دانلود مقاله ISI انگلیسی شماره 39216
عنوان فارسی مقاله

کشف شاخصهای سرمی با پیش بینی توسعه یک اپیزود افسردگی پس از آن در اختلال اضطراب اجتماعی

کد مقاله سال انتشار مقاله انگلیسی ترجمه فارسی تعداد کلمات
39216 2015 9 صفحه PDF سفارش دهید محاسبه نشده
خرید مقاله
پس از پرداخت، فوراً می توانید مقاله را دانلود فرمایید.
عنوان انگلیسی
Discovery of serum biomarkers predicting development of a subsequent depressive episode in social anxiety disorder
منبع

Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)

Journal : Brain, Behavior, and Immunity, Volume 48, August 2015, Pages 123–131

کلمات کلیدی
هراس اجتماعی - غمگین - اختلال عاطفی - اختلال خلقی - پیش بینی - اختلال افسردگی ماژور - عامل خطرساز - افسردگی پیش آگهی - افسردگی ثانویه - دیستایمیک
پیش نمایش مقاله
پیش نمایش مقاله کشف شاخصهای سرمی با پیش بینی توسعه یک اپیزود افسردگی پس از آن در اختلال اضطراب اجتماعی

چکیده انگلیسی

Abstract Although social anxiety disorder (SAD) is strongly associated with the subsequent development of a depressive disorder (major depressive disorder or dysthymia), no underlying biological risk factors are known. We aimed to identify biomarkers which predict depressive episodes in SAD patients over a 2-year follow-up period. One hundred sixty-five multiplexed immunoassay analytes were investigated in blood serum of 143 SAD patients without co-morbid depressive disorders, recruited within the Netherlands Study of Depression and Anxiety (NESDA). Predictive performance of identified biomarkers, clinical variables and self-report inventories was assessed using receiver operating characteristics curves (ROC) and represented by the area under the ROC curve (AUC). Stepwise logistic regression resulted in the selection of four serum analytes (AXL receptor tyrosine kinase, vascular cell adhesion molecule 1, vitronectin, collagen IV) and four additional variables (Inventory of Depressive Symptomatology, Beck Anxiety Inventory somatic subscale, depressive disorder lifetime diagnosis, BMI) as optimal set of patient parameters. When combined, an AUC of 0.86 was achieved for the identification of SAD individuals who later developed a depressive disorder. Throughout our analyses, biomarkers yielded superior discriminative performance compared to clinical variables and self-report inventories alone. We report the discovery of a serum marker panel with good predictive performance to identify SAD individuals prone to develop subsequent depressive episodes in a naturalistic cohort design. Furthermore, we emphasise the importance to combine biological markers, clinical variables and self-report inventories for disease course predictions in psychiatry. Following replication in independent cohorts, validated biomarkers could help to identify SAD patients at risk of developing a depressive disorder, thus facilitating early intervention.

