174 Original Paper
Authors
T. J. Stamm1, D. Becker1, L. M. Sondergeld1, K. Wiethoff1, C. Hiemke2, G. O’Malley3, R. Ricken1, M. Bauer4, M. Adli1
Affiliations
Affiliation addresses are listed at the end of the article
Key words ▶ depression ● ▶ early response ● ▶ therapeutic drug monitoring ● ▶ treatment algorithm ●
Abstract
received revised accepted
27.09.2013 30.05.2014 06.06.2014
Bibliography DOI http://dx.doi.org/ 10.1055/s-0034-1383565 Pharmacopsychiatry 2014; 47: 174–179 © Georg Thieme Verlag KG Stuttgart · New York ISSN 0176-3679 Correspondence T. J. Stamm, MD Department of Psychiatry and Psychotherapy Charité Universitätsmedizin Berlin Campus Mitte Chariteplatz 1 10117 Berlin Germany Tel.: + 49/30/450 517009 Fax: + 49/30/450 517953 [email protected]
▼
Introduction: Early assessment of a therapeutic response is a central goal in antidepressant treatment. The present study examined the potential for therapeutic drug monitoring and symptom rating to predict venlafaxine treatment efficacy (measured by overall patient response and remission). Methods: 88 patients were uptitrated homogenously to 225 mg/day venlafaxine. Serum concentrations of venlafaxine (VEN) and its active metabolite O-desmethylvenlafaxine (ODV) were measured at week 2. Continuous psychopathometric ratings were measured for up to 6 weeks by independent study raters. Results: An early improvement was significantly more common in venlafaxine responders than
Introduction
▼
Despite the introduction of a variety of novel pharmacological agents as treatment options for antidepressant therapy, treatment resistance remains a common issue in the management of depressed patients. 30–40 % of depressed patients who are appropriately treated with an antidepressant substance do not respond. Several treatment algorithms for a rational and standardized treatment of major depressive disorders have been proposed to minimize the rate of treatment refractory depression [1–5]. Antidepressant monotherapy is usually the first line of treatment, however it remains that little is known about potential clinical or biological predictors for selecting the most promising substance. In general, monotherapy is recommended for at least 4–6 weeks until a change in the therapeutic strategy is warranted. This is the case despite growing evidence showing that nonresponse to an antidepressant treatment trial
Stamm TJ et al. Prediction of Antidepressant Response … Pharmacopsychiatry 2014; 47: 174–179
non-responders (χ2; p = 0.007). While ODV serum levels were significantly higher in responders (t test; p = 0.006), VEN serum levels, sum level of VEN + ODV and the ratio of ODV/VEN levels were not. Moreover, patients who showed an early response combined with an ODV serum level above the median of 222 ng/mL were significantly more likely to achieve full response (binary logistic model; p < 0.01). Sensitivity (84 % for early response) and specificity (81 % for combination of early response and therapeutic drug monitoring) were sufficient to qualify as a reasonable screening instrument. Conclusion: Our results indicate that early improvement and ODV serum concentration are predictive of therapeutic outcome and can thus be used to guide use of the antidepressant venlafaxine.
may be detectable at an earlier stage. Nakajima et al. [6] and Szegedi et al. [7, 8] respectively reported significant predictive values of an early response: they observed a reduction of 20 % in the psychopathometric evaluation of depressive symptomatology after 2 weeks to be significantly predictive of later achievement of full remission. The usefulness of therapeutic drug monitoring (TDM) for the prediction of treatment outcome has been discussed [9]. There seems to be no relationship between dose, serum level and clinical outcome in treatment with selective serotonine reuptake inhibitors (SSRI), while for tricyclic-based substances, a serum-level response relationship has been described somewhat contradictorily. For the serotonin-norepinephrine reuptake inhibitor (SNRI) venlafaxine, a relationship between dose, clinical outcome and a favourable response with higher summation serum levels has been shown [10]. However, several methodological constrictions mean the interpretability of these studies is limited. In a similar vein, in the case of citalopram,
This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
Prediction of Antidepressant Response to Venlafaxine by a Combination of Early Response Assessment and Therapeutic Drug Monitoring
Original Paper 175
serum concentrations on day 7 of treatment have been found to be predictive of response measured at week 5 [11]. The present study examines the measurement of early response and serum levels as predictive of later therapeutic outcome in a homogenous population of clinically well-characterized inpatients who were treated with venlafaxine. In detail, we followed 3 assumptions. Firstly, we assumed an association between early response and eventual response. Secondly, we hypothesized that venlafaxine and antidepressant serum levels at week 2 of treatment are also associated with response. Thirdly, we assumed that the combination of both parameters explains more variance in response than each single parameter.
