Treatment response heterogeneity in the predominant negative symptoms of schizophrenia: Analysis of amisulpride vs placebo in three clinical trials

Treatment response heterogeneity in the predominant negative symptoms of schizophrenia: Analysis of amisulpride vs placebo in three clinical trials

Schizophrenia Research 156 (2014) 107–114 Contents lists available at ScienceDirect Schizophrenia Research journal homepage: www.elsevier.com/locate...

506KB Sizes 1 Downloads 33 Views

Schizophrenia Research 156 (2014) 107–114

Contents lists available at ScienceDirect

Schizophrenia Research journal homepage: www.elsevier.com/locate/schres

Treatment response heterogeneity in the predominant negative symptoms of schizophrenia: Analysis of amisulpride vs placebo in three clinical trials Stephen Z. Levine a,⁎, Stefan Leucht b a b

Department of Community Mental Health, University of Haifa, Israel Klinik und Poliklinik für Psychiatrie und Psychotherapie, Klinikum rechts der Isar, Technische Universität München, München, Germany

a r t i c l e

i n f o

Article history: Received 13 January 2014 Received in revised form 9 March 2014 Accepted 2 April 2014 Available online 29 April 2014 Keywords: Scale for the Assessment of Negative Symptoms Symptom response Trajectories Heterogeneity Treatment Negative symptoms

a b s t r a c t Background: The extent of heterogeneity in response to the psychopharmacological treatment of negative symptoms is unknown. Aims: To examine the extent of heterogeneity in response to the treatment of predominantly negative symptoms of schizophrenia. Method: Data were analyzed from three clinical trials that compared placebo or amisulpride for up to 60 days. Trial participants had predominantly negative symptoms of schizophrenia (n = 485). Heterogeneity of percentage reduction on the Scale for the Assessment of Negative Symptoms (SANS) was examined with trajectorygroup based modeling followed by descriptive statistics and the prediction of trajectory group membership with logistic regression modeling. Analyses were repeated separately for the placebo and amisulpride groups. Results: Trajectory group-based modeling identified groups of non- (n = 297, 61.2%), gradual–moderate (n = 135, 27.8%) and rapid- (n = 53, 10.9%) responders. At baseline compared to non-responders, rapid-responders had consistently significantly (p b .05) higher SANS total and subscale scores. Percent SANS improvement at endpoint was greatest for the rapid-responders group, a finding that replicated stratifying by placebo and amisulpride treatment groups. Similarly, in the total sample and stratifying by placebo and amisulpride groups, dropout was not significantly associated with trajectory group membership. Conclusions: Trajectories of treatment response to the psychopharmacological medication of the negative symptoms of schizophrenia demonstrate substantial heterogeneity. Approximately half of the patients included in our analysis showed little improvement, and the most severely ill at baseline responded the most. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Negative symptoms of schizophrenia (anhedonia, social withdrawal, affective flattening, and demotivation) are associated with deficits in cognitive, social and real-world functioning (Bowie et al., 2006; Harvey et al., 2006; Kirkpatrick et al., 2006). As the course of the disorder develops, negative symptoms persist in the long term to become more predominant than positive symptoms (Lieberman et al., 2001). Antipsychotics, however, are mainly effective for positive symptoms and the introduction of second-generation antipsychotics has not substantially improved this situation (Leucht et al., 2009). Therefore, there are currently intense efforts to develop psychopharmacological treatments of predominant negative symptoms. At least 15 psychopharmacological trials are currently underway to treat negative symptoms (Arango et al., 2013). A NIMH-MATRICS expert consensus group ⁎ Corresponding author at: Department of Community Mental Health, University of Haifa, Haifa 31905, Israel. Tel.: +972 524896083; fax: +972 3 7617374. E-mail address: [email protected] (S.Z. Levine).

http://dx.doi.org/10.1016/j.schres.2014.04.005 0920-9964/© 2014 Elsevier B.V. All rights reserved.

has been formed on negative symptoms with participants from academia, the Federal Drug Administration and industry (Alphs, 2006; Kirkpatrick and Fischer, 2006; Kirkpatrick et al., 2006; Marder et al., 2011), and conclusions of a recent consensus meeting reported (Marder et al., 2013). The NIMH-MATRICS consensus group has noted methodological and measurement challenges in clinical trials of negative symptoms (Kirkpatrick et al., 2006). This was elaborated on by recent directions on measurement (Daniel, 2013) backed by clinically meaningful thresholds on the SANS (Levine and Leucht, 2013). Despite these developments, little is known as to the extent of heterogeneity in negative symptoms. Reviews and epidemiological research, however, highlight that the course of schizophrenia is heterogeneous (Tandon et al., 2008; Levine et al., 2011). Examination of heterogeneity in symptoms rather than comparing aggregated medication groups may identify (a) groups of responders, and/or non-responders, and/or placebo-responders, (b) key periods of response, (c) the extent of response is for a minority or majority of patients and/or (d) that the inferior medication showed steady but mild response for most patients. To identify the extent of

