Significant outcomes
∙ White blood cell (WBC) deviations were associated with greater Montgomery and Aasberg depression rating scale (MADRS) severity.
∙ Higher and lower WBC levels were associated with specific symptoms.
∙ The findings were most pronounced among men.
Limitations
∙ WBC is a very unspecific immune system marker and has a large intra-individual variation.
∙ We did not measure subtypes of WBCs or other inflammatory markers.
∙ The cross-sectional and explorative design emphasise cautious interpretations.
Introduction
An increased immune system activity has been associated with the aetiology of bipolar disorder (Reference Drexhage, Knijff and Padmos1–Reference Dargel, Godin, Kapczinski, Kupfer and Leboyer3) and higher levels of pro-inflammatory markers may be associated with increased symptom severity and specific symptom domains in bipolar disorder (Reference Hope, Dieset and Agartz4–Reference Hope, Hoseth and Dieset7). The sickness syndrome theory may represent one explanation for these associations, suggesting that pro-inflammatory cytokines affect the development and severity of specific symptoms, such as decreased appetite, increased need for sleep, social withdrawal, and anhedonia (Reference Raison and Miller8,Reference Saper, Romanovsky and Scammell9).
However, despite increasing pre-clinical evidence (Reference Elenkov10), this aspect has only been investigated in a few clinical trials with small study populations (Reference Hope, Dieset and Agartz4,Reference Lotrich, Butters, Aizenstein, Marron, Reynolds and Gildengers5,Reference Krogh, Benros, Jorgensen, Vesterager, Elfving and Nordentoft11,Reference Jokela, Virtanen, Batty and Kivimaki12). In addition, the studies did not investigate specific symptoms (Reference Hope, Dieset and Agartz4,Reference Lotrich, Butters, Aizenstein, Marron, Reynolds and Gildengers5,Reference Kohler, Sylvia and Bowden13) and factors such as body mass index (BMI) may confound these findings (Reference Krogh, Benros, Jorgensen, Vesterager, Elfving and Nordentoft11). Most studies did not include BMI (Reference Hope, Dieset and Agartz4,Reference Lotrich, Butters, Aizenstein, Marron, Reynolds and Gildengers5,Reference Jokela, Virtanen, Batty and Kivimaki12), and in a recent study among depressed individuals, adjustment for BMI neutralised the unadjusted significant associations between pro-inflammatory cytokines and increased symptom severity (Reference Krogh, Benros, Jorgensen, Vesterager, Elfving and Nordentoft11). Also, immuno-suppression may affect psychopathology (Reference Walsh, Watson and Kipnis14,Reference Brod, Rattazzi, Piras and D’Acquisto15), but the trials have mostly investigated pro-inflammatory markers (Reference Hope, Dieset and Agartz4,5,Reference Krogh, Benros, Jorgensen, Vesterager, Elfving and Nordentoft11,Reference Jokela, Virtanen, Batty and Kivimaki12). Finally, important effects of sex hormones have been found during immune responses (Reference Angele, Pratschke, Hubbard and Chaudry16), why possible gender differences need to be investigated.
We aimed to perform an explorative study based on two large similar clinical trials, which included WBC levels. WBC is a marker of overall immune system activity, with high levels indicating an inflammatory response (i.e. leucocytosis) and low levels indicating insufficient production of immune cells (i.e. leucopenia). We hypothesised that WBC levels are associated with specific symptoms, possibly those suggested to be affected by the sickness syndrome (Reference Raison and Miller8,Reference Saper, Romanovsky and Scammell9). We a priori hypothesised gender differences. We thereby expand our previous work associating WBC levels with overall increased symptom severity (Reference Kohler, Sylvia and Bowden13).
Materials and methods
Setting
This study represents secondary analyses based on data from two similar multi-site, randomised comparative effectiveness trials based in the United States, the lithium treatment moderate-dose use study (LiTMUS) (Reference Nierenberg, Sylvia and Leon17) and the clinical and health outcomes initiatives in comparative effectiveness for bipolar disorder study (Bipolar CHOICE) (Reference Nierenberg, Sylvia and Leon18). LiTMUS compared lithium treatment combined with optimised personalised treatment (OPT) to OPT alone. Bipolar CHOICE compared lithium with quetiapine. The rationale, design, and specific methods are reported in detail elsewhere (Reference Nierenberg, Sylvia and Leon17,Reference Nierenberg, Sylvia and Leon18). The present study examined data at study entry (i.e. at baseline) and participants could not be treated with lithium or quetiapine at baseline (>30 days without treatment), as, for example, lithium affects WBC levels (Reference Amitai, Zivony and Kronenberg19).