ogy after 2 weeks, indicated by a ≥ 20 % decrease on the HDRS. Full response has been defined as a reduction of ≥ 50 % on the HDRS. In the case of non-improvement after 4 weeks of treatment (indicated by < 25 % HDRS reduction), patients progressed to the next step of the algorithm. In the case of partial response (HDRS reduction between 25–50 %), the existing strategy was prolonged for a further 2 weeks. Therefore, the maximum length of time for antidepressant monotherapy was 6 weeks. Remission was defined as achieving a total score of < 10 on the HDRS. All raters were trained and experienced with the measure, and throughout the course of the study, inter-rater reliability was
Patients and Methods
Drop Out: N= 39 Missing TDM Blood Sample Or Results
▼
Patients ▶ Fig. 1 presents an overview of the study sample. 204 patients ● were initially screened for participation in this study as part of the third phase of the German Algorithm Project (GAP 3). The GAP is a standardized, stepwise treatment regimen that compares 3 different treatment algorithms and a computer-based treatment regime against treatment as usual across multiple psychiatric settings (6 academic and 3 non-academic psychiatric hospitals). Inclusion criteria for GAP 3 included a past DSM-IV defined major depressive episode and a Hamilton Depression Rating Scale (HDRS) score ≥ 15. Exclusion criteria were bipolar disorder, depression caused by other medical conditions, currently being pregnant or breast-feeding, or any prior antidepressant treatment that could not be discontinued. All intervention study arms in GAP 3 initially involved antidepressant monotherapy. Since venlafaxine was the selected antidepressant in almost 50 % of cases, the present study on response prediction ▶ Fig. 1). focused exclusively on venlafaxine-treated patients (● For 39 of the initial 204 patients (19.1 %), TDM blood samples or results were missing. For another 39 patients (19.1 %), venlafaxine titration was not conducted in accordance with the protocol, and in 38 patients (18.6 %), psychopathometric assessments were lacking, mainly due to drop out. This left a total of 88 participants (62 female, 26 male). Patients were uptitrated homogeneously: all were prescribed 75 mg/day venlafaxine on day 1 + 2, 150 mg/day on day 3 + 4 and 225 mg/day from day 5 onward. Mean age of participants was 45.39 years (SD = 13.02). On average, depression had first manifested at 34.25 years of age (SD = 14.11), with a sample average of 2.52 depressive episodes in total (SD = 2.89). All participants had venlafaxine/ODV plasma levels measured at week 2 and regular psychopathometric ratings measured for a maximum of the first 6 weeks from starting the study. Endpoints of this study were defined as change of treatment regime (including up- or down-titration), discharge from hospital, or achieving remission. Potentially important socio-demographic and disease-related sample characteristics ▶ Table 1. are presented in ●
Assessment of psychopathology Efficacy of treatment response was assessed using the Hamilton Depression Rating Scale (HDRS) at baseline and again every 2 weeks during antidepressant monotherapy. Patients were rated by independent study investigators not involved in the clinical management of patients and blind to group allocation. Early response has been defined as an improvement of psychopathol-
Included Sample: N= 165 VEN Treatment+ TDM Drop Out: N= 39 Diverging VEN Titration Included Sample: N= 126 VEN Treatment 225 mg/day+ TDM Drop Out: N= 38 Missing Psychopathometric Assessment Final Sample: N = 88 VEN Treatment 225 mg/day+ TDM + Psychopathometric Assessment
Fig. 1 Flow chart of the sample. VEN = venlafaxine.