108

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

heterogeneity in symptoms studies apply a statistical method known as group-based trajectory modeling (Jones and Nagin, 2007). Group-based trajectory modeling empirically identifies different groups of patients who assume similar response patterns over time, quantifies the extent of heterogeneity, and identifies key time points of change in symptom response. This method has been used to study depression (PelayoTerán et al., 2014) as well as schizophrenia. Re-analysis of randomized controlled trials of schizophrenia with trial participants selected on the basis of positive symptoms have shown that symptom response is heterogeneous and is typically captured by four or five symptom severity trajectory groups (Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010; Case et al., 2011; Levine et al., 2011; Marques et al., 2011; Stauffer et al., 2011; Levine et al., 2012). A summary of those trajectory studies of symptom response has estimated that approximately 16% of clinical trial participants assume a trajectory course of ‘dramatic response’ based on approximately a 60% to 80% symptom reduction (Levine et al., 2010). In sum, to extend the literature on heterogeneity in symptom response and on negative symptoms of schizophrenia, the current study aims to examine the extent of heterogeneity in negative symptom response in patients with predominantly negative symptoms who received amisulpride or placebo. 2. Methods 2.1. Participants and measures Patients (n = 487) were participants in three double-blind randomized placebo-controlled clinical trials that compared amisulpride with placebo for the treatment of predominant negative symptoms as defined in each study described below (Boyer et al., 1995; Loo et al., 1997; Danion et al., 1999). The trials used similar symptom selection criteria, randomized participants to placebo or amisulpride, had similar diagnostic groups (i.e., absence of early onset), used the Scale for the Assessment of Negative Symptoms (SANS) to assess negative symptoms (Andreasen, 1983) and had overlapping visit schedules. Complete trial documentation, such as the inclusion criteria and efficacy analyses, is reported in the primary study reports (Boyer et al., 1995; Loo et al., 1997; Danion et al., 1999). Across the three trials most participants were male (65.3% n = 318), and had a mean age of 34.04 (SD = 9.4). Each trial may be described as follows- (1) a multicenter trial, symptom inclusion thresholds of ≥75 SANS & ≤60 SAPS, diagnoses of disorganized, catatonic, undifferentiated and residual schizophrenia, and a 6-week washout phase (12-week if neuroleptics were received) followed by a six-week trial with randomization to placebo (n = 34) and or amisulpride (n = 70) conditions (Boyer et al., 1995); (2) a multicenter multinational trial, symptom inclusion thresholds of ≥ 60 SANS & ≤ 50 SAPS, diagnoses of residual schizophrenia, and a 4-week washout period followed by randomization for 12 weeks to amisulpride (n = 159) or placebo (n = 83) (Danion et al., 1999); and (3) a multicenter trial, symptom inclusion thresholds of ≥60 SANS & ≤50 SAPS, diagnoses of subchronic or chronic schizophrenia, and entering the trial directly for 24 weeks randomized to amisulpride (n = 69) or placebo (n = 72) (Loo et al., 1997). 2.2. Analytic approach First, like prior analyses, the percentage SANS reduction (B%) at each week was calculated using the formula B% = (B0 − Bi) ∗ 100 / B0, where B0 — SANS at baseline, Bi — SANS at week i (Leucht et al., 2005). This outcome was chosen since symptom response thresholds on the SANS reduction have been identified. Namely, research has shown that as ‘very much improved’ corresponds to a SANS percent changes of−90 to − 67, ‘much improved’ to − 50 to − 42, and ‘minimally improved’ to−21 to−13 (Levine and Leucht, 2013).

Second, to identify subgroups of patients with similar courses of treatment response, group-based trajectory analysis was computed (Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010). Using available information at each time-period, like mixed modeling (Lieberman et al., 2005), group-based trajectory analysis identifies subgroups of patients that are homogeneous on outcome criteria (i.e., SANS percent improvement) within the group, and significantly dissimilar (i.e., heterogeneous) from other subgroups in substantial ways (Haviland and Nagin, 2005). To identify the appropriate and most parsimonious number of subgroups the Bayes Information Criterion (BIC) is used. To illustrate treatment response at each week, SANS percentage improvement scores were plotted for the resultant treatment response trajectories. Analysis consisted of SANS percent improvement adjusted for baseline and treatment group as exposures, and accounted for non-random dropout (Nagin, 2005; Nagin and Odgers, 2010). Group-based trajectory analyses were computed for the aggregate (i.e., total) sample. These were implemented exactly as they had been in prior schizophrenia research (Levine and Leucht, 2010; Levine et al., 2010) except: that baseline severity and treatment group were accounted for, and the data based on this form of trajectory analysis are assumed to be not missing at random to account appropriately for systematic dropout (Haviland et al., 2011). At step three of the analysis the resultant empirically identified trajectory groups were compared at baseline with pairwise comparisons. At step four, resultant trajectory group membership was predicted using binary logistic modeling with sex, age, baseline SANS, and medication group as covariates in that order. At step five, the aforementioned steps two to five were re-computed stratifying by placebo, and amisulpride groups. At step six, the extent of trajectory group overlap was examined between the total, and stratified amisulpride and placebo samples.

3. Results The baseline trial characteristics are presented in Table 1. Participants (n = 2) from two trials (Boyer et al., 1995; Loo et al., 1997) were removed due to missing baseline SANS scores leaving 485 participants with available scores. Next group-based trajectory modeling was computed. In the total sample, examination of the SANS total percent reduction BICs identified a three trajectory group model (Trajectories and BICs were: 1 = − 6264.90, 2 = − 6058.58, 3 = − 6037.37 and 4 = − 5886.85, respectively). Posterior probabilities of correct classification were high (.93, .83 and .92). Trajectory group 1 was labeled ‘non-responders’ (n = 297, 61.2%), trajectory

Table 1 Baseline characteristics across trials.

Male Female Age SANS Affect Alogia Anhedonia Attention Apathy Total Randomization Amisulpride doses (100 mg) (300 mg) (50 mg) Placebo

M/n

SD/%

318 169 34.0

(65.3) (34.7) (9.4)

23.7 14.2 19.6 8.2 13.8 79.5

(5.9) (3.6) (3.2) (3.0) (3.0) (13.4)

178 36 84 189

Note. Baseline characteristics of the current study pooled clinical trials.

