Evaluating the Exposome Score for Schizophrenia in a Transdiagnostic Psychosis Cohort: Associations With Psychosis Risk, Symptom Severity, and Personality Traits

Abstract

Background

Investigations of causal pathways for psychosis can be guided by the identification of environmental risk factors. A recently developed composite risk tool, the exposome score for schizophrenia (ES-SCZ), which controls for intercorrelations between risk factors, has shown fair to good performance. We tested the transdiagnostic psychosis classifier performance of the ES-SCZ with the Bipolar-Schizophrenia Network for Intermedial Phenotypes data and examined its relationship with clinical-level outcomes.

Study Design

We computed the case-control classifier performance for the ES-SCZ from cross-sectional data on 1055 volunteers with psychotic diagnoses (schizophrenia, schizoaffective, bipolar psychosis) and 510 controls. Multivariate regression models were used to control for the correlations between outcomes and to correct for the effects of age, sex, and family socioeconomic status across outcomes. We estimated association for the ES-SCZ with psychosis and mood symptom severity, the 5-factor model of personality, and function across biologically defined biotypes, traditional diagnostic categories, and controls.

Study Results

ES-SCZ classifier performance for psychosis was fair to good. ES-SCZ associations with personality factor scores were qualitatively similar between psychosis groups and controls with decreased conscientiousness and agreeableness and increased neuroticism. The patterns of associations between ES-SCZ and symptoms differed across biotypes and diagnoses. Biotype 3 and bipolar disorder had consistent within-group associations where greater exposome score predicted more severe symptoms and worse function.

Conclusions

ES-SCZ performance was consistent with previous reports in this transdiagnostic psychosis sample (adjusted odds ratio: 3.331 [2.834, 3.915], P < .001; area under the curve: 0.762 [0.735, 0.789]). Individual differences in ES-SCZ magnitude may be useful for investigating causal pathways between developmentally relevant exposures and symptomatic expression of psychosis.

Introduction

Identifying a set of well-defined and highly predictive environmental risk factors for psychosis is an active area of research^1,2 and represents a promising approach to investigating potential causal pathways between exposures and symptomatic expression.³ Many potential environmental risk factors for psychosis have been identified, though only a small subset have been found to have replicable and stable effects. Currently, there is no widely agreed-upon set of important environmental risk factors to include in risk models, but well-supported and commonly cited factors include: first-/second-generation migrant status,⁴ urbanicity,⁵ season of birth,⁶ childhood trauma and bullying,^7,8 cannabis use,⁹ tobacco use,¹⁰ obstetric and perinatal complications¹¹ and hearing impairment,¹² and young maternal and paternal age.¹³ Other commonly cited factors likely include a more significant portion of genetic liability, which limit their use in environmental risk factor models, including parental history of severe mental illness, geographical population groupings, premorbid IQ, advanced paternal age, parental socioeconomic status (SES), handedness, and personality dimensions.^14,15

Published umbrella studies of meta-analyses and systematic reviews have sought to identify the most robust risk factors as well as their relative effect sizes.^14,16,17 This work has supported the creation of increasingly accurate predictive models that incorporate the additive relative effects of risk factors.^15,18,19 Most recently, an “exposome” score for schizophrenia (ES-SCZ), which controls for the relationships between environmental risk factors themselves, has exhibited substantial improvements in predictive power over past additive effect-weighted models.^20,21 The ES-SCZ exhibits fair to good discriminative performance (ROC AUCs of 0.73-0.84 and 14%-21% of variance explained in schizophrenia case-control status) using only the 5 domains of the childhood trauma questionnaire (CTQ), bullying, cannabis use, winter birth, and hearing impairment.^20,22

The ES-SCZ model was created to capture schizophrenia risk but has also been found to predict bipolar disorder and suicidality to a similar degree as schizophrenia and predicts increased risk for mood disorders and personality factor scores to a significant but lesser extent in general populations.²³ Additionally, score magnitude has been associated with the degree of functional impairment and symptom severity in a more broadly defined psychosis sample.²⁴ Exposure to abuse and neglect in childhood, which is captured in the ES-SCZ, has been independently shown to be positively associated with psychotic symptoms²⁵ and psychological stress is broadly associated with structural and functional brain changes which alter inter- and intrapersonal regulatory capacity.³ The observation that the ES-SCZ captures associations beyond pure diagnostic risk positions this metric as a tool to explore environment-associated clinical-level constructs, such as personality structure or functional capacity, that may mediate the susceptibility to psychotic disorders. These nonspecific associations of ES-SCZ beyond schizophrenia alone, to include bipolar disorder and other mood disorders, have been acknowledged by the developers of the ES-SCZ.²⁰ And this transdiagnostic quality accords well with a dimensional approach to characterizing severe mental illness, supported by neurocognitive and neuroimaging biomarker findings.^26,27