Table 1 Clinical characteristics of the GAP 3 TDM sample. Criteria Diagnosis ICD-10 F32. ICD-10 F33. total Degree of severity (ICD-10 Criteria) slight moderate severe psychotic total Additional diagnosis alcohol dependency substance dependency dysthymia social phobia panic disorder somatisation disorder combined personality disorder eating disorder missing value no other disorder total Gender female male total Duration of episode 1 month < 3 months < 6 months < 1 year < 2 years to > 10 years unknown
n
%n
39 49 88
44.3 55.7 100
2 41 41 4 88
2.3 46.6 46.6 4.5 100
4 1 3 1 1 1 7 1 1 68 88
4.5 1.1 2.3 1.1 1.1 1.1 7.9 1.1 1.1 78.7 100
62 26 88
70.5 29.5 100
14 21 22 14 13 4
15.9 23.9 25.0 15.9 14.8 4.5
Stamm TJ et al. Prediction of Antidepressant Response … Pharmacopsychiatry 2014; 47: 174–179
This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
Initial Sample: N= 204 VEN Treatment
176 Original Paper
Therapeutic drug monitoring of venlafaxine Serum blood samples were obtained 14 days after commencing venlafaxine treatment across homogenous conditions (i. e., with fasting and withdraw in the morning before first medication intake). Venlafaxine and its active main metabolite O-desmethylvenlafaxine (ODV) were measured as previously described by Shams et al. [12]. Analyzing the influence of ODV serum level was an explorative process. To do so, we used a median split and contrasted a low and a high serum level group. Additional nonpsychiatric medication consumption was recorded on all accounts.
Statistical analysis Statistical analysis was conducted using SPSS Version 18. Frequencies were compared using the Chi-square (χ2) test. Normally distributed variables were compared using parametric t tests. Non-normally distributed values were compared using Mann-Whitney U and Kolomogorov-Smirnov tests. Binary logistic regression was used for analyzing the influence of the venlafaxine (VEN), ODV, and ODV/VEN-ratio serum level on achieving a therapeutic response (≥ 50 % on HDRS).
syndrome. More than half of the patients reported enduring the depressive episode for at least 6 months at the time of the study. 19 of the 88 patients had an additional psychiatric diagnosis. For further detailed information concerning the sample characteris▶ Table 1. tics, see ●
Early response as a predictor of therapeutic outcome In respect to response criteria, out of the 88 participants treated with venlafaxine 225 mg/day, 43 (51 %) achieved remission and 55 (63 %) showed a full response. Age, sex, length of current depressive episode and psychiatric comorbidity were not observed to significantly influence the therapeutic outcome. 37 of the later 43 remitted patients (86 %) had an early response on the HDRS after 2 weeks of treatment, while only 28 of the 45 non-remitted (62 %) showed an early response. This constitutes a significant difference in rates of early response (χ2 = 6 465; p = 0.011). On average, patients who showed an early response remitted 7.35 days earlier than patients without early signs of psychopathology improvement. Out of the 55 responding patients, 46 (84 %) showed an early response after 14 days. In the group of non-responders, only 14 of 33 (42 %) patients had an early response. Again, this constituted a significant difference in presentation of an early response for those who did vs. those who did not eventually respond to treatment (χ2 = 7 256; p = 0.007).
Results
Serum concentration of venlafaxine as a predictor of therapeutic outcome
Sample characteristics
Mean serum level was 121.1 for venlafaxine (SD = 143.8) and 240.2 (SD = 94.6) for ODV, resulting in a VEN + ODV serum level of 361.3 (SD = 162.7). Neither remission nor response was significantly influenced by plasma level of venlafaxine or the sum level of venlafaxine and ODV (all p-values > 0.05). ODV levels, however, were significantly higher in responders ▶ Fig. 2). compared to non-responders [t(86) = 2.831, p = 0.006] (● Patients who showed an ODV serum level above the median of 222 ng/mL were observed as significantly more likely to achieve response [χ2(1,N = 88) = 5.867; p = 0.015]. In regard to remission, such a finding could not be shown [χ2(1,N = 88) = 2.228; p = 0.135]. While ODV serum levels are uncorrelated with venlafaxine level (r = − 0.115; p = 0.01), overall serum level and ODV serum level were highly correlated (r = 0.817; p = 0.01). Co-medication was analyzed with regard to a possible influence ▶ Table 3). Serum levels of patients on the CYP2D6 system (● receiving co-medication deemed as having the potential to inhibit the CYP2D6 system did not differ from those who received no such medication.
▼
All included patients were diagnosed with a depressive episode according to DSM-IV. The mean HDRS score was 26.5 at baseline with 46.6 % of the study sample experiencing a severe depressive
ODV Serum level in ng/ml after 2 weeks of treatment
500 400 300 200 100 0
Yes
No Response
Fig. 2 Serum level of O-desmethylvenlafaxine after 2 weeks of treatment with 225 mg venlafaxine and response.