(36.6) (7.4) (17.2) (38.8)

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

group 2 was labeled ‘gradual-moderate responders’ (n = 135, 27.8%), and trajectory group 3 was labeled ‘rapid-responders’ (n = 53, 10.9%) (see Fig. 1). Based on pairwise comparisons, at baseline compared to non-responders, rapid-responders had consistently higher scores on the baseline SANS total and all subscales (Table 2). On outcome, SANS percent improvement using last-value carried forward was greatest for rapid-responders, followed by gradualmoderate responders followed by non-responders across all scales. Trajectory groups did not statistically significantly differ on dropout (Table 2). Binary logistic modeling showed that prediction of trajectory group membership was significantly (p b .05) associated with treatment group. Placebo predicted membership in the trajectory group of non-responders, whereas amisulpride predicted membership in the trajectory group of gradual-moderate responders (Table 3). Higher baseline SANS total scores significantly (p b .05) predicted assuming a rapid-responder trajectory (Table 3). 3.1. Stratified analyses Next analyses were stratified by treatment group. First, trial participants randomized to amisulpride were analyzed. Trajectory analysis of their SANS total percent reduction scores showed BICs identified a three trajectory group model was a good fit to the data (Trajectories and BICs were: 1 = − 3875.80, 2 = − 3704.85, 3 = − 3657.65, 4 = −3662.97, respectively). Posterior probabilities of correct classification were high (.94, .90, and .96). Corresponding with Fig. 2 the trajectory groups were labeled 1 ‘non-responders’ (N = 163, 54.9%), 2 ‘gradualmoderate responders’ (N = 97, 32.7%), and 3 ‘rapid responders’ (N = 37, 12.5%). Based on pairwise comparisons, at baseline compared to non-responders, rapid responders had higher scores on the SANS total and subscales except Avolition-apathy (Table 4). On outcome percent SANS improvement with last value carried forward was greatest for rapid-responders, followed by the gradual-moderate responders and then the non-responders across all scales. Trajectory groups did not statistically significantly differ on dropout (Table 4). Higher baseline SANS total scores significantly (p b .05) predicted assuming a rapid responder trajectory (Table 4). Second, trial participants randomized to placebo were analyzed. Trajectory analysis of their SANS total percent reduction scores showed that a two trajectory model was a good fit to the data (Trajectory group and BICs were: 1 = −2374.63, 2 = −2341.07, 3 = −2365.85, respectively). Correct classifications based on posterior probabilities were high (.96 and .91). Based on the trajectory groups in Fig. 2, groups

109

were labeled as 1) non-responders (n = 156, 83%), 2) and 3) placeboresponders (n = 32, 17%). Baseline characteristics did not differ significantly between placebo trajectory groups (Table 4). Compared to nonresponders, placebo-responders had higher SANS total and subscales improvement scores only (Table 5). 3.2. Trajectory group overlap The overlap between the total and stratified trajectory groups was examined. Of the amisulpride non-responder trajectory 95.1% (n = 155) were classified as belonging to the total trajectory group of nonresponders. A total of 4.9% (n = 8) of amisulpride non-responders were classified in the gradual-moderate group. Gradual-moderate (N = 97) and rapid-responder (N = 37) trajectory groups were 100% concordant with the total groups of the same name. These results indicate a strong statistically significant association between the total and stratified amisulpride trajectory groups (Kendall's τ = .96, p b .01). In the placebo non-responder trajectory group 91.0% (n = 142) were classified as non-responders in the total trajectory group. However, 9.0% (n = 14) of the placebo non-responders were classified in the moderate-gradual responder trajectory group of the aggregate group analysis. Of the placebo-responder trajectory group 50% (n = 16) were classified as gradual-moderate responders and 50% (n = 16) as rapid-responders in the total-sample analysis. This trend was statistically significant indicating moderately strong overlap between the total and stratified placebo trajectory groups (Kendall's τ = .81, p b .01). 4. Discussion Based on the reanalysis of three large clinical trials of psychopharmacological treatment of predominantly negative symptoms of schizophrenia the current results empirically identified the extent of heterogeneity in treatment response. Specifically, trajectory groups were empirically that consisted of non-responders, gradual-moderate responders and rapid-responders. On aggregate and replicating in stratified analyses, the non-responder groups consisted of the majority of participants (aggregate: 61.2%; placebo: 83%; amisulpride: 54.9%), and yielded the lowest SANS percent total reductions by the last visit (aggregate: 13.2%; placebo: 11.5%; amisulpride: 16.3%). Unlike the nonresponder groups, across aggregate and stratified analyses the rapidresponder groups consisted of the least participants (aggregate: 10.9%; amisulpride: 12.5%; placebo: 17%) and yielded the largest SANS percent total reductions by the last visit (aggregate: 75.7%; placebo: 64.12;

SANS % improvement 90.00 80.00 Rapid-responders 70.00 60.00 50.00

Gradual-moderate responders

40.00 30.00 20.00

Non-responders

10.00 0.00 8-14

26-30

56-60

Days Fig. 1. Symptom response trajectories of the total sample. Note. Trajectory groups with 95% confidence intervals. 1) Non-responders (n = 297, 61.2%), 2) gradual–moderate responders (n = 135, 27.8%) 3) rapid-responders (n = 53, 10.9%).

