- Email: [email protected]
DIFFERENCES BETWEEN EARLY RESPONDERS AND EARLY NON-RESPONDERS TO ATYPICAL ANTIPSYCHOTICS ON SYMPTOM AND FUNCTIONAL OUTCOMES IN THE TREATMENT OF SCHIZOPHRENIA
132
Abstracts
CGI-I). In the RCT, we assessed the prediction of ER (>20% total PANSS score reduction) at week 1, 2 and 3 for UR (>30% total PANSS score) or remission (no more than mild symptoms) at endpoint. Results: In 127 adolescents (15.5 years) with schizophrenia (n = 54) or psychosis NOS (n = 73) treated openly for 3 months, CGI-S and CGAS score reductions were significantly more pronounced in the first 4 weeks vs. the following 8 weeks. CGI-I defined ER at 4 weeks predicted UR with a high sensitivity (83%), moderate specificity (69%) and good Positive Predictive Value (PPV: 78%), and Negative Predictive Value (NPV: 76%). ER patients had significantly greater improvements on the CGI-S and GAF score at 4, 8 and 12 weeks than early non-responders. Early nonresponders had significantly more parkinsonism (44.4% vs. 16.7%, p < .05). In the RCT, 293 adolescents (15.5 years) were randomized to aripiprazole 10 mg (N = 99) or 30 mg (N = 97), or placebo (N = 98). Nearly 50% of the aripiprazole PANSS Total reduction was achieved by week 2; up to 75% by week 3. ER at week 3 had better predictive values than ER at week 2. At week 3, ER predicted UR at 6 week with high sensitivity (88%), specificity (83%), PPV (84%), and NPV (87%). Similarly good predictive values were found for remission: sensitivity = 71%, specificity = 76.2%, PPV = 83%, NPV = 62%. Patients achieving ER at week 2 and week 3 had significantly greater reductions in PANSS total, positive and negative scores, CGI-S, CGI-I, and CGAS scores (all p-values < 0.001). The incidence of adverse events between ER and ENR patients was not different. Conclusions: Data in adolescents with schizophrenia-spectrum disorders confirm that most of the treatment response occurs early. Early non-response was a better predictor of later non-response in the RCT than in the naturalistic study that used more global outcome measures. Further studies in adolescents and in real-world settings are needed. In addition, the predictive value of early changes in the CGI-I score need to be examined in youth and adults to test if this more easily implementable scale could serve as a reliable, clinical indicator or early and ultimate response.
measured by the Schizophrenia Objective Functioning Instrument (SOFI), Quality of Life Scale (QLS), and Subjective Wellbeing under Neuroleptics (SWN) scale. A decision tree constructed using classification and regression tree (CART) analysis with pooled data from six randomized, double-blind trials (N = 1494) was applied to HGMN data. Response was defined as a 30% reduction in Positive and Negative Syndrome Scale (PANSS) Total score by Week 8 of treatment. Analyzed predictors were change in individual PANSS items at Weeks 1 and 2. Results: Early response to risperidone was observed in 27.6% of patients. Compared to early non-responders, early responders to risperidone showed significantly more improvement from baseline to endpoint on the SOFI total score and 4 subdomains (p < .001), the QLS total score and 4 subdomains (p < .01), and the SWN total score and 5 subdomains (p < .05). Most of these differences in functioning were already evident and significantly different between the early response and early nonresponse groups by 2 weeks of treatment. From the pooled data set, a 2-step, 6-PANSS item decision tree was created which utilized a 2-point decrease in at least two of five positive PANSS items at Week 2 (Step 1) followed by a 2-point decrease in the PANSS excitement item at Week 2 (Step 2). Using this approach with HGMN data, response could be predicted in most patients (93%) with a high positive predictive value (70%) and negative predictive value (77%). Conclusion: Patients who show an early response to antipsychotic treatment as measured by improvement in psychiatric symptom severity show early and consistent improvement across multiple domains of functioning, a finding that was concordant between both physician- and patient-rated quality of life scales. Additional analyses are ongoing to explore specific symptoms and/or functional domains whose early response may predict subsequent clinically relevant global outcomes.