Combination of early improvement and ODV serum levels as outcome predictors A combination of early improvement (at least 20 % reduction on the HDRS) and a high level of ODV was observed to be predictive
Table 2 Results from stepwise binary logistic regression on early response and ODV serum level (*grouped) as predictive of eventual response to venlafaxine treatment. Regression
Wald
Variable
Coefficient B (SE)
Test
df
p
(OR)
Odds ratio
Lower
Upper
early response ODV serum level* constant
1.33 (0.51) 1.09 (0.48) − 2.04 (0.84)
6.84 5.19 5.90
1 1 1
0.009* 0.023* 0.015*
3.77 2.99 0.13
1 395 1 165
10 170 7 654
*p < 0.05; R2early response = 0.106; R2ODV Serum Level (grouped) = 0.181 (Nagelkerke)
Stamm TJ et al. Prediction of Antidepressant Response … Pharmacopsychiatry 2014; 47: 174–179
95 % C.l. for OR
This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
assessed across administrators at training sessions to ensure high quality ratings.
Original Paper 177 ▶ Table for response in a binary logistic model (all p’s < 0.05, see ● 2). Early response thereby explained 10.6 % of the variance (odds ratio = 3.77). The addition of ODV serum levels into the model in a second step saw an increase in the explained variance to 18.1 % (odds ratio = 2.99). Regarding the value as a screening instrument for treatment response, early response showed high sensitivity (84 %), but low specificity (42 %). Measurement of ODV revealed mid-range sensitivity (60 %) and specificity (70 %). The
combination of early response and TDM resulted in low sensitiv▶ Fig. 3 shows the distribuity (51 %), but high specificity (81 %). ● tion of the combined TDM and early response measurements in ▶ Table 2 for regression coefficients, responding patients. See ● significance values, effect sizes and all further statistics related to prediction of eventual response.
Discussion
Medication
n
%n
CYP2D6Inhibitors
36 52
40.9 59.1
8 5 3 32 18 6 3 3 2
9.1 5.7 3.4 36.4 20.5 6.8 3.4 3.4 2.3
4 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1
4.5 1.1 1.1 1.1 1.1 2.3 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1
30
No. of participants with Early Response
No. of participants with no Early Response
patients without comedication patients with comedication psychiatric medication antipsychotics olanzapine risperidone sedatives/hypnotics lorazepam zopiclone zolpidem chloral hydrate alprazolam other medication beta-blockers bisoprolol pindolol atenolol metoprolol tartrate amlodipine enalapril simvastatin atorvastatin omeprazol pantoprazol allopurinol metformin aspirin clopidogrel lactulose conjugated estrogenes ginkgo extract mistletoe extract
20
10
0
Early response In accordance with other studies and a recent meta-analysis on the predictive value of the early response assessment, we succeeded in showing a strong influence of early changes in psychopathology on later treatment success [8, 12, 13]. Our results support the hypothesis of an early onset for the medical treatment of depression, at least in venlafaxine-treated patients, as has previously been described elsewhere [12].
x x
x
TDM Regarding the results on TDM, we were able to demonstrate a strong relationship between clinical response and serum ODV, but not between response and venlafaxine or venlafaxine + ODV. Several factors may explain why we successfully detected a response effect with the active metabolite, but not with the active component. Firstly, research has indicated ODV to be more efficacious than venlafaxine: Desvenlafaxine, a novel salt form of isolated ODV has proven to be an effective compound in
Fig. 3 Distribution of combined TDM and early response measurements grouped by patient response success. Mean O-desmethylvenlafaxine serum level was grouped by median split.
30
20
10
0
Yes
No
x x
The results of this study underline the importance of early response assessment and therapeutic drug monitoring in patients with depression who are treated with venlafaxine. Having observed changes in psychopathometrics and TDM after 2 weeks of antidepressant treatment, we hereby propose a simple, early and effective instrument for predicting therapeutic outcome. Given the high sensitivity (84 %) for assessing early response alone, and a high specificity (81 %) when TDM is combined with assessment of early response, the results demonstrate the likelihood of successfully identifying non-responders early on in antidepressant treatment. The proposed technique of screening for non-responders might therefore be useful in guiding patients’ treatment strategies and recommendations in a timely manner.