Participants
For LiTMUS, 338 patients were screened and 283 (84%) were randomised. For Bipolar CHOICE, 692 were screened and 482 (70%) were randomised. In both studies, participants were aged between 18 and 62 years, and limited inclusion and exclusion criteria were defined to maximise heterogeneity of the sample and thus, generalisability of the results. The main inclusion criteria were a DSM-IV-TR bipolar I or II diagnosis and a clinical global impression scale for bipolar disorder ≥3 (Reference Spearing, Post, Leverich, Brandt and Nolen20) (i.e. mildly symptomatic).
In both studies, the extended Mini-International Neuropsychiatric Interview (Reference Sheehan, Lecrubier and Sheehan21) was used to determine psychiatric and substance use diagnoses. Mood symptom severity was measured with the MADRS in LiTMUS (Reference Montgomery and Asberg22) and with the bipolar inventory of symptoms scale (BISS) in Bipolar CHOICE (Reference Bowden, Singh and Thompson23–Reference Thompson, Gonzalez, Singh, Schoolfield, Katz and Bowden25). In order to pool data across studies, we extracted the MADRS score from the BISS items within the Bipolar CHOICE study, which has been validated (Reference Gonzalez, Bowden and Katz24). Clinical interviews obtained demographic information, mental and medical history (hypertension, diabetes, and hyperlipidemia), and current medications. In both studies, baseline WBC count was assessed in each patient at study entry via a fasting blood draw from an antecubital vein and analysed immediately at the local biochemical laboratory at the hospital where the patient participated (no specific time of day, WBC is expressed in International units, i.e. ×109/l).
Statistical analysis
We identified baseline variables possibly influencing WBC levels and symptom severity: gender, age, BMI, race, current smoking, and the medical conditions of diabetes, hypertension and hyperlipidemia. We performed all analyses on men and women separately.
In both studies, mood symptom severity was greater among patients with the highest and lowest WBC counts. Therefore, as a first step, we investigated the association between the deviation from the median WBC and symptom severity. This means that we compared participants with higher or lower WBC counts at baseline (i.e. values higher or lower than the median WBC of all participants are grouped together and expressed as a positive value) to the median baseline WBC of all participants. We grouped participants based on their deviation from the median WBC of 6.9×109/l into the following categories: <1×109/l, 1–1.99×109/l, 2–2.99×109/l, and ≥3×109/l. Similar cut-offs have been used in previous studies (Reference Kannel, Anderson and Wilson26). As the deviation from the median WBC and MADRS symptom severity fitted a linear relationship in both studies, we performed analysis of variance (ANOVA) and linear regression analysis. We set those individuals who deviated <1×109/l as the reference group and report p-values. If an overall ANOVA was significant, we performed pairwise comparison between the different groups by applying Bonferroni, Scheffe, and Sidak multiple comparison tests. For the linear regression analysis, we report a coefficient including 95% confidence intervals (95% CI). The coefficient corresponds to the change in symptom severity for each 1.0×109/l deviation from the median WBC. We were aware that this approach grouped individuals with high or low WBC levels, but we were interested whether any deviation from the median WBC would affect psychopathology.
Second, we specifically investigated if higher or lower WBC levels were associated with symptom severity. We divided participants into the following baseline WBC categories: <4.5×109/l, 4.5–5.99×109/l, 6–6.99×109/l, 7–7.99×109/l, 8–9.99×109/l, ≥10×109/l. We set individuals with a WBC of 7–7.99×109/l as the reference group and performed multinomial logistic regression analysis reporting relative risk ratio including 95% CI.
For all the above-mentioned analyses, we first performed basic models adjusted for age. Second, we adjusted all analyses for age, BMI, race, current smoking, and a diagnosis of diabetes, hypertension, or hyperlipidemia.