doi:10.1016/j.schres.2010.02.088
doi:10.1016/j.schres.2010.02.087
DIFFERENCES BETWEEN EARLY RESPONDERS AND EARLY NON-RESPONDERS TO ATYPICAL ANTIPSYCHOTICS ON SYMPTOM AND FUNCTIONAL OUTCOMES IN THE TREATMENT OF SCHIZOPHRENIA Bruce Kinon1, Lei Chen1, Virginia Stauffer1, Haya Ascher-Svanum1, Wei Zhou1, Sara Kollack-Walker1, John Kane2, Shitij Kapur3 1 Eli Lilly & Company, Indianapolis, IN, USA; 2Zucker Hillside Hospital, Glen Oaks, NY, USA; 3Institute of Psychiatry, Kings College, London, UK
Objectives: In this study, we extend the findings from a prospective clinical trial (Study HGMN) assessing the effects of early response to an atypical antipsychotic across multiple functional outcome measures. In addition, we applied a data-driven decision tree to the data in order to assess the value of using specific predictors to differentiate responders from non-responders. Methods: This was a randomized, double-blind, flexible-dose, 12week study that enrolled chronically-ill patients (n = 628) diagnosed with schizophrenia or schizoaffective disorder who were experiencing an acute symptom exacerbation. Patients were initially assigned to risperidone drug therapy (2-6 mg/day), and their response status at 2 weeks determined. Early responders continued with risperidone therapy, whereas early non-responders were randomized (1:1) in a double-blind manner to either continue on risperidone or switch another atypical antipsychotic for 10 additional weeks of therapy. Subsequent improvement in functioning was
EARLY RESPONSE TO ANTIPSYCHOTICS – WHAT CHANGES EARLY AND HOW? Shitij Kapur1, Tamara Arenovich2, Ofer Agid2 1 Institute of Psychiatry, Kings College, London, UK; Addiction and Mental Health, Toronto, Canada
2
Centre for
Background: While the 'early onset' of antipsychotic response is now relatively well established – two subsequent questions emerge: what symptoms change early and do they have different trajectories of response? While the majority of the PANSS items and all 'factors' (whether one uses three subscales or five factors) show some degree of improvement – it raises the interesting question of whether all these factors and subscales show a similar trajectory of improvement. Methods: To resolve this issue we examined it in a large dataset of 420 individuals treated with olanzapine, haloperidol and placebo – the presence of a large number of placebo-treated patients gives one the possibility of testing whether drugs and placebo have similar trajectories. The data were modelled as a Growth Mixture Model in M-Plus. Results: The trajectory of response of the psychotic symptoms rejected a single-trajectory model, and instead best-fit a four trajectory model. The most common trajectory showed modest change (about 20% improvement over six weeks) – at the margin of what is clinically detectable. A small subset of patients showed significant (50%) or dramatic (∼70%) response in symptoms. The latter were significantly over-represented in the drug-treated dataset.
Abstracts
CGI-I). In the RCT, we assessed the prediction of ER (>20% total PANSS score reduction) at week 1, 2 and 3 for UR (>30% total PANSS score) or remission (no more than mild symptoms) at endpoint. Results: In 127 adolescents (15.5 years) with schizophrenia (n = 54) or psychosis NOS (n = 73) treated openly for 3 months, CGI-S and CGAS score reductions were significantly more pronounced in the first 4 weeks vs. the following 8 weeks. CGI-I defined ER at 4 weeks predicted UR with a high sensitivity (83%), moderate specificity (69%) and good Positive Predictive Value (PPV: 78%), and Negative Predictive Value (NPV: 76%). ER patients had significantly greater improvements on the CGI-S and GAF score at 4, 8 and 12 weeks than early non-responders. Early nonresponders had significantly more parkinsonism (44.4% vs. 16.7%, p < .05). In the RCT, 293 adolescents (15.5 years) were randomized to aripiprazole 10 mg (N = 99) or 30 mg (N = 97), or placebo (N = 98). Nearly 50% of the aripiprazole PANSS Total reduction was achieved by week 2; up to 75% by week 3. ER at week 3 had better predictive values than ER at week 2. At week 3, ER predicted UR at 6 week with high sensitivity (88%), specificity (83%), PPV (84%), and NPV (87%). Similarly good predictive values were found for remission: sensitivity = 71%, specificity = 76.2%, PPV = 83%, NPV = 62%. Patients achieving ER at week 2 and week 3 had significantly greater reductions in PANSS total, positive and negative scores, CGI-S, CGI-I, and CGAS scores (all p-values < 0.001). The incidence of adverse events between ER and ENR patients was not different. Conclusions: Data in adolescents with schizophrenia-spectrum disorders confirm that most of the treatment response occurs early. Early non-response was a better predictor of later non-response in the RCT than in the naturalistic study that used more global outcome measures. Further studies in adolescents and in real-world settings are needed. In addition, the predictive value of early changes in the CGI-I score need to be examined in youth and adults to test if this more easily implementable scale could serve as a reliable, clinical indicator or early and ultimate response.