مقدمه انگلیسی

. Introduction Social anxiety disorder (SAD; also referred to as “social phobia”) is among the most common anxiety spectrum disorders with a 12-month prevalence ranging between 2% and 7% (Kessler et al., 2012 and Wittchen et al., 2011). Defined by a marked fear of social situations, the affected individual avoids situations associated with exposure to possible scrutiny by others (American-Psychiatric-Association, 2013). The age of onset is usually in childhood or adolescence (Ballenger et al., 1998). Despite the associated distress and impairment, only half of the patients fulfilling diagnostic criteria for SAD ever seek help. This results in a median delay of over two decades until correct diagnosis and initial treatment, the longest delay amongst all psychiatric disorders investigated in the US National Comorbidity Survey Replication (Wang et al., 2005). In addition to the characteristic chronic course, SAD patients frequently (20–30%) present with co-morbid major depressive disorder (MDD) (Stein et al., 1990, Merikangas and Angst, 1995 and Lewinsohn et al., 1997), with SAD being the most prevalent co-morbid anxiety disorder in patients suffering from depressive disorders (Pini et al., 1997 and Rush et al., 2005). Co-morbidity is associated with a more severe and chronic disease course and worse clinical outcome (Stein et al., 2001 and Beesdo et al., 2007). The vast majority of SAD patients present initially with social anxiety symptoms (Kessler et al., 1999) and develop a co-morbid depression on average within 5 years (Beesdo et al., 2007). Consistent with these findings, SAD has been shown to be an important predictor/risk factor of a subsequent depressive disorder independent of the age of onset (Stein et al., 2001 and Beesdo et al., 2007). Furthermore, apart from a SAD lifetime history, distinct psychological constructs within the SAD symptom spectrum (e.g. behavioral inhibition) have also been shown to be predictive of the future onset of depression (Beesdo et al., 2007). Other characteristics of anxiety disorders that have been linked to an increased risk of developing a depressive disorder include the level of anxiety-associated impairment (Bittner et al., 2004) and the presence of multiple anxiety disorders (Woodward and Fergusson, 2001) or panic attacks (Goodwin, 2002). However, little is known about the molecular mechanisms involved in the onset of either anxiety or depressive disorders. Changes in cortisol awaking response (Adam et al., 2010) and serum interleukin 6 (Khandaker et al., 2014) have been reported to be predictors for the future onset of depression in adolescence. So far no biomarkers have been associated with the development of depressive episodes in anxiety disorder patients. Awareness of factors that predict increased susceptibility for the onset of a depressive disorder within the SAD patient population could lead to an improved clinical outcome due to early intervention (Kessler et al., 1999). In the present study, we investigated molecular changes in serum collected from patients diagnosed with SAD without current co-morbid depressive disorders with the objective to identify a molecular biomarker panel aiding in the prediction of the onset of a depressive disorder within a 2-year follow-up period. We analyzed the discriminative power of serum protein changes at the time of baseline clinical assessment of 165 analytes using multiplexed immunoassays. Biomarkers were initially identified in 72 patients diagnosed solely with SAD and the analysis was then expanded to include SAD patients with other co-morbid anxiety disorders (totalling 143 SAD patients) in order to account for multiple anxiety spectrum co-morbidities in SAD individuals. A selection of these candidate biomarkers was combined into an optimized panel of four serum analytes. Finally, we evaluated the predictive performance of the identified biomarker panels alone and in combination with psychiatric and somatic patient variables selected from structured patient assessments in order to determine their potential for clinical application.

