Neuropattern: A new translational tool to detect and treat stress pathology. II. The Teltow study

Abstract

The present study was designed to test the clinical utility of Neuropattern (NP), a newly developed translational diagnostic tool. NP consists of biological and psychological measures that facilitate the identification of functional changes (called “neuropatterns”) in patients with stress-related health problems. In this prospective, randomized control trial, we expected NP to improve therapeutic efficacy, as compared with the usual treatment. NP was applied to 101 in-patients suffering from various mental disorders (mainly depression, anxiety disorders, and adjustment disorders), and scoring high on the Symptom Checklist-90-R (SCL-90-R) somatization scale. The patients (73% females, mean ± standard deviation age 46 ± 9.03 years) were randomly assigned to two groups: in the experimental group (n = 51), physicians received results from NP diagnostics, while in the control group (n = 50), this information was not available until discharge from the hospital. Improvements of symptoms in consequence of treatment were monitored by two self-rating scales, the SCL-90-R and Short Form-12 health survey, and a physician's clinical global rating (Beeinträchtigungs–Schwere Score). There was a significantly greater improvement in the experimental group in the self-rating assessments on symptom severity (p = 0.03) and quality of life (p = 0.05), but not in the observer rating of emotional, physical, and social-communicative functioning (p = 0.13). Treatment efficacy in patients can be improved by providing the attendant physician and the patient with diagnostic information and treatment recommendations by NP. The role of concrete mediators of treatment efficacy awaits further research.

Keywords::

• Endophenotypes
• neuropattern
• randomized control trial
• somatization
• stress
• translational stress medicine

Introduction

The diagnostic classification of mental illness according to the Diagnostic and Statistical Manual of Mental Disorders (DSM) or International Classification of Diseases (ICD) is currently based on symptoms of patients. This symptom-oriented approach, however, disregards etiological assumptions and contributes little to the development of treatments addressing the causal mechanisms of symptoms. The National Institute of Mental Health thus launched the “Research Domain Criteria” project to encourage the implementation of research that incorporates translational information into medical care (Miller 2010).

To overcome the symptom-oriented approach of current classification systems and to translate knowledge from the psychobiology of stress into health care systems, we previously introduced Neuropattern (NP), a new translational diagnostic tool for stress-related disorders (see, Hellhammer and Hellhammer 2008; Hellhammer et al. 2011). The term “stress-related disorders” refers to mental and physical symptoms that are evoked by activation of stress response networks in the brain (Hellhammer and Hellhammer 2008). The new diagnostic tool integrates interdisciplinary knowledge of stress pathology. Because of the enormous complexity of mechanisms, the focus is on the measurement of the activity and reactivity of interfaces that participate in the crosstalk between the brain and the body, e.g. the hypothalamic–pituitary–adrenal (HPA) axis, the autonomic nervous systems, and stress-related brain systems (locus caeruleus–noradrenergic system and dorsal raphe–serotonergic system). Each of these interfaces is operationalized by specific constellations of psychological, biological, and symptom measures. Depending on individual constellations on these measures, we described 13 endophenotypes of the stress response, called neuropatterns, which represent estimated states of activity and reactivity of these interfaces (Hellhammer et al. 2011). In order to make NP implementable in clinical routine and to relieve the physician, most measures can easily be taken by the patient. In the current version, the patient has to fill in questionnaires, collect saliva over several days, and make an overnight long-term electrocardiogram (ECG) recording. After all measures are completed, a central laboratory integrates all diagnostic data and generates diagnostic reports for the patient and the physician. It is the explicit goal of this diagnostic tool to individualize treatment recommendations, by recommending selected pharmacological and psychotherapeutic treatments with respect to the individual diagnostic results.

There is much research on the psychobiological basis of stress-related disorders and different approaches have been proposed (Juster et al. 2010; Sonino and Peruzzi 2009). However, to the best of our knowledge, NP is the first practical diagnostic tool for individualized treatment in clinical routine. The effects of NP are currently tested in a series of randomized controlled trials (RCTs). We here report data from the first RCT, investigating whether informing the treating physicians about patients' individual NP results and personalized treatment recommendations can help to improve therapeutic success, as compared with treatment as usual. As NP diagnostics were conceptualized to assess stress-related dysfunctions of the autonomic nervous system and the pituitary–adrenal axis, we decided to select patients with high scores on a somatization scale. We expected that NP information would result in a stronger reduction of perceived symptom severity and improved mental and physical health of the patient.

Notably, this study was a first pilot study of NP in routine clinical practice. As patients were already being treated by expert physicians, we hypothesized that NP could provide a valuable contribution to treatment success. If so, we planned to run a replication study, which might justify a complex and costly monitoring of potential mediators of such treatment effects.