As a primary analysis, we performed all the above-mentioned analyses (i.e. ANOVA, linear regression and multinomial logistic regression analysis) on the 10 individual item scores on the MADRS among all individuals from both trials (n=765). As a second step, we also performed multinomial logistic regression analysis on different WBC levels and specific symptoms on the BISS scale (i.e. the remaining 34 items not included in the MADRS scale) for the participants (n=482) in the Bipolar CHOICE study. For all analyses on specific symptoms, we divided individuals depending on the severity of a single specific symptom (1–2, 3–4, or 5–6) and compared with individuals without this symptom (i.e. a score of 0).
In all the above-mentioned analyses, we corrected for multiple testing by dividing the p-value of 0.05 with the amount of tests performed.
All statistical analyses were performed using STATA version 14.
Sensitivity analyses
Carbamazepine and clozapine can lower WBC levels (Reference Lally and MacCabe27). Therefore, we performed all the above-mentioned analyses where we excluded individuals who used these compounds at baseline to explore whether this had an impact on our findings.
Results
The median WBC across both studies (n=765) was 6.9×109/l and ranged from 3×109/l to 18×109/l, which is similar to the general US adult population with a mean of 6.9×109/l (range 1.5–100×109/l) (Reference Liu and Taioli28). The baseline characteristics based on WBC count are shown in Table 1, and Supplementary Material Table 1 shows the baseline characteristics in both studies divided by gender. Individuals with low WBC levels had lower BMI and were less often smokers. Individuals with high WBC levels had higher BMI, were more often smokers, and were younger.
Baseline information on 765 patients with bipolar disorder from the lithium treatment moderate-dose use study (n=283) and the clinical and health outcomes initiatives in comparative effectiveness for bipolar disorder study (n=482), depending on white blood cell counts (WBC) at study entry
BMI, body mass index.
WBC deviations and specific MADRS symptoms
Higher and lower WBC levels were associated with higher MADRS total scores, which was more present among men despite no significant differences between men and women at any point (Fig. 1).
White blood cell (WBC) count and Montgomery and Aasberg depression rating scale (MADRS) severity at baseline among 765 individuals with bipolar disorder. There were no significant differences between men and women for any WBC levels.
Tables 2 and 3 show the gender-separate association between the deviation from the median WBC and the 10 specific MADRS items. Among 322 men, WBC deviations were associated with increased severity on the following specific MADRS symptoms: reported sadness, inner tension, reduced appetite, inability to feel, and suicidal thoughts (all p<0.05, but only inner tension and inability to feel showed a corrected p-value <0.005). Among 443 women, WBC deviations showed a trend towards increased severity on apparent sadness (p=0.08), reduced appetite (p=0.06), and lassitude (p=0.07).
Analysis of variance* showing the correlation at study entry between the 10 specific Montgomery and Aasberg depression rating scale items and the deviation from the median white blood cell (WBC) count of 6.9×109/l among 322 men and 443 women with bipolar disorder from the lithium treatment moderate-dose use study and clinical and health outcomes initiatives in comparative effectiveness for bipolar disorder study
* All analyses were adjusted for age, body mass index, current smoking, race, and a diagnosis of diabetes, hypertension or hyperlipidemia.
† The analyses for men and women were corrected for multiple testing by dividing the p-value of 0.05 with 10 (the amount of tests performed among men and women, respectively), that is, the corrected p-value is 0.005.
Linear regression analyses showing the association between the deviation from the median white blood cell (WBC) count and the 10 specific Montgomery and Aasberg depression rating scale (MADRS) items among 322 men and 443 women with bipolar disorder from the lithium treatment moderate-dose use study and clinical and health outcomes initiatives in comparative effectiveness for bipolar disorder study
The median WBC in this population was 6.9×109/l.
* Model 1 was adjusted for age; Model 2 was adjusted for age, body mass index, current smoking, race, and a diagnosis of diabetes, hypertension, or hyperlipidemia.
† The coefficient indicates the change in MADRS score for every deviation in white blood cell count, for example, a coefficient of 1.10 indicates an increase in MADRS of 1.10 for each 1.0×109/l deviation (both higher and lower) from the median WBC of 6.9×109/l.
‡ The analyses for men and women were corrected for multiple testing by dividing the p-value of 0.05 with 10 (the amount of tests performed among men and women, respectively), that is, the corrected p-value is 0.005.