نتیجه گیری انگلیسی

3. Results 3.1. Demographics An overview of demographic characteristics is provided in Table 1 and Table 2. Individuals with SAD who developed a depressive disorder within 2-year follow-up (SAD converters, n = 22) did not differ significantly from those individuals who did not develop a depressive disorder (SAD non-converters, n = 50) in relation to baseline age, education, BMI, presence of non-psychiatric chronic disorders, smoking status, baseline diagnosis of alcohol use disorders, lifetime history of depressive disorders, FQ scores, BAI scores and psychiatric medication, psychotherapeutic treatment or non-psychiatric medication ( Table 1). However, SAD converters scored significantly higher on the baseline IDS scale compared to SAD non-converters (p-value 0.012) and this difference remained significant for the 2-year re-assessment (p-value 8.47E-03, note that due to the cross-sectional nature of the study design not all SAD converters were experiencing an acute depressive episode at the time point of the 2-year follow-up e.g. due to psychiatric treatment having been initiated following baseline assessment). There was a trend (p-value 0.072) towards a gender bias in the analyzed individuals, with more females in the SAD converter group. Expanding the sample size to include individuals with SAD and co-morbid PDA, PD, AP or GAD (n = 143), those who developed a depressive disorder within 2-year follow-up (SADcom converters, n = 47) had higher baseline IDS scores (p-value 2.04E-04) compared to patients who did not (SADcom non-converters, n = 96) and higher IDS scores at the 2-year follow-up time point (p-value 1.36E-05) ( Table 2). None of the other sociodemographic, health-related and clinical characteristics, self-report inventories or psychiatric treatment variables differed between the groups. Table 1. Group characteristics of SAD depression converters and non-converters without co-morbid anxiety disorders. Values are presented as mean ± standard deviation or as percentage. SAD non-converter = individuals with SAD without comorbid anxiety diagnosis, who did not develop a depressive disorder during 2-year follow-up; SAD converter = individuals with SAD without comorbid anxiety diagnosis, who developed a depressive disorder during 2-year follow-up. BMI = body mass index; FQ = Fear Questionnaire; BAI = Beck Anxiety Inventory; IDS = Inventory of Depressive Symptomatology; SSRI = selective serotonin re-uptake inhibitors; SNRI = serotonin-norepinephrine re-uptake inhibitors; TCA = tricyclic antidepressants; ANX = anxiolytics including benzodiazepines. SAD non-converter SAD converter p-Value n Number 50 22 Diagnoses Social anxiety disorder 50 22 Panic disorder with agoraphobia 0 0 Panic disorder without agoraphobia 0 0 Agoraphobia 0 0 Generalized anxiety disorder 0 0 Number of anxiety diagnoses – One anxiety diagnosis 50 22 Two anxiety diagnoses 0 0 Baseline alcohol use disorder diagnosis 34.0% 31.8% 1.000 Depressive disorder lifetime diagnosis 60.0% 59.1% 1.000 Demographics and health indicators Age 41.2 ± 12.9 42.0 ± 12.3 0.850 Females 48.0% 72.7% 0.072 Education 0.690 Basic (<6 years) 6.0% 9.1% Medium (7–12 years) 56.0% 45.5% High (>12 years) 38.0% 45.5% BMI 26.8 ± 6.3 24.7 ± 4.2 0.269 Smoking state 0.257 Current smokers 26.0% 45.5% Former smokers 40.0% 31.8% Non smokers 34.0% 22.7% Treated chronic diseases Total chronic non-psychiatric disorders 32.0% 40.9% 0.592 Cancers 2.0% 4.5% 0.521 Cardiovascular diseases 6.0% 4.5% 1.000 Respiratory diseases 2.0% 9.1% 0.219 Diabetes 2.0% 0.0% 1.000 Self-report symptom inventories FQ 34.6 ± 19.4 38.4 ± 15.6 0.206 FQ social phobia subscale 18.7 ± 8.2 21.9 ± 7.8 0.174 FQ agoraphobia phobia subscale 8.0 ± 8.9 8.1 ± 6.9 0.486 FQ blood and injury phobia subscale 7.9 ± 7.6 8.4 ± 7.2 0.745 BAI 12.4 ± 7.6 13.1 ± 6.4 0.448 BAI somatic subscale 7.4 ± 4.9 8.1 ± 4.9 0.473 BAI subjective subscale 5.1 ± 3.6 5.0 ± 2.8 0.844 Baseline IDS 20.4 ± 9.4 25.9 ± 8.4 0.012 2-year follow-up IDS 15.1 ± 8.7 21.5 ± 9.8 8.47E-03 Psychiatric treatment Psychotherapy 34.0% 36.4% 1.000 Any antidepressant 18.0% 4.5% 0.161 SSRI 14.0% 4.5% 0.421 SNRI 6.0% 0.0% 0.548 TCA 6.0% 0.0% 