No
Yes Response
Response ODV serum level ODV
Authors
T. J. Stamm1, D. Becker1, L. M. Sondergeld1, K. Wiethoff1, C. Hiemke2, G. O’Malley3, R. Ricken1, M. Bauer4, M. Adli1
Affiliations
Affiliation addresses are listed at the end of the article
Key words ▶ depression ● ▶ early response ● ▶ therapeutic drug monitoring ● ▶ treatment algorithm ●
Abstract
received revised accepted
27.09.2013 30.05.2014 06.06.2014
Bibliography DOI http://dx.doi.org/ 10.1055/s-0034-1383565 Pharmacopsychiatry 2014; 47: 174–179 © Georg Thieme Verlag KG Stuttgart · New York ISSN 0176-3679 Correspondence T. J. Stamm, MD Department of Psychiatry and Psychotherapy Charité Universitätsmedizin Berlin Campus Mitte Chariteplatz 1 10117 Berlin Germany Tel.: + 49/30/450 517009 Fax: + 49/30/450 517953 [email protected]
▼
Introduction: Early assessment of a therapeutic response is a central goal in antidepressant treatment. The present study examined the potential for therapeutic drug monitoring and symptom rating to predict venlafaxine treatment efficacy (measured by overall patient response and remission). Methods: 88 patients were uptitrated homogenously to 225 mg/day venlafaxine. Serum concentrations of venlafaxine (VEN) and its active metabolite O-desmethylvenlafaxine (ODV) were measured at week 2. Continuous psychopathometric ratings were measured for up to 6 weeks by independent study raters. Results: An early improvement was significantly more common in venlafaxine responders than
Introduction
▼
Despite the introduction of a variety of novel pharmacological agents as treatment options for antidepressant therapy, treatment resistance remains a common issue in the management of depressed patients. 30–40 % of depressed patients who are appropriately treated with an antidepressant substance do not respond. Several treatment algorithms for a rational and standardized treatment of major depressive disorders have been proposed to minimize the rate of treatment refractory depression [1–5]. Antidepressant monotherapy is usually the first line of treatment, however it remains that little is known about potential clinical or biological predictors for selecting the most promising substance. In general, monotherapy is recommended for at least 4–6 weeks until a change in the therapeutic strategy is warranted. This is the case despite growing evidence showing that nonresponse to an antidepressant treatment trial
Stamm TJ et al. Prediction of Antidepressant Response … Pharmacopsychiatry 2014; 47: 174–179
non-responders (χ2; p = 0.007). While ODV serum levels were significantly higher in responders (t test; p = 0.006), VEN serum levels, sum level of VEN + ODV and the ratio of ODV/VEN levels were not. Moreover, patients who showed an early response combined with an ODV serum level above the median of 222 ng/mL were significantly more likely to achieve full response (binary logistic model; p < 0.01). Sensitivity (84 % for early response) and specificity (81 % for combination of early response and therapeutic drug monitoring) were sufficient to qualify as a reasonable screening instrument. Conclusion: Our results indicate that early improvement and ODV serum concentration are predictive of therapeutic outcome and can thus be used to guide use of the antidepressant venlafaxine.
may be detectable at an earlier stage. Nakajima et al. [6] and Szegedi et al. [7, 8] respectively reported significant predictive values of an early response: they observed a reduction of 20 % in the psychopathometric evaluation of depressive symptomatology after 2 weeks to be significantly predictive of later achievement of full remission. The usefulness of therapeutic drug monitoring (TDM) for the prediction of treatment outcome has been discussed [9]. There seems to be no relationship between dose, serum level and clinical outcome in treatment with selective serotonine reuptake inhibitors (SSRI), while for tricyclic-based substances, a serum-level response relationship has been described somewhat contradictorily. For the serotonin-norepinephrine reuptake inhibitor (SNRI) venlafaxine, a relationship between dose, clinical outcome and a favourable response with higher summation serum levels has been shown [10]. However, several methodological constrictions mean the interpretability of these studies is limited. In a similar vein, in the case of citalopram,
This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
Prediction of Antidepressant Response to Venlafaxine by a Combination of Early Response Assessment and Therapeutic Drug Monitoring
Original Paper 175
serum concentrations on day 7 of treatment have been found to be predictive of response measured at week 5 [11]. The present study examines the measurement of early response and serum levels as predictive of later therapeutic outcome in a homogenous population of clinically well-characterized inpatients who were treated with venlafaxine. In detail, we followed 3 assumptions. Firstly, we assumed an association between early response and eventual response. Secondly, we hypothesized that venlafaxine and antidepressant serum levels at week 2 of treatment are also associated with response. Thirdly, we assumed that the combination of both parameters explains more variance in response than each single parameter.