110

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

Table 2 Trajectory group membership at baseline and outcome characteristics across trials. Trajectory groups:

SANS total Affective flattening Alogia Anhedonia–asociality Attentional impairment Avolition-apathy Age Male Female Placebo Amisulpride (aggregate) Dosing percent by dose 50 mg (n, %) 100 mg (n, %) 300 mg (n, %)

Significant (p b .05) pairwise

3 Rapid-responders (53, 10.9%)

2 Gradual-moderate responders (135, 27.8%)

1 Non-responders (297, 61.2%) M/N

(SE/%)

M/N

(SE/%)

M/N

(SE/%)

Comparisons

79.01 23.63 14.31 19.23 8.27 13.57 34.67 193 104 142 155

(.75) (.34) (.20) (.18) (.17) (.18) (.55) (64.98) (35.02) (47.81) (52.19)

78.21 23.10 13.53 19.87 7.73 13.98 33.25 91 44 30 105

(1.08) (.48) (.29) (.26) (.26) (.23) (.81) (67.41) (32.59) (22.22) (77.78)

85.51 25.87 15.66 20.58 9.04 14.36 32.72 33 20 16 37

(2.29) (.85) (.62) (.45) (.47) (.40) (1.21) (62.26) (37.74) (30.19) (69.81)

3N1 3N1 3 N 1, 1 N 2 2&3N 1 2b3 NS NS NS P b .05

57 85 13

(67.86) (47.75) (37.14)

21 72 12

(25.00) (40.45) (34.29)

6 21 10

(7.14) (11.80) (28.57)

NS P b .05 P b .05

Percent change at endpoint

M/N

(SE/%)

M/N

(SE/%)

M/N

(SE/%)

Comparisons

SANS total Affective flattening Alogia Anhedonia-asociality Attentional impairment Avolition-apathy Dropout Completer

9.16 8.12 8.78 5.29 6.79 9.76 123 174

(1.36) (3.37) (1.44) (1.16) (2.66) (1.62) (41.41) (58.59)

(1.42) (1.94) (1.89) (1.71) (2.88) (1.89) (36.30) (63.70)

74.47 79.97 71.09 63.91 83.28 76.38 18 35

(1.91) (2.46) (2.79) (3.29) (3.21) (3.01) (33.96) (66.04)

3 3 3 3 3 3

Outcomes

43.48 48.84 44.52 33.88 53.59 42.94 49 86

N N N N N N

2 & 1, 1 2 & 1, 1 2 & 1, 1 2 & 1, 1 2 & 1, 1 2 & 1, 1

N N N N N N

2 2 2 2 2 2

NS

Note. Trajectory numbers correspond with Fig. 1. The upper section of this table depicts the baseline features associated with each of the empirically identified trajectory-groups. The lower section of this Table depicts outcomes associated with each of the empirically identified trajectory-groups. All statistics for the pairwise comparisons are shown comparisons are shown in the online supplement.

amisulpride: 74.8%). At baseline compared to non-responders, rapidresponders had consistently higher baseline SANS total and subscales scores. SANS percent improvement based on last value carried forward was greatest for rapid-responders group. This finding replicated stratifying the placebo and amisulpride treatment groups. Similarly, in the total sample and stratifying for placebo and amisulpride groups, dropout was not significantly associated with trajectory group membership. Comparison with prior trajectory analyses is complex, since unlike prior analysis the current study uses predominant negative symptom selection criteria (Levine and Leucht, 2010; Levine and Rabinowitz, 2010; Levine et al., 2010; Case et al., 2011; Levine et al., 2011;

Marques et al., 2011; Stauffer et al., 2011; Levine et al., 2012). Nonetheless, unlike prior research, in the current study dropout was not significantly associated with trajectory group membership. Aggregate analysis in positive symptom studies demonstrates that dropout is systematically associated with lack of response (Rabinowitz and Davidov, 2008), a criticism that applies to standard trajectory analyses (Rabinowitz, 2009). Unlike other methods, however, the current trajectory analysis accounts for non-random attrition (Haviland et al., 2011). Second, like prior analyses, assuming a better response trajectory was associated with higher baseline scores, and greater improvement on endpoint severity scores. The result that the more severely ill patients were

Table 3 Examination of trajectory group membership with binary logistic regression. Trajectory group

Variables in each binary logistic regression modela

Beta

Standard error

P-value

Odds ratiob

95% confidence interval range

1) Non-responders

Sex Age Baseline SANS Treat Intercept Sex Age SANS total baseline Treat Intercept Sex Age Baseline SANS Treat Intercept

0.07 0.02 −0.01 −1.04 0.95 −0.20 −0.01 −0.01 1.09 −0.02 0.22 −0.02 0.04 0.39 −5.06

0.20 0.01 0.01 0.21 0.76 0.22 0.01 0.01 0.24 0.83 0.31 0.02 0.01 0.32 1.18

0.72 0.12 0.41 0.00 0.21 0.37 0.33 0.11 0.00 0.98 0.48 0.32 0.00 0.22 0.00

1.08 1.02 0.99 0.36 2.59 0.82 0.99 0.99 2.97 0.98 1.24 0.98 1.04 1.47 0.01

0.72 1.00 0.98 0.24

1.60 1.04 1.01 0.53

0.53 0.97 0.97 1.87

1.26 1.01 1.00 4.70

0.68 0.95 1.02 0.79

2.27 1.02 1.06 2.76

2) Gradual-moderate-responders

3) Rapid-responders

Note. a Order of entry, sex, age, baseline SANS, medication group. b Underlined odds ratios reflect statistically significant (p b .05) values.

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

111

SANS % improvement 90.00 80.00

Rapid-responders

70.00 60.00 Gradual-moderate responders

50.00 40.00 30.00 20.00

Non-responders

10.00 0.00 8-14

26-30

56-60

Days SANS % improvement 80.00 70.00 Placebo-responders

60.00 50.00 40.00 30.00 20.00

Non-responders

10.00 0.00 -10.00 8-14

26-30

56-60

Days Fig. 2. Stratified symptom response trajectories by psychopharmacological medication group. Note. Amisulpride trajectory groups 1) non-responders (N = 163, 54.9%), 2) gradual– moderate responders (N = 97, 32.7%), and 3) rapid responders (N = 37, 12.5%), placebo trajectory groups: 1 non-responders (n = 156, 83%), and 2 placebo-responders (n = 32, 17%).