Materials and methods

Participants

Patients were treated as in-patients in a department of behavioral and psychosomatic medicine (Center Seehof, Teltow, Germany). Patients with one or more diagnoses of depression, anxiety disorders, personality disorders, somatoform disorders, or adjustment disorder, and high somatization, as defined by a somatization score T>60 in the Symptom Checklist (SCL-90-R; Franke 2002) qualified for inclusion in the study. The SCL-90-R subscale somatization includes 12 items from physical stress to functional disabilities. We excluded patients whose medication would directly interfere with biological measures (e.g. cortisone treatment), non-native speakers, and patients who were expected to stay less than 6 weeks in the hospital. As shown in Figure 1, 139 in-patients met the inclusion and exclusion criteria, 106 gave their written informed consent to participate in the study, 5 dropped out for various reasons; so data from 101 patients could be analyzed. For this sample, the mean age was 46.7 (standard deviation, SD = 9.03) years with 73% females. Subjects stayed in the rehabilitation center for 47.5 (SD = 10.26) days on average.

Figure 1.  Diagram of the study design. Flow charts are shown for the experimental and control groups. Missing = participants with missing data for follow-up. SCL-90-R, symptom checklist-90-R; SF-12, short form health questionnaire.

Randomization

The study was designed as a parallel two-group pre/post-test trial with balanced randomization. Upon study entry, subjects were randomly assigned to one of two groups according to patient-identification number on hospitalization. For the experimental group (n = 51), the physicians were given access to results from the NP diagnostics soon after admission. For the control group (n = 50), the physicians received this information before discharge from the hospital. Thirteen residents in psychiatry supervised by three senior psychiatrists treated patients with and without feedback.

Study procedure

Randomization and enrollment of patients were organized by a trained study nurse at the study site who was not involved in the trial. Patients who fulfilled all criteria were informed about the study, received written patient information, and were asked for their written consent before study entry. Diagnostic assessments were made with the mini-international neuropsychiatric interview (M.I.N.I.; Sheehan et al. 1998), a short structured diagnostic interview for mental disorders according to DSM-IV or ICD-10.

Collection of NP diagnostic data began about 5 days after admission. Patients had to fill in the patient health questionnaire (Spitzer et al. 1999, 2000), a screener for ICD diagnoses of depressive disorders, somatoform disorders, eating disorders, anxiety disorders, and alcohol abuse. Additionally, the NP questionnaires (NPQs) were used: First, an anamnestic questionnaire was filled in by the physician (NPQ-A). Then, the NPQ-P was completed for psychological and symptom measures of stress. The NPQ-S was used to assess perceived stress responses during the last month and experienced life events that were considered responsible for the onset of the complaints. A third questionnaire (NPQ-PSQ (Pre-/postnatal Stress Questionaire)) retrospectively recorded pre-, peri-, and postnatal adverse events. In addition to the questionnaires, patients were instructed to run an overnight ECG recording for heart rate variability analyses. On three consecutive days, 16 saliva samples for the assessment of cortisol were collected. On the evening of the second day, patients were given 0.25 mg of dexamethasone orally, to assess feedback sensitivity of the HPA axis.

Immediately after data collection, diagnostic information was sent to the NP laboratory, where data were analyzed and underwent an integrated pattern evaluation. Based on this evaluation, a diagnostic report was generated by one of the authors (DH), who was blinded about group allocation. The software-aided data analysis followed a priori defined criteria for the single neuropatterns. The report described the observed endophenotypes and their possible interactions. Additionally, the report provides recommendations for pharmacological and psychotherapeutic treatment options to the physician; furthermore, an explanatory disease model was provided (Hellhammer et al. 2011).

To evaluate the usefulness of NP, patients filled in an abbreviated form of the Short Form (SF-12) health questionnaire that is routinely used by the German Socioeconomic Panel. As in the original SF-12, the items of the shortened questionnaire (10 items) are assigned to subscales termed “mental health” and “physical health” (Bullinger and Kirchberger 1998; Ware et al. 2002; Nuebling et al. 2006; Wagner et al. 2007). The reliability (Cronbach's α) for each scale was good (mental health: α = 0.91) or at least satisfactory (physical health: α = 0.84).

A German version of the SCL-90-R (Derogatis 1992; Franke 2002) was filled in by patients after admission and before discharge. We used the global severity index (GSI) as a highly reliable index of perceived symptom load (Cronbach's α in the present sample = 0.97; Franke 2002).

Physicians performed a clinical global impression rating after admission and before discharge, using three items to globally appraise the patient's emotional, physical, and social-communicative functioning before and after treatment (Beeinträchtigungs–Schwere Score, BSS; Schepank 1995). According to the BSS manual (Schepank 1995), a high inter-rater reliability can be expected for trained raters (r = 0.90). In this study, as each patient was rated by just one physician, no index of inter-rater reliability could be calculated. Instead, we assessed the internal consistency for the three different ratings and determined a Cronbach's α of 0.69.