The Bipolar-Schizophrenia Network for Intermediate Phenotypes (B-SNIP) consortium is one such research program that used a transdiagnostic approach and acquired a broad panel of neurocognitive and physiological biomarkers in a large sample of individuals with psychosis and nonpsychotic controls. Three psychosis phenotypes or Biotypes were identified independent of DSM nosology. While DSM psychosis subgroups represent clinically apparent phenotypes, Biotypes reflect data-driven clusters derived from biologically informed patterns across cognitive performance, psychophysiological responses, and EEG activity. The details of the study design, biotyping procedure, and general sample characteristics have been previously described.²⁶ Briefly, Biotype 1 is characterized by broadly reduced gray matter with extensive cognitive dysfunction, low EEG, and impairments in abstract thinking, Biotype 2 is characterized by decreased frontotemporal gray matter, elevated EEG, and extensive cognitive dysfunction with pronounced thought disorder, and Biotype 3 is characterized by mostly normal cognition and limited gray matter reduction of the frontal, cingulate, and temporal cortex with more mild psychotic symptoms.²⁷ The Biotype clusters were replicated in a large sample (B-SNIP2). That replication sample was also extensively characterized by clinically important information, including demographic and socioeconomic factors which coincidentally reflect the core elements of environmental risk captured in the ES-SCZ.²⁸

Methods

Design and Hypotheses

In this study, we examine the performance of the ES-SCZ in the B-SNIP2 cohort with respect to psychosis case-control risk and the psychotic subtypes as captured by diagnostic categories defined by the DSM and empirically derived Biotypes identified and replicated in the B-SNIP research program. Additionally, we planned to further explore the association between ES-SCZ with clinical-level metrics, including psychotic symptoms, mood symptoms, functional domains, and personality factors. With these goals, we hoped to leverage the large sample size and explicit transdiagnostic approach of the B-SNIP2 data to characterize associations with environmental risk that may differ across biotypes with the potential to reveal more proximate causal mechanisms. Our a priori hypotheses were (1) that the ES-SCZ score would show similar performance as a classifier across conventional psychotic diagnoses in this data set and (2) that there would be a linear positive association with neuroticism, psychotic symptoms, and mood symptoms across the whole sample. Regarding the Biotypes, we hypothesized that ES-SCZ would have the greatest association with symptoms and function for Biotype 3 (BT-3) given the limited cognitive and structure brain abnormalities noted above and previous suggestions the BT-3 psychotic symptoms may be driven by environmental and epigenetic factors.²⁷

Participants

The Bipolar-Schizophrenia Network for Intermediate Phenotypes consortium replication study (B-SNIP2) enrolled individuals with the DSM-defined diagnoses of schizophrenia, schizoaffective disorder, and bipolar disorder with psychotic features (Table 1). Clinical evaluations were completed using the Structured Clinical Interview for DSM-IV-TR.²⁹ Volunteers with psychosis were outpatients who were medicated and stable. Volunteers without a personal history of psychotic or mood disorders and without a first- or second-degree family history of schizophrenia or bipolar spectrum disorders served as controls. Both psychosis (n = 1055) and control (n = 510) groups completed a battery of laboratory-based tests that were used to cluster and replicate biotypes in the initial B-SNIP1 cohort as well as clinically relevant self-report rating scales as noted below.^26,28

Measures

Standard clinician-administered, self-report instruments were used to quantify historical and current environmental exposures, personality traits, global function, mood symptoms, and psychosis symptoms. Childhood exposure to adversity was quantified with the CTQ Short Form which assesses 5 domains of trauma: physical, emotional, and sexual abuse and physical and emotional neglect.³⁰ Cannabis use was quantified by self-report of ever/never and frequency of use before age 15, between 15 and 18, and over 18 years of age. SES was calculated separately for participant and family of origin using the Hollingshead model which combines level of education and professional status. The Positive and Negative Syndrome Scale (PANSS) subscales (Positive, Negative, and General Psychopathology) were used to assess psychosis symptoms in the volunteers.³¹ Depression severity was measured with The Montgomery-Åsberg Depression Rating Scale (MADRS). Mania symptoms were captured with the Young Mania Rating Scale (YMRS). Anxiety symptoms were measured with the Clinical Anxiety Scale (CAS). Personality dimensions were measured with a 120-item variant of the NEO-PI-R personality inventory structured around a 5-factor model.^32,33