ogy after 2 weeks, indicated by a ≥ 20 % decrease on the HDRS. Full response has been defined as a reduction of ≥ 50 % on the HDRS. In the case of non-improvement after 4 weeks of treatment (indicated by < 25 % HDRS reduction), patients progressed to the next step of the algorithm. In the case of partial response (HDRS reduction between 25–50 %), the existing strategy was prolonged for a further 2 weeks. Therefore, the maximum length of time for antidepressant monotherapy was 6 weeks. Remission was defined as achieving a total score of < 10 on the HDRS. All raters were trained and experienced with the measure, and throughout the course of the study, inter-rater reliability was
Patients and Methods
Drop Out: N= 39 Missing TDM Blood Sample Or Results
▼
Patients ▶ Fig. 1 presents an overview of the study sample. 204 patients ● were initially screened for participation in this study as part of the third phase of the German Algorithm Project (GAP 3). The GAP is a standardized, stepwise treatment regimen that compares 3 different treatment algorithms and a computer-based treatment regime against treatment as usual across multiple psychiatric settings (6 academic and 3 non-academic psychiatric hospitals). Inclusion criteria for GAP 3 included a past DSM-IV defined major depressive episode and a Hamilton Depression Rating Scale (HDRS) score ≥ 15. Exclusion criteria were bipolar disorder, depression caused by other medical conditions, currently being pregnant or breast-feeding, or any prior antidepressant treatment that could not be discontinued. All intervention study arms in GAP 3 initially involved antidepressant monotherapy. Since venlafaxine was the selected antidepressant in almost 50 % of cases, the present study on response prediction ▶ Fig. 1). focused exclusively on venlafaxine-treated patients (● For 39 of the initial 204 patients (19.1 %), TDM blood samples or results were missing. For another 39 patients (19.1 %), venlafaxine titration was not conducted in accordance with the protocol, and in 38 patients (18.6 %), psychopathometric assessments were lacking, mainly due to drop out. This left a total of 88 participants (62 female, 26 male). Patients were uptitrated homogeneously: all were prescribed 75 mg/day venlafaxine on day 1 + 2, 150 mg/day on day 3 + 4 and 225 mg/day from day 5 onward. Mean age of participants was 45.39 years (SD = 13.02). On average, depression had first manifested at 34.25 years of age (SD = 14.11), with a sample average of 2.52 depressive episodes in total (SD = 2.89). All participants had venlafaxine/ODV plasma levels measured at week 2 and regular psychopathometric ratings measured for a maximum of the first 6 weeks from starting the study. Endpoints of this study were defined as change of treatment regime (including up- or down-titration), discharge from hospital, or achieving remission. Potentially important socio-demographic and disease-related sample characteristics ▶ Table 1. are presented in ●
Assessment of psychopathology Efficacy of treatment response was assessed using the Hamilton Depression Rating Scale (HDRS) at baseline and again every 2 weeks during antidepressant monotherapy. Patients were rated by independent study investigators not involved in the clinical management of patients and blind to group allocation. Early response has been defined as an improvement of psychopathol-
Included Sample: N= 165 VEN Treatment+ TDM Drop Out: N= 39 Diverging VEN Titration Included Sample: N= 126 VEN Treatment 225 mg/day+ TDM Drop Out: N= 38 Missing Psychopathometric Assessment Final Sample: N = 88 VEN Treatment 225 mg/day+ TDM + Psychopathometric Assessment
Fig. 1 Flow chart of the sample. VEN = venlafaxine.