associated with the more responsive trajectory groups may be in part simply explained by more room for improvement. Making a simple comparison with internal medicine, patients with very high cholesterol levels will benefit more from statins than those with little elevation. The results also raise the law of initial value hypothesis that states that “the higher the initial value, the greater the organism's response” and would therefore explain the greater symptom reductions among the severer patients (Wilder, 1957). The reason for this law may be a statistical artifact, such as regression to the mean, but it could also be an actual clinical manifestation, because similar observations have been made in other psychiatric disorders. For instance, pooled analysis of two new generation antidepressants (fluoxetine and venlafaxine) has reported this effect in depression (Fournier et al., 2010). Accordingly, the law of initial value was supported for predominant negative symptoms, as it has been in, for example, depression (Kirsch et al., 2008). This implies that that high baseline symptom severity appears to be a study-design consideration for future clinical trials of predominantly negative symptoms. The clinical implication is that it is appropriate that doctors be aware that the most severely ill patients will respond best, which may appear counterintuitive. For mildly ill patients, more care by doctors appears to be appropriate in prescribing medication, because the risk–benefit balance may be worse due to the same amount of side-effects balanced by less efficacy. The current trajectory analysis is also relevant for the debate as to whether the onset of action of antipsychotics is “delayed” or “early”.

The “delayed-onset” hypothesis posits that response attains its therapeutic level a “delay” of two to three weeks occurs prior to the onset of treatment response. This hypothesis was tested in the treatment of positive symptoms of schizophrenia and subsequently rejected by meta-analysis (Agid et al., 2003), and the rejection was confirmed by large-scale individual-level analysis (Leucht et al., 2005). A prior analysis did not clearly identify an early responder effect comparing placebo and amisulpride in these data (Levine and Leucht, 2012). Accordingly, by going beyond the placebo-treatment divide, the current results show that a select sub-group of approximately 10% of patients assumed a trajectory of rapid response and 28% of gradual-moderate response. This suggests that we did also not observe evidence for the delayed onset of antipsychotic drug action hypothesis that posited an initial delay in response is (e.g., at one or two weeks) followed (e.g., at three or four weeks) by symptomatic reduction, unless we speculate that the large group of non-responders would have responded at a later stage after the study analyses ended (at 56–60 days). 4.1. Limitations and conclusions There are several limitations to the current study. First, we were unable to examine between trial differences due to sample size considerations. Second, due to the trial selection criteria (i.e., high negative SANS and low positive SAPS scores) the current study included patients with predominantly negative symptoms who may not be representative of

112

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

Table 4 Trajectory group membership at baseline and outcome characteristics across trials. Amisulpride

Placebo

1) Non-responders (N = 163, 54.9%)

2) Gradualmoderate responders (N = 97, 32.7%)

3) Rapid responders (N = 37, 12.5%)

Significant (p b .05) pairwise

1) Non-responders (n = 156, 83%)

2) Placeboresponders (n = 32, 17%)

Significant (p b .05) pairwise

Comparisons

M/N

(SE/%)

M/N

(SE/%)

Comparisons

3N 3N 3N 3N NS NS NS

78.12 22.92 13.90 19.41 8.29 13.60 34.99 104 52

(1.11) (.54) (.30) (.25) (.24) (.25) (.77) (66.67) (33.33)

81.69 24.91 14.28 20.41 7.72 14.38 32.72 24 8

(2.48) (1.03) (.73) (.53) (.60) (.51) (1.77) (75.00) (25.00)

NS NS NS NS NS NS NS NS

6.67 1.73 6.21 4.27 9.04 8.61 87

(2.34) (6.20) (2.25) (1.83) (3.60) (2.32) (55.77)

64.94 67.46 65.88 56.41 76.50 64.44 17

(3.12) (3.83) (4.40) (4.76) (4.71) (3.99) (53.13)

2N 1 2N 1 2N 1 2N 1 2N 1 2N 1 NS

M/N

(SE/%)

M/N

(SE/%)

M/N

(SE/%)

SANS total Affective flattening Alogia Anhedonia–asociality Attentional impairment Avolition–apathy Age Male Female 50 mg (% per dose) 100 mg 300 mg

79.67 24.05 14.55 19.20 8.24 13.63 34.41 105 58 57 93 13

(.94) (.41) (.23) (.25) (.22) (.23) (.75) (64.42) (35.58) (67.86) (52.25) (37.14)

78.31 23.43 13.70 19.67 7.69 13.81 32.70 61 36 21 64 12

(1.22) (.49) (.35) (.31) (.31) (.27) (.88) (62.89) (37.11) (25.00) (35.96) (34.29)

85.76 25.46 15.68 20.70 9.43 14.49 33.35 23 14 6 21 10

(2.95) (1.12) (.76) (.58) (.52) (.50) (1.49) (62.16) (37.84) (7.14) (11.80) (28.57)

Outcomes SANS total Affective flattening Alogia Anhedonia–asociality Attentional impairment Avolition–apathy Dropout

13.97 17.33 14.61 8.20 7.68 13.16 50

(1.19) (1.63) (1.65) (1.30) (3.57) (2.06) (30.67)

45.61 51.83 45.23 35.38 57.09 45.18 28

(1.51) (2.02) (2.19) (1.89) (3.32) (2.15) (28.87)

73.73 80.42 68.99 62.17 82.17 76.18 8

(2.27) (3.04) (3.27) (3.83) (3.58) (3.81) (21.62)

2 2 1 2,1

P b .05 NS P b .05 3N 3N 3N 3N 3N 3N NS

2 & 1, 1 2 & 1, 1 2 & 1, 1 2 & 1, 1 2 & 1, 1 2 & 1, 1

N N N N N N

2 2 2 2 2 2

Note. Trajectory numbers correspond with Fig. 2. The upper section of this Table depicts the baseline features associated with each of the empirically identified trajectory-groups. The lower section of this Table depicts outcomes associated with each of the empirically identified trajectory-groups. All statistics for the pairwise comparisons are shown comparisons are shown in the online supplement. SE: Standard error. M: Mean. N: Sample size.