Exposome Score for Schizophrenia

The ES-SCZ was calculated for each participant as the validated additive log-odds weighted model²² which includes binarized environmental exposures with the following weights: cannabis use (1.31), winter birth (December to March, 0.03), emotional abuse (0.78), physical abuse (−0.39), sexual abuse (0.86), emotional neglect (0.44), and physical neglect (0.25). Hearing impairment was assessed using audiometry and was an exclusionary criterion (0 for all). Bullying was not systematically captured in the data set; thus, its effect is not represented in this study. Binarization cutoffs of continuous measures for the ES-SCZ are: cannabis (once per week or greater) and CTQ subscale scores for emotional abuse (>8), physical abuse (>7), sexual abuse (>5), emotional neglect (>9), and physical neglect (>7).

Analysis

All statistical analyses were performed in R³⁴ except for receiver operating characteristic data and area under the curve (AUC) which were calculated using Matlab’s “perfcurve” function³⁵ with bootstrapped 95% confidence intervals using 10 000 bootstrap samples. Group differences for continuous variables were detected using t test and Kruskal-Wallis tests as appropriate for the data distribution. Differences in ES-SCZ distributions between diagnostic categories and biotypes were computed using ANCOVA with post hoc group comparison using Tukey’s honestly significant difference. Error around means is reported as upper and lower values of the 95% confidence intervals. Logistic regression was used to model the linear relationship between ES-SCZ and odds of a given psychosis label vs controls in the Wilkinson notation form of “group label ~ ES-SCZ.” A linear relationship between risk score and psychosis group was modeled as previous work demonstrates a dose-dependent response.²⁰ Odds ratios (ORs) are reported as “adjusted” with respect to age, sex, and family SES. Explained variance is reported as the McFadden pseudo-R-squared for logistic regression. The associations of ES-SCZ (independent variable) with psychosis symptoms, mood symptoms, personality traits, and functioning (dependent variables) were quantified with multivariate linear models to control for the covariance between dependent variables and better estimate the associations with ES-SCZ. Continuous variables were z-scored prior to model fitting and are thus reported as β weights to allow for clearer comparison across outcomes. Age, sex, and family SES were included as covariates in all linear models as these were found to alter the strength of the associations in sensitivity analyses. Given the a priori hypotheses regarding Biotypes, a nested hypothesis testing approach was used. The inflation of the false discovery rate due to multiple comparisons was managed using the Holm-Bonferroni method.

Results

ES-SCZ Case-Control Classifier Performance

The ES-SCZ scores were markedly different between control and psychosis groups (mean difference 0.940 [0.836, 1.044] points or 0.912 [0.811, 1.013] standard deviations, t₁₃₆₂ = 17.765, P < .001). The ES-SCZ yielded an adjusted OR (aOR) of 3.331 ([2.834, 3.915], P < .001) and predicted an estimated 16.3% of the variance in case-control status after correcting for age, sex, and family SES. Figure 1A presents mean differences by psychosis subgroup which were all significantly larger than controls (all P < .001). Case-control classifier performance revealed an AUC of 0.762 [0.735, 0.789] (Figure 1B).

ES-SCZ Values Across DSM and Biotype Psychosis Subgroups and Controls

An analysis of covariance revealed a small but significantly increased ES-SCZ score for schizoaffective disorder vs schizophrenia (t₉₆₉ = 2.430, P = .041, corrected for multigroup comparisons) but not between other diagnoses (Figure 1A). There were no differences in mean ES-SCZ scores across Biotypes (all P > .2). The raw ES-SCZ values ranged from −0.39 to 3.64 and subgroup means and standard deviations were as follows, controls: 0.61 (0.80), composite psychosis: 1.60 (0.97), Biotype 1: 1.67 (1.00), Biotype 2: 1.57 (0.96), Biotype 3: 1.55 (0.96), Schizophrenia: 1.49 (0.97), Schizoaffective: 1.72 (0.98), Bipolar with Psychosis: 1.59 (0.95).