Table 1 Clinical characteristics of the GAP 3 TDM sample. Criteria Diagnosis ICD-10 F32. ICD-10 F33. total Degree of severity (ICD-10 Criteria) slight moderate severe psychotic total Additional diagnosis alcohol dependency substance dependency dysthymia social phobia panic disorder somatisation disorder combined personality disorder eating disorder missing value no other disorder total Gender female male total Duration of episode 1 month < 3 months < 6 months < 1 year < 2 years to > 10 years unknown
n
%n
39 49 88
44.3 55.7 100
2 41 41 4 88
2.3 46.6 46.6 4.5 100
4 1 3 1 1 1 7 1 1 68 88
4.5 1.1 2.3 1.1 1.1 1.1 7.9 1.1 1.1 78.7 100
62 26 88
70.5 29.5 100
14 21 22 14 13 4
15.9 23.9 25.0 15.9 14.8 4.5
Stamm TJ et al. Prediction of Antidepressant Response … Pharmacopsychiatry 2014; 47: 174–179
This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
Initial Sample: N= 204 VEN Treatment
176 Original Paper
Therapeutic drug monitoring of venlafaxine Serum blood samples were obtained 14 days after commencing venlafaxine treatment across homogenous conditions (i. e., with fasting and withdraw in the morning before first medication intake). Venlafaxine and its active main metabolite O-desmethylvenlafaxine (ODV) were measured as previously described by Shams et al. [12]. Analyzing the influence of ODV serum level was an explorative process. To do so, we used a median split and contrasted a low and a high serum level group. Additional nonpsychiatric medication consumption was recorded on all accounts.
Statistical analysis Statistical analysis was conducted using SPSS Version 18. Frequencies were compared using the Chi-square (χ2) test. Normally distributed variables were compared using parametric t tests. Non-normally distributed values were compared using Mann-Whitney U and Kolomogorov-Smirnov tests. Binary logistic regression was used for analyzing the influence of the venlafaxine (VEN), ODV, and ODV/VEN-ratio serum level on achieving a therapeutic response (≥ 50 % on HDRS).
syndrome. More than half of the patients reported enduring the depressive episode for at least 6 months at the time of the study. 19 of the 88 patients had an additional psychiatric diagnosis. For further detailed information concerning the sample characteris▶ Table 1. tics, see ●
Early response as a predictor of therapeutic outcome In respect to response criteria, out of the 88 participants treated with venlafaxine 225 mg/day, 43 (51 %) achieved remission and 55 (63 %) showed a full response. Age, sex, length of current depressive episode and psychiatric comorbidity were not observed to significantly influence the therapeutic outcome. 37 of the later 43 remitted patients (86 %) had an early response on the HDRS after 2 weeks of treatment, while only 28 of the 45 non-remitted (62 %) showed an early response. This constitutes a significant difference in rates of early response (χ2 = 6 465; p = 0.011). On average, patients who showed an early response remitted 7.35 days earlier than patients without early signs of psychopathology improvement. Out of the 55 responding patients, 46 (84 %) showed an early response after 14 days. In the group of non-responders, only 14 of 33 (42 %) patients had an early response. Again, this constituted a significant difference in presentation of an early response for those who did vs. those who did not eventually respond to treatment (χ2 = 7 256; p = 0.007).
Results
Serum concentration of venlafaxine as a predictor of therapeutic outcome
Sample characteristics
Mean serum level was 121.1 for venlafaxine (SD = 143.8) and 240.2 (SD = 94.6) for ODV, resulting in a VEN + ODV serum level of 361.3 (SD = 162.7). Neither remission nor response was significantly influenced by plasma level of venlafaxine or the sum level of venlafaxine and ODV (all p-values > 0.05). ODV levels, however, were significantly higher in responders ▶ Fig. 2). compared to non-responders [t(86) = 2.831, p = 0.006] (● Patients who showed an ODV serum level above the median of 222 ng/mL were observed as significantly more likely to achieve response [χ2(1,N = 88) = 5.867; p = 0.015]. In regard to remission, such a finding could not be shown [χ2(1,N = 88) = 2.228; p = 0.135]. While ODV serum levels are uncorrelated with venlafaxine level (r = − 0.115; p = 0.01), overall serum level and ODV serum level were highly correlated (r = 0.817; p = 0.01). Co-medication was analyzed with regard to a possible influence ▶ Table 3). Serum levels of patients on the CYP2D6 system (● receiving co-medication deemed as having the potential to inhibit the CYP2D6 system did not differ from those who received no such medication.
▼
All included patients were diagnosed with a depressive episode according to DSM-IV. The mean HDRS score was 26.5 at baseline with 46.6 % of the study sample experiencing a severe depressive
ODV Serum level in ng/ml after 2 weeks of treatment
500 400 300 200 100 0
Yes
No Response
Fig. 2 Serum level of O-desmethylvenlafaxine after 2 weeks of treatment with 225 mg venlafaxine and response.
