be of interest to find out the effect of a long-term anti-inflammatory treatment with a drug such as salsalate on CRP levels, FTD and overall schizophrenia symptomatology, especially as negative FTD has been shown to have predictive value of a worsened course of illness (Wilcox et al., 2012).
In conclusion, CRP predicted specific psychotic symptoms as well as certain symptoms of FTD. A treatment that lowers CRP levels, might reduce these specific symptoms and therefore might enhance the course of illness. Not only the quality of life for the patients with schizophrenia would increase, but also the immense costs associated with schizophrenia would decrease.
References
Adami, C., Sorci, G., Blasi, E., Agneletti, A. L., Bistoni, F., & Donato, R. (2001). S100B Expression in and Effects on Microglia. Glia, 33, 131–142.
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author.
Anderson, K., Wherle, L., Park, M., Nelson, K., & Nguyen, L. D. (2014). Salsalate, an old, inexpensive drug with potential new indications: A review of the evidence from 3 recent studies. American Health and Drug
Benefits, 7(4), 231–235.
Bechdolf, A., Pukrop, R., Köhn, D., Tschinkel, S., Veith, V., Schultze-Lutter, F., … Klosterkötter, J. (2005).
Subjective quality of life in subjects at risk for a first episode of psychosis: A comparison with first episode schizophrenia patients and healthy controls. Schizophrenia Research, 79(1), 137–143.
https://doi.org/10.1016/j.schres.2005.06.008
Begemann, M. J. H., Linszen, M. M. J., De Boer, J. N., Hovenga, W. D., Gangadin, S. S., Schutte, M. J. L., & Sommer, I. E. C. (2019). Atopy increases risk of psychotic experiences: A large population-based study.
Frontiers in Psychiatry, 10(JULY), 1–7. https://doi.org/10.3389/fpsyt.2019.00453
Bolu, A., Aydn, M. S., Akgün, A., Cokun, A., Garip, B., Öznur, T., … Uzun, Ö. (2019). Serum levels of high sensitivity c-reactive protein in drug-naïve first-episode psychosis and acute exacerbation of schizophrenia. Clinical Psychopharmacology and Neuroscience, 17(2), 244–249.
https://doi.org/10.9758/cpn.2019.17.2.244
Boozalis, T., Teixeira, A. L., Cho, R. Y.-J., & Okusaga, O. (2018). C-Reactive Protein Correlates with Negative Symptoms in Patients with Schizophrenia. Frontiers in Public Health, 5(January), 1–6.
https://doi.org/10.3389/fpubh.2017.00360
Bora, E., Yalincetin, B., Akdede, B. B., & Alptekin, K. (2019). Neurocognitive and linguistic correlates of positive and negative formal thought disorder: A meta-analysis. Schizophrenia Research, 209, 2–11.
https://doi.org/10.1016/j.schres.2019.05.025
Chang, C. H., Lane, H. Y., Liu, C. Y., Cheng, P. C., Chen, S. J., & Lin, C. H. (2019). C-reactive protein is associated with severity of thought and language dysfunction in patients with schizophrenia. Neuropsychiatric
Disease and Treatment, 15, 2621–2627. https://doi.org/10.2147/NDT.S223278
Chong, H. Y., Teoh, S. L., Wu, D. B.-C., Kotirum, S., Chiou, C.-F., & Chaiyakunapruk, N. (2016). Global economic burden of schizophrenia : a systematic review. Neuropsychiatric Disease and Treatment, 12, 357–373. https://doi.org/http://dx.doi.org/10.2147/NDT.S96649
Comparelli, A., Corigliano, V., Forcina, F., Bargagna, P., Montalbani, B., Falcone, G., … Pompili, M. (2020). The Complex Relationship among Formal Thought Disorders, Neurocognition, and Functioning in Nonacutely Ill Schizophrenia Patients. Journal of Nervous and Mental Disease, 208(1), 48–55.
https://doi.org/10.1097/NMD.0000000000001087
Dickerson, F., Stallings, C., Origoni, A., Boronow, J., & Yolken, R. (2007). C-reactive protein is associated with the severity of cognitive impairment but not of psychiatric symptoms in individuals with schizophrenia.