0.548 ANX 12.0% 4.5% 0.427 Non-psychiatric medication Any cardiovascular medication 14.0% 13.6% 1.000 Any anti-inflammatory medication 4.0% 4.5% 1.000 Any hypolipidemic medication 6.0% 0.0% 0.548 Table options Table 2. Group characteristics of SAD depression converters and non-converters with and without co-morbid anxiety disorders. Values are presented as mean ± standard deviation or as percentage. SADcom non-converter = individuals with SAD with or without comorbid anxiety diagnoses, who did not develop a depressive disorder during 2-year follow-up; SADcom converter = individuals with or without SAD without comorbid anxiety diagnoses, who developed a depressive disorder during 2-year follow-up. BMI = body mass index; FQ = Fear Questionnaire; BAI = Beck Anxiety Inventory; IDS = Inventory of Depressive Symptomatology; SSRI = selective serotonin re-uptake inhibitors; SNRI = serotonin-norepinephrine re-uptake inhibitors; TCA = tricyclic antidepressants; ANX = anxiolytics including benzodiazepines. SADcom non-converter SADcom converter p-Value n Number 96 47 Diagnoses Social anxiety disorder 96 47 Panic disorder with agoraphobia 23 12 Panic disorder without agoraphobia 7 3 Agoraphobia 11 3 Generalized anxiety disorder 5 7 Number of anxiety diagnoses 0.678 One anxiety diagnosis 50 22 Two anxiety diagnoses 46 25 Baseline alcohol use disorder diagnosis 30.2% 31.9% 0.849 Depressive disorder lifetime diagnosis 58.3% 72.3% 0.140 Demographics and health indicators Age 39.8 ± 12.7 41.5 ± 11.1 0.433 Females 60.4% 72.3% 0.195 Education 0.817 Basic (<6 years) 5.2% 4.3% Medium (7–12 years) 58.3% 63.8% High (>12 years) 36.5% 31.9% BMI 25.9 ± 6.0 25.8 ± 5.5 1.000 Smoking state 0.225 Current smokers 28.1% 42.6% Former smokers 40.6% 31.9% Non smokers 31.3% 25.5% Treated chronic diseases Total chronic non-psychiatric disorders 37.5% 42.6% 0.588 Cancers 2.1% 2.1% 1.000 Cardiovascular diseases 5.2% 4.3% 1.000 Respiratory diseases 8.3% 12.8% 0.389 Diabetes 3.1% 2.1% 1.000 Self-report symptom inventories FQ 34.5 ± 18.4 37.0 ± 16.2 0.255 FQ social phobia subscale 17.0 ± 8.6 19.4 ± 8.1 0.117 FQ agoraphobia phobia subscale 10.1 ± 9.1 9.6 ± 8.6 0.873 FQ blood and injury phobia subscale 7.4 ± 7.4 8.1 ± 6.5 0.309 BAI 13.8 ± 8.5 14.8 ± 7.0 0.215 BAI somatic subscale 8.2 ± 5.8 9.0 ± 5.4 0.256 BAI subjective subscale 5.6 ± 3.7 5.8 ± 3.1 0.530 Baseline IDS 21.2 ± 9.3 27.0 ± 8.3 2.04E-04 2-year follow-up IDS 15.0 ± 8.4 22.9 ± 10.0 1.36E-05 Psychiatric treatment Psychotherapy 33.3% 46.8% 0.143 Any antidepressant 19.8% 12.8% 0.355 SSRI 17.7% 12.8% 0.629 SNRI 3.1% 0.0% 0.551 TCA 3.1% 0.0% 0.551 ANX 14.6% 12.8% 1.000 Non-psychiatric medication Any cardiovascular medication 13.5% 14.9% 0.803 Any anti-inflammatory medication 3.1% 2.1% 1.000 Any hypolipidemic medication 5.2% 4.3% 1.000 Table options 3.2. Biomarker discovery in SAD patients without other co-morbid anxiety disorders Stepwise logistic regression analyses predicting SAD depression converter status for all 165 individual serum analytes (single analyte effects) resulted in the identification of two significantly associated analytes (adjusted p-values <0.05): AXL receptor tyrosine kinase (AXL) and vascular cell adhesion molecule 1 (VCAM1) ( Table 3). For AXL two self-report inventories were selected as covariates: the baseline IDS and the total BAI scores. Baseline IDS and the total BAI score were also selected for VCAM1 along with the BAI somatic and subjective subscale. The additional covariates for VCAM1 suggested that the total BAI score and the somatic and subjective subscales provided independent information for the underlying model. Serum concentrations for both analytes, AXL and VCAM1, were found to be lower in SAD converters in both cases. To model the joint effect of analytes and covariates, we applied stepwise logistic regression to AXL ( Fig. 1a), VCAM1 and all available covariates. This resulted in the selection of AXL, baseline IDS score and the BAI total score as the optimal set for prediction of SAD converter status ( Fig. 1b). A combination of the clinical covariates and self-report inventories alone resulted in a poor to fair performance (AUC = 0.70, 95%CI = 