those who receive routine clinical care. For instance, patients in our database had signed informed consent, thus they had to be relatively cooperative and not extremely thought disordered. Suicidal patients were excluded from the trials. So were patients with additional substance abuse that is a frequent problem in some settings, patients with significant nonpsychiatric illnesses and patients younger than 18 or older than 65. Also, only one antipsychotic (amisulpride) was compared to placebo, yet it would be more clinically relevant to compare across standard antipsychotics used for schizophrenia. Nevertheless, to the best of our knowledge, few studies have addressed the effects of other antipsychotics than amisulpride in patients with predominant negative symptoms (Leucht et al., 2009). Third, the trials examined in the current study provide a sufficient duration to examine the short-term course of response (to 60 days). The study visit schedules, however, are not designed to assess response trajectories. This is relevant since the percent SANS reduction depends on the time gap between visits that varied across and within trials. Also, the duration may be too short to determine change in negative symptoms. Different results (possibly with more responders) may have emerged had different schedules and a longer follow-up period been available. Analysis of dose at baseline highlighted an effect of greater dosing with improved trajectory membership. Namely, of those receiving 300 mg amisulpride, 28.57% assumed the trajectory of rapid-response, a figure that fell to 11.8% on 100 mg amisulpride and 7.14% on 50 mg amisulpride. Modest weighting may be given to this finding, however, since the sample size in 5 of 9 cells was under 22. Accordingly, we were unable to include dose in the logistic regression models. This suggests that a dose of 300 mg was a marker rather than necessarily a predictor of rapid-response. The current study examined amisulpride since there are no largescale clozapine studies in patients with predominant negative symptoms

with a well-defined negative symptom measure. Future research should examine heterogeneity of clozapine in the treatment of predominant negative symptoms. This heterogeneity may explain the mixed findings regarding the efficacy of clozapine identified by systematic reviews of the treatment of negative symptoms (Murphy et al., 2006; Arango et al., 2013). The available data do not contain duration of untreated psychosis (Wyatt, 1991; Birchwood et al., 1998). This is a limitation and highlights a key direction for future study. It is relevant since longer durations of untreated psychosis have been shown to have prognostic value in first-episode cohorts (Marshall et al., 2005; Perkins et al., 2005), more negative symptoms (Compton et al., 2011), and heterogeneity in response of positive symptoms (Levine et al., 2010). Also research has suggested that reduction in the duration of untreated psychosis may affect the core pathobiological deficit process through prevention of negative symptom escalation (Melle et al., 2008). In conclusion, based on three randomized double blind clinical trials of predominant negative symptoms in schizophrenia, the current results empirically identified three groups as non-responders, gradualmoderate responders and rapid responders. These group characterizations: were consistent after stratification by psychopharmacological medication, showed distinct symptom severity and response profiles and were unrelated to dropout. This suggests that treatment response to the psychopharmacological medication of the negative symptoms of schizophrenia demonstrates substantial heterogeneity. From a clinical point of view, the current study results confirm that negative symptoms are difficult to treat: the largest proportion of patients fell in the non-responder group who comprised 61.2% of the total sample on the average yielded only a 13.2% SANS total score reduction by the last visit. Alternatively, the results also provide a more optimistic perspective for clinicians in that they can expect that their most severely ill patients will benefit most from treatment.

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

113

Table 5 Examination of trajectory group membership with binary logistic regression stratified by medication group. Amisulpride

Variables in each binary logistic regression modela

Beta

Standard error

P-value

Odds ratiob

95% confidence interval range

1) Non-responders

Sex Age Baseline SANS Intercept Sex Age Baseline SANS Intercept Sex Age Baseline SANS Intercept