The study protocol was reviewed and approved by the institutional review board of the Deutsche Rentenversicherung Bund (German Federal Pension Insurance Agency). The study was conducted in accordance with the declaration of Helsinki.

Intervention

NP diagnostic reports were sent to the physicians in the experimental group. The physicians were free to use this information in the treatment of the patients. They also could inform the patient about the diagnostic results, if this was considered appropriate. Physicians treating subjects of the control group received the NP diagnostic report once the treatment period had ended, at the time the patients were discharged.

As patients may suffer from all types of mental disorders, initially a thorough diagnostic work-up was done. Depending on the result, individualized treatments were applied. This could include psychotherapy, pharmacotherapy, somatic medical treatment, occupational therapy, sports therapy, physiotherapy, or social support. Psychotherapy was based on principles of cognitive behavior therapy (Linden and Hautzinger 2011) with at least two single sessions and two group sessions per week and regular supervision. Because of the complexity of the treatment regimen, the study was not designed to assess treatment of individual patients in detail.

Statistical analysis

After completion of the study, statistical analyses were performed using IBM SPSS statistical software package version 17 (IBM SPSS, Chicago, IL, USA). Interim analyses were not conducted. Data were checked for normality and appropriate statistical tests were chosen for analyses. Descriptive statistics were used to describe the sample. Baseline differences in treatment groups are reported using mean value and SD, and were subsequently tested by t-tests and χ² tests. To analyze effects of change, gain scores were computed and single one-sided t-tests were applied. Cases with missing data were case-wise excluded from analyses.

Results

Sample characteristics are shown in Table I. Experimental and control groups were compared for initial differences in age, sex, level of education, diagnosis, duration of hospital stay, application for a pension, as well as for differences in pre-scores with SCL-90-R, SF-12, and BSS. According to the selection of patients, all showed a high score on the SCL-90-R subscale of “somatization.” Similarly, the global SCL-90-R score and the SF-12 scores showed intensities of clinical relevance. The diagnoses according to the M.I.N.I. standardized diagnostic interview showed that patients were suffering from various mental disorders, most frequently from depression, adjustment disorders, and anxiety disorders, with no significant differences between groups (for each test: p>0.16).

Table I.  Sample characteristics.

	Control group (n = 50)	Experimental group (n = 51)	p
Age: M (SD)	46.86 (9.43)	46.57 (8.72)	0.87
Sex (n; male:female)	12:39	15:35	0.51
Level of education (n)			0.61
Special education school	1	0
Mid-level secondary school	32	29
High-school graduate	15	17
College graduate	2	3
Duration of hospitalization in days: M (SD)	49 (11.26)	46.14 (9.19)	0.16
Application for a pension (n)	4	5	0.51
Health and well-being (pre-score)
SCL-90-R (somatization): M (SD)	1.52 (0.50)	1.49 (0.55)	0.76
SCL-90-R (GSI): M (SD)	1.40 (0.61)	1.40 (0.67)	0.99
SF-12 (Mental health): M (SD)	2.40 (0.63)	2.27 (0.62)	0.31
SF-12 (Physical health): M (SD)	2.64 (0.79)	2.80 (0.85)	0.32
BSS: M (SD)	5.24 (1.79)	5.61 (1.98)	0.33
Diagnoses (M.I.N.I.)
Depression (n)	34	32	0.68
Generalized anxiety disorder (n)	21	20	0.84
Personality disorder (n)	8	5	0.39
Somatoform disorder (n)	2	2	>0.99
Adjustment disorder (n)	23	24	>0.99

Note: M = mean, SD = standard deviation, n = number of participants. BSS, Beeinträchtigungs-Schwere Score; GSI, global severity index; M.I.N.I.: mini-international neuropsychiatric interview; SCL-90-R, symptom checklist-90-revised; SF-12, short form health questionnaire. Statistical comparisons by Student's t-tests or χ² tests.

Among the 101 patients, there were 68 different constellations of neuropatterns, from which different individualized treatment recommendations were derived. As expected, most patients fulfilled criteria for serotonin hypoactivity (n = 47), norepinephrine hyperactivity (n = 36), norepinephrine hyper-reactivity (n = 36), and/or corticotropin releasing factor-hyperactivity (n = 28). Four patients did not qualify for any neuropattern.

Comparison of treatment outcome

Percentage pre–post differences in outcome measures between groups are shown in Figure 2. There was a significantly better improvement in the experimental group than in the control group for the GSI (t = − 1.88, df (degree of freedom) = 96, p_one-tailed = 0.03). The same was found for the SF-12 mental health subscale (t = − 1.62, df = 91, p_one-tailed = 0.05), while there was no significant difference between groups for the SF-12 physical health score (t = − 0.002, df = 96, p_one-tailed = 0.56), or the observer's BSS rating on emotional, physical, and social communicative functioning (t = − 1.13, df = 99, p_one-tailed = 0.13).