Higher and lower WBC levels are associated with specific MADRS symptoms
Among men, higher WBC levels (compared with men with WBC 7–7.99×109/l) showed a trend towards higher scores on reported sadness, apparent sadness, inner tension, reduced appetite, concentration difficulties, inability to feel, pessimistic thoughts, and suicidal thoughts (most p-values <0.2 with several <0.05) (Supplementary Material Table 2). Lower WBC levels showed a trend towards higher scores on reported sadness, inner tension, reduced sleep, reduced appetite, concentration difficulties, and suicidal thoughts (most p-values <0.3 with several <0.05).
Among women, higher WBC levels (compared with women with WBC 7–7.99×109/l) showed a trend towards increased severity of apparent sadness, reduced appetite, concentration difficulties, lassitude, inability to feel, and pessimistic thoughts (most p-values <0.3 with several <0.05) (Supplementary Material Table 3). Lower WBC levels showed a trend towards increased severity of reported and apparent sadness, reduced sleep, reduced appetite, concentration difficulties, lassitude, inability to feel, and pessimistic thoughts (most p-values <0.2 with several <0.1). Among men and women, lower WBC levels showed a trend towards increased severity of reduced sleep, and higher WBC levels showed a trend towards lower severity for reduced sleep (most p-values <0.2 with several <0.05). The latter was most pronounced among men.
However, the majority of findings with p<0.05 became non-significant after multiple testing.
Bipolar CHOICE: WBC levels and specific BISS symptoms
Finally, we assessed the remaining 34 BISS items among 199 men (Supplementary Material Table 4) and 283 women (Supplementary Material Table 5) from the Bipolar CHOICE study. Higher WBC levels were associated with an increment of severity on several depressive symptoms, in particular social withdrawal and anxiety symptoms among men and women. Also lower WBC levels were associated with increased severity of several depressive symptoms among men and women. Higher and lower WBC levels showed a trend towards increased severity of several manic and psychotic symptoms, which was most pronounced among women. However, due to small groups because of the detailed division, several multinomial regression analyses did not reach significance but only showed a trend. Furthermore, the majority of findings with p<0.05 became non-significant after multiple testing.
Sensitivity analyses
At study entry, three individuals used clozapine and 34 used carbamazepine. Exclusion of these users did not change the results (not shown).
Discussion
The present study found that higher WBC deviations from the median WBC were associated with greater severity of several specific bipolar disorder symptoms. Both lower and higher WBC levels were associated with an increased severity of several symptoms. In particular, the following MADRS items were affected: sadness, inner tension, reduced appetite, concentration difficulties, inability to feel, and suicidal thoughts. All the above-mentioned findings were most pronounced among men. We thereby expand our previous findings associating WBC deviations with increased overall mood symptom severity (Reference Kohler, Sylvia and Bowden13), and the present results give the most detailed overview to date between very specific bipolar disorder psychopathology and alterations in the immune system. As our analyses were exploratory, the design was cross-sectional, and the majority of tests did not survive multiple testing, future studies need to confirm these findings. Large, high-quality clinical trials should include a combination of markers on the overall immune system activity and a wide range of specific pro- and anti-inflammatory markers, preferably including sex hormone levels, and investigate associations with specific psychopathology over time.
Immune system alterations and symptom-specific severity in bipolar disorder
Our findings support the sickness syndrome theory, suggesting that peripheral cytokines and prostaglandins, produced during an inflammatory response, can cross the blood-brain barrier and thus, affect the CNS (Reference Raison and Miller8–Reference Elenkov10). Although we only had WBC counts, increased WBC levels (i.e. >10×109) indicate leucocytosis and hence an acute inflammation. Furthermore, our results indicate that even low WBC levels affect psychopathology, representing a less expected finding. This may be explained by potential neuroprotective properties of immune cells, such as CD4+ T-cells (Reference Brod, Rattazzi, Piras and D’Acquisto15), suggesting that also immune suppression may affect mood symptom severity.
In addition, suicidal thoughts and risky behaviour showed a trend towards increased severity among men with higher or lower WBC levels. Previous studies have indicated that CNS affection during inflammatory responses may lead to self-harming and even suicidal behaviour (Reference Pandey, Rizavi and Ren29). Finally, the fact that most results were most pronounced among men may be explained by the differential effects of sex hormones during immune responses, as female sex hormones may exhibit protective effects, whereas male sex hormones can act suppressive on cell-mediated immune responses (Reference Angele, Pratschke, Hubbard and Chaudry16).