−0.08 0.02 −0.01 0.12 0.02 −0.02 −0.01 1.00 0.15 −0.01 0.04 −5.00

0.24 0.01 0.01 0.92 0.26 0.01 0.01 1.00 0.37 0.02 0.01 1.41

0.74 0.15 0.54 0.90 0.94 0.19 0.15 0.32 0.69 0.78 0.00 0.00

0.92 1.02 1.00 1.12 1.02 0.98 0.99 2.73 1.16 1.00 1.04 0.01

0.57 0.99 0.98

1.49 1.05 1.01

0.61 0.96 0.97

1.69 1.01 1.01

0.56 0.96 1.01

2.40 1.03 1.06

Sex Age Baseline SANS Intercept Sex Age Baseline SANS Intercept

0.32 0.02 −0.02 1.79 −0.32 −0.02 0.02 −1.79

0.45 0.02 0.01 1.49 0.45 0.02 0.01 1.49

0.47 0.31 0.23 0.23 0.47 0.31 0.23 0.23

1.38 1.02 0.98 5.97 0.72 0.98 1.02 0.17

0.57 0.98 0.96

3.33 1.07 1.01 1.74 1.02 1.05

2) Gradual-moderate responders

3) Rapid-responders

Placebo sample 1) Non-responders

2) Placebo-responders

0.30 0.94 0.99

Note. a Order of entry, sex, age, baseline SANS, medication group. b Underlined odds ratios reflect statistically significant (p b .05) values. Funding The current study was not funded. Contributors Levine analyzed and interpreted the data and drafted the manuscript. Leucht interpreted the data and provided critical manuscript feedback. Conflict of Interest In the last three years Stefan Leucht has received honoraria for lectures from Abbvie, Astra Zeneca, BristolMyersSquibb, ICON, EliLilly, Janssen, Johnson & Johnson, Roche, SanofiAventis, Lundbeck and Pfizer; for consulting/advisory boards from Roche, EliLilly, Medavante, BristolMyersSquibb, Alkermes, Janssen, Johnson & Johnson and Lundbeck. EliLilly has provided medication for a study with Stefan Leucht as primary investigator. Stephen Z Levine has received research support, and/or consultancy fees and/or travel support from Hoffmann-La Roche, Shire Pharmaceuticals, Eli Lilly, the Claims Conference, Maccabi Health Insurance and the Israel Science Foundation. Acknowledgments The authors acknowledge SanofiAventis with thanks for providing the data for this manuscript.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.schres.2014.04.005. References Agid, O., Kapur, S., Arenovich, T., Zipursky, R.B., 2003. Delayed-onset hypothesis of antipsychotic action: a hypothesis tested and rejected. Arch. Gen. Psychiatry 60 (12), 1228–1235. Alphs, L., 2006. An industry perspective on the NIMH consensus statement on negative symptoms. Schizophr. Bull. 32 (2), 225–230. Andreasen, N.C., 1983. Scale for the Assessment of Negative Symptoms. University of Iowa, Iowa City. Arango, C., Garibaldi, G., Marder, S.R., 2013. Pharmacological approaches to treating negative symptoms: a review of clinical trials. Schizophr. Res. 150 (2–3), 346–352. Birchwood, M., Todd, P., Jackson, C., 1998. Early intervention in psychosis. The critical period hypothesis. Br. J. Psychiatry Suppl. 172 (33), 53–59. Bowie, C.R., Reichenberg, A., Patterson, T.L., Heaton, R.K., Harvey, P.D., 2006. Determinants of real-world functional performance in schizophrenia subjects: Correlations with cognition, functional capacity, and symptoms. Am. J. Psychiatry 163 (3), 418–425. Boyer, P., Lecrubier, Y., Puech, A.J., Dewailly, J., Aubin, F., 1995. Treatment of negative symptoms in schizophrenia with amisulpride. Br. J. Psychiatry 166 (1), 68–72. Case, M., Stauffer, V.L., Ascher-Svanum, H., Conley, R., Kapur, S., Kane, J.M., KollackWalker, S., Jacob, J., Kinon, B.J., 2011. The heterogeneity of antipsychotic response in the treatment of schizophrenia. Psychol. Med. 41 (6), 1291–1300.

Compton, M.T., Gordon, T.L., Goulding, S.M., Esterberg, M.L., Carter, T., Leiner, A.S., Weiss, P.S., Druss, B.G., Walker, E.F., Kaslow, N.J., 2011. Patient-level predictors and clinical correlates of duration of untreated psychosis among hospitalized first-episode patients. J. Clin. Psychiatry 72 (2), 225–232. Daniel, D.G., 2013. Issues in selection of instruments to measure negative symptoms. Schizophr. Res. 150 (2–3), 343–345. Danion, J.M., Rein, W., Fleurot, O., 1999. Improvement of schizophrenic patients with primary negative symptoms treated with amisulpride. Amisulpride study group. Am. J. Psychiatry 156 (4), 610–616. Fournier, J.C., DeRubeis, R.J., Hollon, S.D., Dimidjian, S., Amsterdam, J.D., Shelton, R.C., Fawcett, J., 2010. Antidepressant drug effects and depression severity: a patientlevel meta-analysis. JAMA 303 (1), 47–53. Harvey, P.D., Koren, D., Reichenberg, A., Bowie, C.R., 2006. Negative symptoms and cognitive deficits: what is the nature of their relationship? Schizophr. Bull. 32 (2), 250–258. Haviland, A.M., Nagin, D.S., 2005. Causal inferences with group based trajectory models. Psychometrika 70 (3), 557–578. Haviland, A.M., Jones, B.L., Nagin, D.S., 2011. Group-based trajectory modeling extended to account for nonrandom participant attrition. Sociol. Methods Res. 40 (2), 367–390. Jones, B.L., Nagin, D.S., 2007. Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociol. Methods Res. 35 (4), 542–571. Kirkpatrick, B., Fischer, B., 2006. Subdomains within the negative symptoms of schizophrenia: commentary. Schizophr. Bull. 32 (2), 246–249. Kirkpatrick, B., Fenton, W.S., Carpenter Jr., W.T., Marder, S.R., 2006. The NIMH-MATRICS consensus statement on negative symptoms. Schizophr. Bull. 32 (2), 214–219. Kirsch, I., Deacon, B.J., Huedo-Medina, T.B., Scoboria, A., Moore, T.J., Johnson, B.T., 2008. Initial severity and antidepressant benefits: a meta-analysis of data submitted to the food and drug administration. PLoS Med. 5 (2), e45. Leucht, S., Busch, R., Hamann, J., Kissling, W., Kane, J.M., 2005. Early-onset hypothesis of antipsychotic drug action: a hypothesis tested, confirmed and extended. Biol. Psychiatry 57 (12), 1543–1549. Leucht, S., Corves, C., Arbter, D., Engel, R.R., Li, C., Davis, J.M., 2009. Second-generation versus first-generation antipsychotic drugs for schizophrenia: a meta-analysis. Lancet 373 (9657), 31–41. Levine, S.Z., Leucht, S., 2010. Elaboration on the early-onset hypothesis of antipsychotic drug action: treatment response trajectories. Biol. Psychiatry 68 (1), 86–92. Levine, S.Z., Leucht, S., 2012. Delayed- and early-onset hypotheses of antipsychotic drug action in the negative symptoms of schizophrenia. Eur. Neuropsychopharmacol. 22 (11), 812–817. Levine, S.Z., Leucht, S., 2013. Identifying clinically meaningful symptom response cut-off values on the sans in predominant negative symptoms. Schizophr. Res. 145 (1–3), 125–127. Levine, S.Z., Rabinowitz, J., 2010. Trajectories and antecedents of treatment response over time in early-episode psychosis. Schizophr. Bull. 36 (3), 624–632. Levine, S.Z., Rabinowitz, J., Case, M., Ascher-Svanum, H., 2010. Treatment response trajectories and their antecedents in recent-onset psychosis: a 2-year prospective study. J. Clin. Psychopharmacol. 30 (4), 446–449. Levine, S.Z., Lurie, I., Kohn, R., Levav, I., 2011. Trajectories of the course of schizophrenia: from progressive deterioration to amelioration over three decades. Schizophr. Res. 126 (1–3), 184–191. Levine, S.Z., Rabinowitz, J., Faries, D., Lawson, A.H., Ascher-Svanum, H., 2012. Treatment response trajectories and antipsychotic medications: examination of up to 18 months of treatment in the CATIE chronic schizophrenia trial. Schizophr. Res. 137 (1–3), 141–146.