Combination of early improvement and ODV serum levels as outcome predictors A combination of early improvement (at least 20 % reduction on the HDRS) and a high level of ODV was observed to be predictive
Table 2 Results from stepwise binary logistic regression on early response and ODV serum level (*grouped) as predictive of eventual response to venlafaxine treatment. Regression
Wald
Variable
Coefficient B (SE)
Test
df
p
(OR)
Odds ratio
Lower
Upper
early response ODV serum level* constant
1.33 (0.51) 1.09 (0.48) − 2.04 (0.84)
6.84 5.19 5.90
1 1 1
0.009* 0.023* 0.015*
3.77 2.99 0.13
1 395 1 165
10 170 7 654
*p < 0.05; R2early response = 0.106; R2ODV Serum Level (grouped) = 0.181 (Nagelkerke)
Stamm TJ et al. Prediction of Antidepressant Response … Pharmacopsychiatry 2014; 47: 174–179
95 % C.l. for OR
This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
assessed across administrators at training sessions to ensure high quality ratings.
Original Paper 177 ▶ Table for response in a binary logistic model (all p’s < 0.05, see ● 2). Early response thereby explained 10.6 % of the variance (odds ratio = 3.77). The addition of ODV serum levels into the model in a second step saw an increase in the explained variance to 18.1 % (odds ratio = 2.99). Regarding the value as a screening instrument for treatment response, early response showed high sensitivity (84 %), but low specificity (42 %). Measurement of ODV revealed mid-range sensitivity (60 %) and specificity (70 %). The
combination of early response and TDM resulted in low sensitiv▶ Fig. 3 shows the distribuity (51 %), but high specificity (81 %). ● tion of the combined TDM and early response measurements in ▶ Table 2 for regression coefficients, responding patients. See ● significance values, effect sizes and all further statistics related to prediction of eventual response.
Discussion
Medication
n
%n
CYP2D6Inhibitors
36 52
40.9 59.1
8 5 3 32 18 6 3 3 2
9.1 5.7 3.4 36.4 20.5 6.8 3.4 3.4 2.3
4 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1
4.5 1.1 1.1 1.1 1.1 2.3 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1
30
No. of participants with Early Response
No. of participants with no Early Response
patients without comedication patients with comedication psychiatric medication antipsychotics olanzapine risperidone sedatives/hypnotics lorazepam zopiclone zolpidem chloral hydrate alprazolam other medication beta-blockers bisoprolol pindolol atenolol metoprolol tartrate amlodipine enalapril simvastatin atorvastatin omeprazol pantoprazol allopurinol metformin aspirin clopidogrel lactulose conjugated estrogenes ginkgo extract mistletoe extract
20
10
0
Early response In accordance with other studies and a recent meta-analysis on the predictive value of the early response assessment, we succeeded in showing a strong influence of early changes in psychopathology on later treatment success [8, 12, 13]. Our results support the hypothesis of an early onset for the medical treatment of depression, at least in venlafaxine-treated patients, as has previously been described elsewhere [12].
x x
x
TDM Regarding the results on TDM, we were able to demonstrate a strong relationship between clinical response and serum ODV, but not between response and venlafaxine or venlafaxine + ODV. Several factors may explain why we successfully detected a response effect with the active metabolite, but not with the active component. Firstly, research has indicated ODV to be more efficacious than venlafaxine: Desvenlafaxine, a novel salt form of isolated ODV has proven to be an effective compound in
Fig. 3 Distribution of combined TDM and early response measurements grouped by patient response success. Mean O-desmethylvenlafaxine serum level was grouped by median split.
30
20
10
0
Yes
No
x x
The results of this study underline the importance of early response assessment and therapeutic drug monitoring in patients with depression who are treated with venlafaxine. Having observed changes in psychopathometrics and TDM after 2 weeks of antidepressant treatment, we hereby propose a simple, early and effective instrument for predicting therapeutic outcome. Given the high sensitivity (84 %) for assessing early response alone, and a high specificity (81 %) when TDM is combined with assessment of early response, the results demonstrate the likelihood of successfully identifying non-responders early on in antidepressant treatment. The proposed technique of screening for non-responders might therefore be useful in guiding patients’ treatment strategies and recommendations in a timely manner.
No
Yes Response
Response ODV serum level ODV