ES-SCZ Associations With Personality Factor Scores Within DSM and Biotype Psychosis Subgroups and Controls

In controls, the ES-SCZ was positively associated with openness and neuroticism and negatively associated with agreeableness, extraversion, and conscientiousness, while in individuals with psychosis there was a similar though attenuated pattern except for extraversion (Table 2). The intercepts for the linear models were significantly different between psychosis and control groups which suggests the attenuated linear association may be the result of higher or lower means scores (see Supplementary Figure S1). Across Biotypes, ES-SCZ was negatively associated with conscientiousness and agreeableness, and positively associated with neuroticism. Only BT-1 was positively associated with openness, while BT-2 and BT-3 had no association. DSM diagnoses displayed greater heterogeneity with schizophrenia and schizoaffective disorder showing negative associations for conscientiousness and agreeableness, and a positive association with neuroticism, while bipolar disorder with psychosis was only associated with openness in the positive direction. A post hoc exploratory analysis of personality facets within the openness dimension that drove the positive association revealed that for both psychosis and control groups significant positive associations for facets 1 and 3 were present which correspond with imagination (“love to daydream”) and emotionality (“feel other’s emotions”).

ES-SCZ Associations With Clinical Psychosis, Mood, and Function Measures Within DSM and Biotype Psychosis Subgroups and Controls

Biotype 3 (BT-3) has been hypothesized to represent individuals with psychotic presentations that are more vulnerable to or exacerbated by environmental exposures. To test the hypothesis that within BT-3 the ES-SCZ is positively associated with symptom severity, a single multivariate multiple regression model was chosen over a series of bivariate correlations. Multivariate models control for the highly correlated nature of symptoms and function to better estimate the unique association between environmental risk and each outcome variable.³⁶ In this model, ES-SCZ was associated with multiple symptom domains for BT-3 in the predicted direction. In support of our a priori hypothesis, BT-3 shows a significant positive association with psychotic, depressive, and manic symptom severity, and negative associations with age at first break and global function after correction for multiple comparisons (PANSS pos β = 0.138 [0.026, 0.249], PANSS gen β = 0.175 [0.065, 0.285], MADRS β = 0.184 [0.065, 0.285], YMRS β = 0.155 [0.039, 0.272], First Break Age β = −0.200 [−0.307, −0.009], GAF β = −0.183 [−0.303, −0.060], participant SES β = −0.165 [−0.272, −0.050]). Follow-up testing of associations for BT-1 and BT-2 reveal largely nonsignificant associations even before correction for multiple comparisons. ES-SCZ associations with depressive and anxious symptoms are the most consistent across all 3 Biotypes. For BT-2, ES-SCZ shows an association with age at first break that is similar to BT-3. Table 3 reports the effect of ES-SCZ on symptom severity for groups; notably, only BT-3 shows a consistent pattern of effects in the expected direction among Biotypes (Figure 2). Among DSM diagnoses, the patterns of ES-SCZ associations are more diffuse, but the Bipolar group demonstrates a pattern that is the most similar to BT-3.

Discussion

In this study, we investigated the discriminative performance of the ES-SCZ and its associations with clinical-level factors and outcomes in the B-SNIP2 sample—a well-characterized cohort with an explicitly transdiagnostic characterization of psychosis. We build on previously published exposome analyses with our novel use of a robust multivariate approach to control for the highly intercorrelated nature of symptoms and function to better isolate the effect of ES-SCZ on outcomes.³⁶ We found that ES-SCZ demonstrated discriminative performance for psychosis similar to previous reports with an aOR of 3.33 and an ROC AUC of 0.762.^20,22 Mean ES-SCZ was not different across B-SNIP Biotypes but the schizoaffective group showed a small but significant increase in exposome risk as compared with the schizophrenia group. We then tested an a priori hypothesis that phenotypic expression for BT-3 is uniquely associated with exposome risk as captured by the ES-SCZ and indeed found a consistent pattern of associations with greater symptom severity and worse function for BT-3, but not for BT-1 or BT-2. Finally, we explored ES-SCZ associations with personality dimensions and functional outcomes including controls and DSM diagnoses for comparison which revealed a complex picture but with a trend for lower personality factor scores across all groups, which was similar in controls. This study adds to the existing exposome literature by using a unified multivariate model to cross-validate and expand upon previously identified associations with ES-SCZ using the standardized clinical measures of psychosis symptoms, mood symptoms, function, and personality factors available from the B-SNIP2 study to better isolate the unique associations between ES-SCZ and clinically relevant outcomes. These findings expand the characterization of the biologically derived B-SNIP psychosis Biotypes by revealing an aggregation of exposome-associated impairments in symptomatic control and clinically relevant outcomes within Biotype 3, suggesting that this subtype is well-suited to explore mechanisms linking environmental risk and psychosis.