114

S.Z. Levine, S. Leucht / Schizophrenia Research 156 (2014) 107–114

Lieberman, J.A., Perkins, D., Belger, A., Chakos, M., Jarskog, F., Boteva, K., Gilmore, J., 2001. The early stages of schizophrenia: speculations on pathogenesis, pathophysiology, and therapeutic approaches. Biol. Psychiatry 50 (11), 884–897. Lieberman, J.A., Stroup, T.S., McEvoy, J.P., Swartz, M.S., Rosenheck, R.A., Perkins, D.O., Keefe, R.S., Davis, S.M., Davis, C.E., Lebowitz, B.D., Severe, J., Hsiao, J.K., 2005. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N. Engl. J. Med. 353 (12), 1209–1223. Loo, H., Poirier-Littre, M.F., Theron, M., Rein, W., Fleurot, O., 1997. Amisulpride versus placebo in the medium-term treatment of the negative symptoms of schizophrenia. Br. J. Psychiatry 170, 18–22. Marder, S.R., Daniel, D.G., Alphs, L., Awad, A.G., Keefe, R.S., 2011. Methodological issues in negative symptom trials. Schizophr. Bull. 37 (2), 250–254. Marder, S.R., Alphs, L., Anghelescu, I.G., Arango, C., Barnes, T.R., Caers, I., Daniel, D.G., Dunayevich, E., Fleischhacker, W.W., Garibaldi, G., Green, M.F., Harvey, P.D., Kahn, R. S., Kane, J.M., Keefe, R.S., Kinon, B., Leucht, S., Lindenmayer, J.P., Malhotra, A.K., Stauffer, V., Umbricht, D., Wesnes, K., Kapur, S., Rabinowitz, J., 2013. Issues and perspectives in designing clinical trials for negative symptoms in schizophrenia. Schizophr. Res. 150 (2–3), 328–333. Marques, T.R., Arenovich, T., Agid, O., Sajeev, G., Muthen, B., Chen, L., Kinon, B.J., Kapur, S., 2011. The different trajectories of antipsychotic response: antipsychotics versus placebo. Psychol. Med. 41 (7), 1481–1488. Marshall, M., Lewis, S., Lockwood, A., Drake, R., Jones, P., Croudace, T., 2005. Association between duration of untreated psychosis and outcome in cohorts of first-episode patients: a systematic review. Arch. Gen. Psychiatry 62 (9), 975–983. Melle, I., Larsen, T.K., Haahr, U., Friis, S., Johannesen, J.O., Opjordsmoen, S., Rund, B.R., Simonsen, E., Vaglum, P., McGlashan, T., 2008. Prevention of negative symptom psychopathologies in first-episode schizophrenia: Two-year effects of reducing the duration of untreated psychosis. Arch. Gen. Psychiatry 65 (6), 634–640.

Murphy, B.P., Chung, Y.C., Park, T.W., McGorry, P.D., 2006. Pharmacological treatment of primary negative symptoms in schizophrenia: a systematic review. Schizophr. Res. 88 (1–3), 5–25. Nagin, D.S., 2005. Group-based Modeling of Development. Harvard Univ Pr. Nagin, D.S., Odgers, C.L., 2010. Group-based trajectory modeling in clinical research. Annu. Rev. Clin. Psychol. 6, 109–138. Pelayo-Terán, J.M., Diaz, F.J., Perez-Iglesias, R., Suarez-Pinilla, P., Tabares-Seisdedos, R., de Leon, J., Crespo-Facorro, B., 2014. Trajectories of symptom dimensions in short-term response to antipsychotic treatment in patients with a first episode of non-affective psychosis. Psychol. Med. 44 (1), 37–50. Perkins, D.O., Gu, H., Boteva, K., Lieberman, J.A., 2005. Relationship between duration of untreated psychosis and outcome in first-episode schizophrenia: a critical review and meta-analysis. Am. J. Psychiatry 162 (10), 1785–1804. Rabinowitz, J., 2009. Design and statistical pitfalls and solutions in examining trajectories of response. 22nd ECNP Congress, Istanbul, Turkey. Rabinowitz, J., Davidov, O., 2008. The association of dropout and outcome in trials of antipsychotic medication and its implications for dealing with missing data. Schizophr. Bull. 34 (2), 286–291. Stauffer, V., Case, M., Kollack-Walker, S., Ascher-Svanum, H., Ball, T., Kapur, S., Kinon, B.J., 2011. Trajectories of response to treatment with atypical antipsychotic medication in patients with schizophrenia pooled from 6 double-blind, randomized clinical trials. Schizophr. Res. 130 (1–3), 11–19. Tandon, R., Keshavan, M.S., Nasrallah, H.A., 2008. Schizophrenia, “just the facts” what we know in 2008. 2. Epidemiology and etiology. Schizophr. Res. 102 (1–3), 1–18. Wilder, J., 1957. The law of initial value in neurology and psychiatry; facts and problems. J. Nerv. Ment. Dis. 125 (1), 73–86. Wyatt, R.J., 1991. Neuroleptics and the natural course of schizophrenia. Schizophr. Bull. 17 (2), 325–351.