WBC as a potential biomarker for more personalised treatment
In depression, studies have emphasised use of specific pro-inflammatory markers to identify sub-groups of patients that may respond better to antidepressants (Reference Uher, Tansey and Dew30) and anti-inflammatory treatment (Reference Raison, Rutherford and Woolwine31,Reference Kohler, Benros and Nordentoft32), which has been less studied in bipolar disorder (Reference Li, Hong and Zhang33–Reference Arabzadeh, Ameli and Zeinoddini35). Our results suggest WBC as a valid and inexpensive biomarker, which is easy to apply and thus, relevant for everyday clinical use. Future studies should investigate whether increased and/or decreased WBC levels are associated with improved response to mood-stabilising treatment and/or anti-inflammatory drugs. These studies may explore gender differences and the combination of specific symptoms with several immune system markers.
Strengths and limitations
The LiTMUS and Bipolar CHOICE studies were very similar trials with broad inclusion and minimal exclusion criteria, thus representing populations seen in everyday clinical practice. Both studies included large sample sizes, and no patients were treated with lithium or quetiapine at the time of assessment. Furthermore, WBC is an easily available and inexpensive measure from peripheral blood, supporting the clinical relevance of our findings. In addition, we were able to adjust for important factors (i.e. BMI, smoking, race, and specific somatic diseases).
In regards to limitations, these explorative analyses were cross-sectional, performed post-hoc and most findings with p<0.05 did not survive multiple testing. Hence, our findings have to be interpreted with caution. In addition, we did not include measures on specific subtypes of immune cells or other markers of immune activation (e.g. C-reactive protein or interleukin-6) and had only one WBC measure, with WBC being a rather non-specific marker of the immune system. In cardiovascular diseases, specific WBC subtypes seem to be associated with the risk increase associated with WBC deviations (Reference Horne, Anderson and John36). Thus, it seems necessary to evaluate specific WBC lines and assess a wider range of both pro-inflammatory and anti-inflammatory markers, and several assessments over time. Furthermore, important effects of sex hormones have been found during immune responses (Reference Angele, Pratschke, Hubbard and Chaudry16), and future studies should include this aspect. In addition, although we adjusted for important somatic diseases, we had no knowledge whether participants suffered of an acute infection or an autoimmune illness, that is, the observed WBC counts may be due to other illnesses. Moreover, we did not adjust for use of other medications. As the Bipolar CHOICE and LiTMUS studies were intended to represent real-world settings, somatic diseases only resulted in exclusion if the somatic disease contraindicated treatment with the study medications. Finally, some analyses on specific symptoms were underpowered, for example, on some psychotic or manic symptoms. This was further limited as we investigated on outpatients with bipolar disorder, with most patients (>60%) being in depressed phases at study entry.
Disclosures
Dr. Nierenberg is a consultant for the Abbott Laboratories, American Psychiatric Association, Appliance Computing Inc. (Mindsite), Basliea, Brain Cells, Inc., Brandeis University, Bristol-Myers Squibb, Clintara, Corcept, Dey Pharmaceuticals, Dainippon Sumitomo (now Sunovion), Eli Lilly and Company, EpiQ, L.P./Mylan Inc., Forest, Genaissance, Genentech, GlaxoSmithKline, Hoffman LaRoche, Infomedic, Lundbeck, Janssen Pharmaceutica, Jazz Pharmaceuticals, Medavante, Merck, Methylation Sciences, Naurex, Novartis, PamLabs, Parexel, Pfizer, PGx Health, Ridge Diagnostics Shire, Schering-Plough, Somerset, Sunovion, Takeda Pharmaceuticals, Targacept, and Teva; consulted through the MGH Clinical Trials Network and Institute (CTNI) for AstraZeneca, Brain Cells, Inc, Dianippon Sumitomo/Sepracor, Johnson and Johnson, Labopharm, Merck, Methylation Science, Novartis, PGx Health, Shire, Schering-Plough, Targacept and Takeda/Lundbeck Pharmaceuticals. He