Exposome and Symptom Severity

The ES-SCZ was introduced in the literature as a measure of exposome-associated risk for schizophrenia and developed using case-control performance metrics; however, it has also been observed to be a predictor of nonpsychotic diagnoses and clinical symptoms. The limited phenotypic specificity observed in this study was anticipated given the previously reported broader associations and supports the hypothesis that the ES-SCZ captures an environmental diathesis associated with developmentally elevant exposures.²² Previous work exploring the ES-SCZ in a general population sample, the Netherlands Mental Health Survey and Incidence Study-2 (NEMESIS-2), identified exposome associations with psychosis risk, schizophrenia spectrum disorder, bipolar disorder, social phobia, major depressive disorder, generalized anxiety disorder, and personality traits, modeled as separate bivariate correlations with ES-SCZ.²³ Given the high among psychiatric symptoms, bivariate associations alone are likely to contain nonunique covariance which distorts the relationship between exposome score and symptom severity. To address this, we employed multivariate linear models which control for the covariance between outcome variables to better estimate each association with ES-SCZ. We used this to first test our a priori hypothesis for the BT-3 group which revealed a pattern of associations which was consistent with the hypothesis that the BT-3 cluster represents individuals with phenotypic expressions that are sensitive to environmental risk. Follow-up testing of BT-1 and BT-2 revealed a weaker and less consistent pattern, suggesting that BT-3 illness symptoms are more sensitive to environmental exposures as captured by the ES-SCZ.

The bipolar psychosis group also demonstrated more consistent associations between symptom severity and ES-SCZ, than the schizophrenia or schizoaffective groups. BT-3 cases are about 50% of the bipolar group (vs ~25% for BT-1 and BT-2) which provides some link between the 2 groups as well. Childhood trauma and other psychosocial stressors are known to predict greater symptoms and dysfunction in bipolar spectrum populations,³⁷ though the consistency of the pattern observed in this study and its direct comparison with schizophrenia and schizoaffective is a novel finding to our knowledge. One possible explanation for these shared associations with ES-SCZ and symptoms for BT-3 and bipolar groups is that both have lower mean symptom scores and higher mean function relative to their comparison groups. While not a clear example of regression to the mean, ceiling or floor effects may have made those measures less sensitive to detect an association with ES-SCZ in the other groups. Given that symptom severity scores were measured on individuals receiving treatment for their psychiatric disorders, they also reflect the degree to which symptoms have responded to medication. In this framing, symptoms in BT-3 and bipolar disorder with psychosis respond less well to treatment as ES-SCZ increases in magnitude which suggests that a potential mechanistic link between ES-SCZ and psychosis may involve processes that do not respond as well to standard psychopharmacological approaches (antipsychotics and mood stabilizers).

Personality and Illness Susceptibility

In this transdiagnostic sample, the ES-SCZ was associated with increased neuroticism and decreased conscientiousness and agreeableness for the composite psychosis group (Supplementary Table S1) and the control group. The ES-SCZ-associated personality differences in the control group result in a personality profile that is like that observed in psychotic illness.^32,38 Within both DSM and Biotype psychosis subgroups there was heterogeneity in the pattern of ES-SCZ associations with individual factors scores, though a trend of decreased agreeableness and increased neuroticism was present across subgroups. To our knowledge, this is the first direct comparison of personality differences associated with ES-SCZ magnitude across a transdiagnostic sample. We did not have a priori hypotheses across psychosis subtypes. As an exploratory component of the analysis without prior effect sizes for contextualization, our study may have been underpowered to detect some differences. The differences in patterns of associations are a potential target for further investigation to explore phenotypic convergence and divergence across subgroups.