0.57–0.83, sensitivity = 55%, specificity = 82%, PPV = 57%, NPV = 81%, ACC = 74%). AXL alone resulted in a good performance (AUC = 0.84, 95%CI = 0.75–0.93, sensitivity = 100%, specificity = 56%, PPV = 50%, NPV = 100%, ACC = 69%). The combination of AXL, baseline IDS score and BAI total score resulted in a further improvement of performance (AUC = 0.88, 95%CI = 0.80–0.96, sensitivity = 95%, specificity = 70%, PPV = 58%, NPV = 97%, ACC = 78%). Table 3. Analytes significantly associated with depression conversion/non-conversion status in different SAD ± anxiety disorder comorbidity groups. Values in parentheses represent the number of non-converters/converters. Abbrev. = analyte abbreviation; Coeff. = logistic regression coefficient; Std. error = standard error; PDA = panic disorder with agoraphobia; PD = panic disorder without agoraphobia; AP = agoraphobia; GAD = generalized anxiety disorder; IDS = baseline Inventory of Depressive Symptomatology; BAI = Beck Anxiety Inventory; BAIsub = Beck Anxiety Inventory subjective subscale; BAIsom = Beck Anxiety Inventory somatic subscale; FQsp = Fear Questionnaire social phobia subscale; BMI = body mass index; RD = chronic treated respiratory diseases; adj. p-value = Benjamini–Hochberg procedure ( Benjamini and Hochberg, 1995) post hoc corrected p-value. Group Analyte Abbrev. Coeff. Std. error p-Value Covariates adj. p-value SAD (n = 50/22) Receptor tyrosine kinase AXL AXL −13.42 3.64 2.29E-04 IDS + BAI 0.034 Vascular cell adhesion molecule 1 VCAM1 −20.91 5.92 4.11E-04 BMI + IDS + BAI + BAIsom + BAIsub 0.034 SAD ± PDA (n = 73/34) Receptor tyrosine kinase AXL AXL −8.09 2.19 2.21E-04 IDS + BAIsub 0.035 Vascular cell adhesion molecule 1 VCAM1 −11.75 3.39 5.37E-04 IDS + BAIsub 0.043 SAD ± PD (n = 57/25) Receptor tyrosine kinase AXL AXL −13.78 3.59 1.23E-04 FQsp 0.020 Vascular cell adhesion molecule 1 VCAM1 −16.30 4.65 4.61E-04 BMI + IDS + BAI + BAIsom + BAIsub 0.037 Insulin-like growth factor binding protein 3 IGFBP3 −21.70 6.40 6.98E-04 Gender + BMI + IDS 0.037 Collagen IV CollIV −6.40 1.94 9.71E-04 Gender + BMI + IDS + BAI 0.039 Vitronectin Vitro 12.90 4.00 1.26E-03 BMI + IDS + BAI 0.040 SAD ± AP (n = 61/25) Receptor tyrosine kinase AXL AXL −9.71 2.58 1.70E-04 IDS + BAIsub 0.025 Vascular cell adhesion molecule 1 VCAM1 −14.72 4.09 3.14E-04 BMI + IDS + BAIsub 0.025 Insulin-like growth factor binding protein 3 IGFBP3 −21.65 6.42 7.39E-04 Gender + BMI + IDS 0.039 SAD ± GAD (n = 55/29) Receptor tyrosine kinase AXL AXL −11.4 3.11 2.59E-04 IDS + RD 0.041 SAD ± PDA/PD/AP/GAD (n = 96/47) Receptor tyrosine kinase AXL AXL −6.50 1.74 1.83E-04 IDS 0.029 Table options Evaluation of the performance of biomarkers, clinical variables and self-report ... Fig. 1. Evaluation of the performance of biomarkers, clinical variables and self-report inventories predictive of depression conversion in SAD patients without other co-morbid anxiety disorder diagnoses (SAD converters (n = 22); SAD non-converters (n = 50)). (a) Boxplots of the unadjusted serum concentration of AXL in SAD converters and SAD non-converters without co-morbid anxiety disorders. (b) ROC curve analysis showing the performance of the biomarker AXL alone (blue), selected clinical and self-report covariates alone (black) and a combination of clinical and self-report variables and AXL (red) in predicting the development of a depressive disorder within 2-year follow-up. Values in parentheses represent the AUC. Stepwise selection using only clinical and self-report variables yielded an AUC = 0.72, 95%CI = 0.60–0.85 using baseline IDS and BMI (ROC curve not shown). AXL = AXL receptor tyrosine kinase; IDS = baseline Inventory of Depressive Symptomatology; BAI = Beck Anxiety Inventory; BMI = body mass index. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Figure options 3.3. Biomarker discovery in SAD patients with and without other co-morbid anxiety disorders Logistic regression predicting depressive disorder converter status in the group of SAD patients with or without co-morbid anxiety disorders (single analyte effects) resulted in the identification of five serum analytes (AXL, VCAM1 and three additional candidate markers: insulin-like growth factor binding protein 3 (IGFBP3), collagen IV (CollIV) and vitronectin (Table 3)). The association between these analytes and converter/non-converter status did not differ between SAD patients with and without co-morbid anxiety disorders (see Supplementary Table 2 for detailed anxiety disorder association heterogeneity analyses). SADcom converters displayed decreased serum concentrations of AXL (Fig. 2a), VCAM1, IGFBP3 and CollIV and increased concentrations of vitronectin compared to SADcom non-converters. Stepwise logistic regression was applied to AXL, VCAM1, IGFBP3, CollIV, vitronectin and all available clinical covariates and self-report inventories (joint effects model) and resulted in the selection of AXL, VCAM1, CollIV, vitronectin, baseline IDS, BAI somatic subscale, lifetime history of depressive disorder and BMI as predictors of SADcom converter status in our naturalistic cohort design (Fig. 2b). When we allowed for co-morbid anxiety disorders, a similar degree of discrimination was obtained to that observed in the SAD patient sample without co-morbid anxiety disorders, using only a combination of clinical variables and self-report inventories. Baseline IDS, BAI somatic subscale, lifetime history of depressive disorder and BMI resulted in a poor to fair performance (AUC = 0.70, 95%CI = 0.62–0.79, sensitivity = 68%, specificity = 67%, PPV = 50%, NPV = 81%, ACC = 67%). The biological marker set comprising AXL, VCAM1, IGFBP3 and CollIV reached a fair performance (AUC = 0.78, 95%CI = 0.70–0.86, sensitivity = 70%, specificity = 78%, PPV = 61%, NPV = 84%, ACC = 75%). Finally, the combination of biological markers, clinical variables and self-report inventories reached a good performance (AUC = 0.86, 95%CI = 0.79–0.92, sensitivity = 77%, specificity = 81%, PPV = 66%, NPV = 88%, ACC = 80%). In order to test whether this final panel combining converter-related serum analytes, clinical variables and self-report inventories predicted subsequent depressive episodes in the other populations assessed in NESDA, we applied the fitted eight marker SAD regression model to four other anxiety disorder groups (PDA, PD, AP or GAD) and healthy controls (all five groups without a depressive disorder baseline diagnosis). The predictive performance ranged between fail and poor in all anxiety disorder groups (PDA: 51 non-converters, 26 converters, AUC = 0.64; PD: 32 non-converters, 18 converters, AUC = 0.56; AP: 39 non-converters, 9 converters, AUC = 0.58; GAD: 32 non-converters, 7 converters, AUC = 0.64) and was fair in healthy controls (373 non-converters, 26 converters, AUC = 0.72). Evaluation of the performance of biomarkers, clinical variables and self-report ... Fig. 2. Evaluation of the performance of biomarkers, clinical variables and self-report inventories predictive of depression conversion in SAD patients with and without co-morbid anxiety diagnoses (SAD converters (n = 47); SAD non-converter (n = 96)). (a) Boxplots of the unadjusted serum concentration of AXL in SAD converters and SAD non-converters with and without co-morbid anxiety disorders. Black dots represent SAD patients without a co-morbid anxiety disorder and red dots represent SAD patients with a co-morbid anxiety disorder. (b) ROC curve analysis showing the performance of AXL, VCAM1, vitronectin and collagen IV (blue), selected clinical and self-report covariates alone (black) and a combination of clinical variables and self-report and biomarkers (red) in predicting the development of a depressive disorder within 2-year follow-up. Values in parentheses represent the AUC. Stepwise selection using only clinical and self-report variables yielded an AUC = 0.69, 95%CI = 0.60–0.78 using the baseline IDS alone (ROC curve not shown). AXL = AXL receptor tyrosine kinase; VCAM1 = vascular cell adhesion molecule 1; Vitro = vitronectin; CollIV = collagen IV; IDS = baseline Inventory of Depressive Symptomatology; BAIsom = Beck Anxiety Inventory somatic subscale; DepLH = depressive disorder lifetime diagnosis; BMI = body mass index. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

خرید مقاله
پس از پرداخت، فوراً می توانید مقاله را دانلود فرمایید.