Schizophrenia Research, 93(1–3), 261–265. https://doi.org/10.1016/j.schres.2007.03.022
Eaton, W. W., Byrne, M., Ewald, H., Mors, O., Chen, C. Y., Agerbo, E., & Mortensen, P. B. (2006). Association of schizophrenia and autoimmune diseases: Linkage of Danish national registers. American Journal of
Psychiatry, 163(3), 521–528. https://doi.org/10.1176/appi.ajp.163.3.521
Feigenson, K. A., Kusnecov, A. W., & Silverstein, S. M. (2014). Inflammation and the Two-Hit Hypothesis of Schizophrenia. Neurosci Biobehav Rev., 38, 72–93. https://doi.org/10.1016/j.neubiorev.2013.11.006 Fernandes, B. S., Steiner, J., Bernstein, H. G., Dodd, S., Pasco, J. A., Dean, O. M., … Berk, M. (2016). C-reactive
Molecular Psychiatry, 21(4), 554–564. https://doi.org/10.1038/mp.2015.87
Fleischman, A., Shoelson, S. E., Bernier, R., & Goldfine, A. B. (2008). Salsalate improves glycemia and inflammatory parameters in obese young adults. Diabetes Care, 31(2), 289–294.
https://doi.org/10.2337/dc07-1338
Fond, G., Lançon, C., Auquier, P., & Boyer, L. (2018). C-reactive protein as a peripheral biomarker in schizophrenia. An updated systematic review. Frontiers in Psychiatry, 9(AUG), 1–12.
https://doi.org/10.3389/fpsyt.2018.00392
Fox, J., & Weisberg, S. (2011). An {R} Companion to Applied Regression, Second Edition. Thousand Oaks CA: sage. Retrieved from http://socserv.socsci.mcmaster.ca/jfox/Books/Companion
Gibney, S. M., & Drexhage, H. A. (2013). Evidence for a dysregulated immune system in the etiology of psychiatric disorders. Journal of Neuroimmune Pharmacology, 8(4), 900–920.
https://doi.org/10.1007/s11481-013-9462-8
Hage, F. G., & Szalai, A. J. (2007). C-Reactive Protein Gene Polymorphisms, C-Reactive Protein Blood Levels, and Cardiovascular Disease Risk. Journal of the American College of Cardiology, 50(12), 1115–1122.
https://doi.org/10.1016/j.jacc.2007.06.012
Hsuchou, H., Kastin, A. J., Mishra, P. K., & Pan, W. (2012). C-Reactive Protein Increases BBB Permeability : Implications for Obesity and Neuroinflammation. Cellular Physiology and Biochemistry, 30, 1109–1119. https://doi.org/10.1159/000343302
Jibson, M. D., & Tandon, R. (1998). New atypical antipsychotic medications. Journal Of Psychiatric Research,
32(3–4), 215–228. https://doi.org/https://doi.org/10.1016/S0022-3956(98)00023-5
Joseph, J., Depp, C., Martin, A. S., Daly, R. E., Glorioso, D. K., Palmer, B. W., & Jeste, D. V. (2015). Associations of high sensitivity C-reactive protein levels in schizophrenia and comparison groups. Schizophrenia
Research, 168(1–2), 456–460. https://doi.org/10.1016/j.schres.2015.08.019
Kahn, R. S., Sommer, I. E., Murray, R. M., Meyer-Lindenberg, A., Weinberger, D. R., Cannon, T. D., … Insel, T. R. (2015). Schizophrenia. Nature Reviews Disease Primers, 1(November).
https://doi.org/10.1038/nrdp.2015.67
Kay, S. R., Fiszbein, A., & Opler, L. A. (1987). The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophrenia Bulletin, 13(2), 261–276. https://doi.org/10.1093/schbul/13.2.261 Kerns, J. G., & Berenbaum, H. (2002). Cognitive Impairments Associated With Formal Thought Disorder in
People With Schizophrenia. Journal of Abnormal Psychology, 111(2), 211–224. https://doi.org/10.1037//0021-843X.111.2.211
Khandaker, G. M., Zimbron, J., Lewis, G., & Jones, P. B. (2013). Prenatal maternal infection, neurodevelopment and adult schizophrenia: A systematic review of population-based studies. Psychological Medicine, 43(2), 239–257. https://doi.org/10.1017/S0033291712000736
Kircher, T., Krug, A., Stratmann, M., Ghazi, S., Schales, C., Frauenheim, M., … Nagels, A. (2014). A rating scale for the assessment of objective and subjective formal thought and language disorder (TALD). Schizophrenia
Research, 160(1–3), 216–221. https://doi.org/10.1016/j.schres.2014.10.024
Koska, J., Ortega, E., Bunt, J. C., Gasser, A., Impson, J., Hanson, R. L., … Krakoff, J. (2009). The effect of salsalate on insulin action and glucose tolerance in obese non-diabetic patients: Results of a randomised double-blind placebo-controlled study. Diabetologia, 52(3), 385–393. https://doi.org/10.1007/s00125-008-1239-x
Kushner, I., Rzewnicki, D., & Samols, D. (2006). What does minor elevation of C-reactive protein signify?