The dynamics of person-environment interactions cannot be disentangled in this cross-sectional design, personality structure is thought to reflect a relatively stable psychological continuum which is sensitive to environmental exposures and influences susceptibility to the expression of mental illness phenotypes.³⁹ In accordance with a diathesis-stress framework of mental illness, exposome-associated changes in personality structure (increased openness and neuroticism, decreased conscientiousness, extraversion, and agreeableness) may be interpreted as one domain of increased susceptibility to stress-induced emergence of clinical-level affective and psychotic symptoms during later periods of stress.^40,41 While personality structure is considered relatively stable, there is significant developmental variability within individual.⁴² As such, exposome-associated personality changes may be a target for early intervention.⁴³ If the ES-SCZ indeed captures causal factors that exacerbate a psychological diathesis, then a similar pattern of associations with personality differences across psychosis and control groups represents a potential causal link between environment and susceptibility to stress-induced mental illness. Further work is needed to characterize the causal pathways between exposome risk, developmental changes in brain and behavior, and susceptibility to mental illness.

Limitations and Future Directions

A major strength of this study is its incorporation of a previously validated measures of exposome risk with a large and well-characterized data set in combination with the use of multivariate models to control for the highly correlated nature of symptoms, function, and personality. The analysis was centered on specific a priori hypotheses for BT-3 and ES-SCZ-associated clinical outcomes to limit the risk of false discovery. Post hoc exploratory analyses across groups must be interpreted more cautiously and represent a limitation of this approach.

The case-control classifier performance in this study was comparable with previously published reports, though the B-SNIP2 cohort is a significantly different population than the model was trained on. It is based in the United States and contains a lower proportion of White-identifying individuals in comparison to the European cohorts used for initial ES-SCZ model estimation and testing. The consistent performance of the ES-SCZ in this sample suggests that the model was not over-fit to the original population and captures more general exposome risk associations. It is also important to note that the B-SNIP2 cohort excluded hearing impairment and did not systematically capture bullying. As these are both highly weighted factors in the ES-SCZ, this represents a limitation of this study. Additionally, the 9 exposome factors used in the development of the ES-SCZ were chosen for their convenience in a previously collected dataset.²⁰ The overall performance of the tool may be improved through future work which includes a broader set of environmental risk factors in a data set large enough to train an expanded model.

Another limitation of this study is the observational and cross-sectional nature of the data as past symptoms, substance use, and childhood trauma are derived from self-report measures. Therefore, strong causal inference between exposures and outcomes would be ill-advised. While causality and dynamic temporal factors cannot be determined with this analysis of previously collected cross-sectional data, the ES-SCZ-associated differences within and between group offer a starting point for more rigorous investigations of intermediate constructs, such as personality structure, that may be part of a modifiable causal chain between environment and psychosis. While personality dimensions demonstrate stability over time, they are not immutable and may represent a modifiable diathesis target. Future work is needed to explore the potential mediating role of exposome-associated personality structures with symptomatic expression and the role that individual differences in brain organization and genetic factors may play in these associations within the Biotype 3 cluster and bipolar psychosis.

In conclusion, our findings from the application of the ES-SCZ to the B-SNIP2 study data further support its use as a research tool to study environmentally associated psychosis risk as well as individual differences in symptomatic expression, function, and personality structure in those with psychotic illness as well as in nonpsychotic populations. The Biotype 3 cluster offers a biologically informed psychosis phenotype that may be useful for exploring the causal links between exposome and psychotic illness.

Funding

This research was collectively supported by numerous NIH grants: B.K.: MH101078. C.A.T.: MH077851; MH096913; MH127179; MH124813. E.G.: MH124804; MH127162; MH103368. B.A.C.: MH124803; MH126398; MH096900; MH124806; MH103366; MH124802; MH127172. G.P.: MH127158; MH124802; MH077945; MH096957. M.K.: MH124807; MH127174; MH078113; MH096942. D.P.: MH117315; UL1TR002378; TL1TR002382.

Conflicts of Interest

Conflict of interest disclosures: B.K.: None. W.Y.: None. M.K.: B-SNIP Diagnostics, Board of Managers. D.P.: None. V.J.T.: None. G.P.: B-SNIP Diagnostics, Board of Managers. S.K.: B-SNIP Diagnostics, Board of Managers. J.M.: B-SNIP Diagnostics, Board of Managers. E.G.: B-SNIP Diagnostics, Board of Managers; Consultant: Kynexis Corporation. E.I.: B-SNIP Diagnostics, Board of Managers; Consultant: Janssen Pharmaceuticals; Consultant: Alkermes. S.K.H.: None. B.A.C.: B-SNIP Diagnostics, Board of Managers; Kynexis Corporation, Scientific Advisory Board. C.A.T.: B-SNIP Diagnostics, Board of Managers; Kynexis Corporation, Scientific Advisory Board and receives a retainer; Bristol Myers Squibb, Scientific Advisory Board.

References