receives grant/research support from American Foundation for Suicide Prevention, AHRQ, Brain and Behavior Research Foundation, Bristol-Myers Squibb, Cederroth, Cephalon, Cyberonics, Elan, Eli Lilly, Forest, GlaxoSmithKline, Janssen Pharmaceutica, Lichtwer Pharma, Marriott Foundation, Mylan, NIMH, PamLabs, PCORI, Pfizer Pharmaceuticals, Shire, Stanley Foundation, Takeda, and Wyeth-Ayerst. Honoraria include Belvoir Publishing, University of Texas Southwestern Dallas, Brandeis University, Bristol-Myers Squibb, Hillside Hospital, American Drug Utilization Review, American Society for Clinical Psychopharmacology, Baystate Medical Center, Columbia University, CRICO, Dartmouth Medical School, Health New England, Harold Grinspoon Charitable Foundation, IMEDEX, Israel Society for Biological Psychiatry, Johns Hopkins University, MJ Consulting, New York State, Medscape, MBL Publishing, MGH Psychiatry Academy, National Association of Continuing Education, Physicians Postgraduate Press, SUNY Buffalo, University of Wisconsin, University of Pisa, University of Michigan, University of Miami, University of Wisconsin at Madison, APSARD, ISBD, SciMed, Slack Publishing and Wolters Klower Publishing ASCP, NCDEU, Rush Medical College, Yale University School of Medicine, NNDC, Nova Southeastern University, NAMI, Institute of Medicine, CME Institute, ISCTM. He was currently or formerly on the advisory boards of Appliance Computing, Inc., Brain Cells, Inc., Eli Lilly and Company, Genentech, Johnson and Johnson, Takeda/Lundbeck, Targacept, and InfoMedic. He owns stock options in Appliance Computing, Inc., Brain Cells, Inc, and Medavante; has copyrights to the clinical positive affect scale and the MGH Structured Clinical Interview for the MADRS exclusively licensed to the MGH Clinical Trials Network and Institute (CTNI).
Dr. Tohen was a full time employee at Lilly (1997–2008). He has received honoraria from, or consulted for, Abbott, Actavis, AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline, Lilly, Johnson & Johnson, Otsuka, Merck, Sunovion, Forest, Gedeon Richter, Roche, Elan, Alkermes, Allergan, Lundbeck, Teva, Pamlab, Wyeth and Wiley Publishing. His spouse was a full time employee at Lilly (1998–2013).
During the last 3 years, Dr. Terence Ketter has had financial interests/arrangements or affiliations with organisations that could be perceived as real or apparent conflicts of interest. Dr. Ketter has received Grant/Research Support (through Stanford University) from the Agency for Healthcare Research and Quality, AstraZeneca Pharmaceuticals LP, Cephalon Inc. (now Teva Pharmaceuticals), Eli Lilly and Company, Pfizer, Inc., and Sunovion Pharmaceuticals; has served as a Consultant/Advisory Board Member for Allergan, Inc., Avanir Pharmaceuticals, Depotmed, Forest Pharmaceuticals, Genentech, Janssen Pharmaceuticals, Merck & Co., Inc., Myriad Genetic Laboratories, Inc., ProPhase, Sunovion Pharmaceuticals, and Teva Pharmaceuticals; has received Lecture Honoraria (NOT Speaker’s Bureau payments) from Abbott Laboratories, Inc, GlaxoSmithKline, Otsuka Pharmaceuticals, Pfizer, Inc., and Sunovion Pharmaceuticals; and has received Royalties from American Psychiatric Publishing, Inc. In addition Dr. Ketter’s spouse is an employee of and stockholder of Janssen Pharmaceuticals.
Acknowledgements
Authors’ Contribution: O.K.F., A.A.N., and L.G.S. planned this study and prepared the analysis plan. O.K.F. and A.A.N. had access to all data and performed the analyses. All authors contributed to the interpretation of the results. O.K.F., A.A.N., and L.G.S. wrote the first draft of the manuscript which was critically revised by all authors, and all authors have approved the final manuscript.
Financial Support
Bipolar CHOICE was funded by the Agency for Healthcare Research and Quality (AHRQ), 1R01HS019371-01, and LiTMUS was funded by NIMH contract NO1MH80001. The funders had no role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.
Conflicts of Interest
The other authors declare no competing interests.
Ethical Standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Supplementary materials
To view supplementary material for this article, please visit https://doi.org/10.1017/neu.2017.34