measured by the Schizophrenia Objective Functioning Instrument (SOFI), Quality of Life Scale (QLS), and Subjective Wellbeing under Neuroleptics (SWN) scale. A decision tree constructed using classification and regression tree (CART) analysis with pooled data from six randomized, double-blind trials (N = 1494) was applied to HGMN data. Response was defined as a 30% reduction in Positive and Negative Syndrome Scale (PANSS) Total score by Week 8 of treatment. Analyzed predictors were change in individual PANSS items at Weeks 1 and 2. Results: Early response to risperidone was observed in 27.6% of patients. Compared to early non-responders, early responders to risperidone showed significantly more improvement from baseline to endpoint on the SOFI total score and 4 subdomains (p < .001), the QLS total score and 4 subdomains (p < .01), and the SWN total score and 5 subdomains (p < .05). Most of these differences in functioning were already evident and significantly different between the early response and early nonresponse groups by 2 weeks of treatment. From the pooled data set, a 2-step, 6-PANSS item decision tree was created which utilized a 2-point decrease in at least two of five positive PANSS items at Week 2 (Step 1) followed by a 2-point decrease in the PANSS excitement item at Week 2 (Step 2). Using this approach with HGMN data, response could be predicted in most patients (93%) with a high positive predictive value (70%) and negative predictive value (77%). Conclusion: Patients who show an early response to antipsychotic treatment as measured by improvement in psychiatric symptom severity show early and consistent improvement across multiple domains of functioning, a finding that was concordant between both physician- and patient-rated quality of life scales. Additional analyses are ongoing to explore specific symptoms and/or functional domains whose early response may predict subsequent clinically relevant global outcomes.
doi:10.1016/j.schres.2010.02.088
doi:10.1016/j.schres.2010.02.087
DIFFERENCES BETWEEN EARLY RESPONDERS AND EARLY NON-RESPONDERS TO ATYPICAL ANTIPSYCHOTICS ON SYMPTOM AND FUNCTIONAL OUTCOMES IN THE TREATMENT OF SCHIZOPHRENIA Bruce Kinon1, Lei Chen1, Virginia Stauffer1, Haya Ascher-Svanum1, Wei Zhou1, Sara Kollack-Walker1, John Kane2, Shitij Kapur3 1 Eli Lilly & Company, Indianapolis, IN, USA; 2Zucker Hillside Hospital, Glen Oaks, NY, USA; 3Institute of Psychiatry, Kings College, London, UK
Objectives: In this study, we extend the findings from a prospective clinical trial (Study HGMN) assessing the effects of early response to an atypical antipsychotic across multiple functional outcome measures. In addition, we applied a data-driven decision tree to the data in order to assess the value of using specific predictors to differentiate responders from non-responders. Methods: This was a randomized, double-blind, flexible-dose, 12week study that enrolled chronically-ill patients (n = 628) diagnosed with schizophrenia or schizoaffective disorder who were experiencing an acute symptom exacerbation. Patients were initially assigned to risperidone drug therapy (2-6 mg/day), and their response status at 2 weeks determined. Early responders continued with risperidone therapy, whereas early non-responders were randomized (1:1) in a double-blind manner to either continue on risperidone or switch another atypical antipsychotic for 10 additional weeks of therapy. Subsequent improvement in functioning was
EARLY RESPONSE TO ANTIPSYCHOTICS – WHAT CHANGES EARLY AND HOW? Shitij Kapur1, Tamara Arenovich2, Ofer Agid2 1 Institute of Psychiatry, Kings College, London, UK; Addiction and Mental Health, Toronto, Canada
2
Centre for
Background: While the 'early onset' of antipsychotic response is now relatively well established – two subsequent questions emerge: what symptoms change early and do they have different trajectories of response? While the majority of the PANSS items and all 'factors' (whether one uses three subscales or five factors) show some degree of improvement – it raises the interesting question of whether all these factors and subscales show a similar trajectory of improvement. Methods: To resolve this issue we examined it in a large dataset of 420 individuals treated with olanzapine, haloperidol and placebo – the presence of a large number of placebo-treated patients gives one the possibility of testing whether drugs and placebo have similar trajectories. The data were modelled as a Growth Mixture Model in M-Plus. Results: The trajectory of response of the psychotic symptoms rejected a single-trajectory model, and instead best-fit a four trajectory model. The most common trajectory showed modest change (about 20% improvement over six weeks) – at the margin of what is clinically detectable. A small subset of patients showed significant (50%) or dramatic (∼70%) response in symptoms. The latter were significantly over-represented in the drug-treated dataset.