American Journal of Medicine, 119(2), 166.e17-166.e28. https://doi.org/10.1016/j.amjmed.2005.06.057
Lopresti, A. L., Maker, G. L., Hood, S. D., & Drummond, P. D. (2014). A review of peripheral biomarkers in major depression: The potential of inflammatory and oxidative stress biomarkers. Progress in
https://doi.org/10.1016/j.pnpbp.2013.09.017
Müller, N., Wagner, J. K., Krause, D., Weidinger, E., Wildenauer, A., Obermeier, M., … Schwarz, M. J. (2012). Impaired monocyte activation in schizophrenia. Psychiatry Research, 198(3), 341–346.
https://doi.org/10.1016/j.psychres.2011.12.049
R Core Team. (2017). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.r-project.org/
Rifai, N., & Ridker, M. P. (2003). Population Distributions of C-reactive Protein in Ap- parently Healthy Men and Women in the United States: Implication for Clinical Interpretation,. Clinical Chemistry, 49(4), 666–669. Roche, E., Creed, L., Macmahon, D., Brennan, D., & Clarke, M. (2015). The Epidemiology and Associated
Phenomenology of Formal Thought Disorder: A Systematic Review. Schizophrenia Bulletin, 41(4), 951– 962. https://doi.org/10.1093/schbul/sbu129
Schwarcz, R., Bruno, J. P., Muchowski, P. J., & Wu, H. (2012). Kynurenines in the mammalian brain : when physiology meets pathology. Nature Reviews Neuroscience, 13, 465–477.
https://doi.org/10.1038/nrn3257
Schwieler, L., Larsson, M. K., Skogh, E., Kegel, M. E., Orhan, F., Abdelmoaty, S., … Engberg, G. (2015). Increased levels of IL-6 in the cerebrospinal fluid of patients with chronic schizophrenia — significance for activation of the kynurenine pathway. Psychiatry Neuroscience, 40(2), 126–133.
https://doi.org/10.1503/jpn.140126
Shcherbakova, I., Neshkova, E., Dotsenko, V., Platonova, T., Shcherbakova, E., & Yarovaya, G. (1999). The possible role of plasma kallikrein – kinin system and leukocyte elastase in pathogenesis of schizophrenia.
Immunopharmacology, 43, 273–279.
Sidlova, M., Prasko, J., Jelenova, D., Kovacsova, A., Latalova, K., Sigmundova, Z., & Vrbova, K. (2011). The quality of life of patients suffering from schizophrenia - a comparison with healthy controls. Biomedical Papers,
155(2), 173–180. https://doi.org/10.5507/bp.2011.010
Smith, J. A., Das, A., Ray, S. K., & Banik, N. L. (2012). Role of pro-inflammatory cytokines released from microglia in neurodegenerative diseases. Brain Research Bulletin, 87(1), 10–20.
https://doi.org/10.1016/j.brainresbull.2011.10.004
Sproston, N. R., & Ashworth, J. J. (2018). Role of C-reactive protein at sites of inflammation and infection.
Frontiers in Immunology, 9(APR), 1–11. https://doi.org/10.3389/fimmu.2018.00754
Tourjman, V., Kouassi, É., Koué, M. È., Rocchetti, M., Fortin-Fournier, S., Fusar-Poli, P., & Potvin, S. (2013). Antipsychotics’ effects on blood levels of cytokines in schizophrenia: A meta-analysis. Schizophrenia
Research, 151(1–3), 43–47. https://doi.org/10.1016/j.schres.2013.10.011
UMCU, (2016, July 8). Raloxifene Augmentation in Patients with a Schizophrenia spectrum Disorder to reduce symptoms and improve cognition. EU Clinical Trials Register. Retrieved from
https://www.clinicaltrialsregister.eu/ctrsearch/search?query=raloxifene+augmentation+psychosis Wilcox, J., Winokur, G., & Tsuang, M. (2012). Predictive value of thought disorder in new-onset psychosis.