Figure 2.  Mean change (pre–post difference in percent) for different measures of treatment success in the experimental and the control groups. Note: *significant group difference, p < 0.05_one-tailed, Student's t-test; error bars show SEM (Standard Error of Mean). For the short form health questionnaire (SF-12), clinical improvement is indicated by a positive difference score (increase), whereas for the SCL-90-R (GSI) and the BSS, an improvement is indicated by a negative difference score (reduction). Number of patients: experimental, n = 51; control, n = 50.

Discussion

This study aimed to investigate the effects of NP information on the outcomes of treatment of patients with stress-related disorders and high scores on the SCL-90-R somatization subscale. Our data show a better improvement in the NP information group than in the control group without such information. With respect to the two self-report measures, patients in the intervention group showed a larger reduction in the severity of symptoms and a greater improvement in self-rated mental health during treatment than those in the control group. However, patients' self-ratings of physical health did not reveal significant group differences. In addition, the emotional, physical, and social communicative functioning ratings by the attending physician (by BSS) did not reach significance, although the experimental group developed in the expected direction.

Despite shared variance between the outcome measures, the questionnaires measured slightly different aspects of health. Hence, it seems not unexpected that results differed among the instruments to some extent. Specifically, the observer's BSS rating did not show significant correlations with the self-rating instruments (r_{SF-12 mental} = − 0.12, r_GSI = 0.22, and r_{SF-12 physical} = 0.25). This is not unexpected as the SCL-90-R and the SF-12 assess perceived symptom load, while the BSS measures degrees of emotional, physical, and socio-communicative functioning, as observed by the rater.

In this initial RCT with NP, the study protocol was not designed to assess any mediators of potential treatment effects. Thus, the improvement in the experimental group cannot be attributed to any specific factor. In this study, we did not control whether the physicians used the diagnostic information to inform the patient, or how much of the information was used, or whether the information was used to adjust the treatment regimen. Hence, it is only possible that NP recommendations were considered by the physician for pharmaco- and/or psychotherapeutic treatment. Improvement, especially in the self-rating measures, may additionally or alternatively be the result of providing an explanatory disease model for the patient. Patients may have appreciated receiving biological data, reassuring their assumptions about biomedical causation of their illness, thus diminishing subjective complaints. However, an exploratory single item moderation analysis did not support the latter assumption. Furthermore, because of the high variability in treatment in the hospital, improvement in the experimental group could possibly be attributed to a different treatment in comparison with the control group (e.g. more physical activity).

Because of these limitations in the present study, the determinants for the positive effect may have varied greatly among the patients. Nevertheless, the data from the present study indicate the need for and appropriateness of further analysis of mediating or causative factors enhancing treatment efficacy in future RCTs. Upcoming studies should implement measures of treatment credibility/expectancy, plausibility, and appreciation of the disease model by the physicians, and adherence to the recommendations given in the NP diagnostic report. Yet another aspect is whether patients with certain diagnoses may particularly profit from NP. Several mental and physical disorders are closely linked to psychobiological stress responses (McEwen 1998; Chrousos 2009; Lupien et al. 2009). Thus, patients with such diagnoses may possibly profit most from the NP diagnostic tool. All these aspects need to be monitored in future studies by ratings of the patients, the physician, and by objective data on direct and indirect costs.

In conclusion, the data reported here will encourage further investigation of the therapeutic efficacy of NP in patients with stress-related disorders. Ongoing studies are carefully monitoring potential moderators or mediators of NP treatment efficacy. Notably, NP was not developed for in-patients treated by psychiatrists, but rather for out-patients treated by family physicians. We expect that NP will be most helpful for general practitioners, as it is not always easy to reliably diagnose stress-related disorders in everyday practice. In particular, physiological stress cannot be reliably perceived and communicated by the patient. Furthermore, in contrast to family physicians, psychiatrists provide expert treatment for these patients. Our currently running studies with out-patients will show whether favorable effects can also be generated in out-patients who are not treated by specialists.

Acknowledgements

The authors would like to thank the on-site study nurse Christine Kuehn for patient recruitment and patient management, Christel Neu for excellent study coordination in Trier as well as all attending physicians and staff at the Center Seehof, Teltow, Germany. This study was supported by a grant of the Rhineland-Palatinate foundation for innovations.

Declarations of interest: This study was supported by a grant of the Rhineland-Palatinate foundation for innovations. The research grant awarded to DH implicates the goal to make Neuropattern commercially available to physicians and participants in future. Neuropattern is a registered trademark. DH holds the copyright on all questionnaires.