Innovations In Clinical Neuroscience

NOV-DEC 2017

A peer-reviewed, evidence-based journal for clinicians in the field of neuroscience

Issue link:

Contents of this Issue


Page 54 of 83

55 ICNS INNOVATIONS IN CLINICAL NEUROSCIENCE November-December 2017 • Volume 14 • Number 11–12 R E V I E W positive and negative symptom severity scores is relatively low, indicating a degree of separation in the underlying neurobiological substrate. 6–9 This finding is consistent with evidence that the neurobiological correlates of positive symptoms are largely different from those that underlie negative symptoms. 11–13 In contrast, the standard (Marder) PANSS factor change scores have been found to be highly correlated with each other. Especially notable is the high degree of correlation between the PANSS positive factor and other Marder factors. We have confirmed this finding in a previous pooled analysis of PANSS data derived from five double-blind, placebo-controlled studies of lurasidone for the treatment of patients with an acute exacerbation of schizophrenia. 14 Moderate- to-high correlations were observed between improvements in standard (Marder) PANSS positive and other PANSS factors, ranging from r=0.52 to r=0.74. Correlations of this magnitude represent a major confound that make interpretation of treatment-related improvement in individual Marder factors challenging. As a result, improvement in symptom severity in one of the Marder factors might not represent a treatment effect on a specific symptom domain, but instead might simply be a nonspecific effect secondary to improvement in PANSS items that are highly correlated with the PANSS positive factor. The high degree of between-factor correlation among Marder PANSS factors has been characterized as an example of pseudospecificity. Pseudospecificity is the term, originally promulgated by the United States Food and Drug Administration (FDA), 15,16 to describe potential pharmacologic symptom targets that are too highly correlated (overlapping) in terms of phenomenology and measured treatment response to justify separate drug treatment claims. POSSIBLE SOLUTIONS TO THE PROBLEM OF PSEUDOSPECIFICIT Y There are two possible approaches that can be used to address the measurement issue that is central to pseudospecificity, both of which have advantages. One approach is to direct research efforts toward the development and validation of new, domain-specific, instruments with minimal-to-no correlation with other outcome domains. Examples of this approach include instruments such as the Negative Symptom Assessment (NSA) Scale 17–19 and the MATRICS Consensus Cognitive Battery (which measures cognitive function). 20,21 However, to our knowledge, evidence has not been presented demonstrating that instruments (such as the NSA) have low levels of correlation with PANSS positive (and other) factors in patients experiencing an acute exacerbation of schizophrenia. A second approach would retain the PANSS as an efficacy measure while using analytic strategies to minimize the degree of between- factor correlation. The advantage of this approach is that it is a cost-effective strategy that would permit acquisition of domain- specific efficacy data using required acute registration trials without the need to conduct, for example, a separate negative symptom trial. In addition, validation of modified PANSS factors with low between-factor correlation would facilitate study replication while preserving decades of treatment research with dozens of agents, many with distinctive receptor binding profiles, that could be retrospectively analyzed for domain-specific efficacy. We briefly summarize here recently reported results of an uncorrelated PANSS score matrix (UPSM) transform designed to reduce pseudospecificity in the assessment of symptom change in patients with schizophrenia. 14 We also report new results that assess whether the UPSM transform can also be applied directly to PANSS severity scores at baseline. TRANSFORMED PANSS FAC TORS: HIGH CORRELATION WITH STANDARD (MARDER) FAC TORS The original endpoint change score analysis 14 was performed on PANSS data derived from five similarly designed, randomized, double-blind, placebo-controlled, six-week treatment studies of lurasidone or active comparator for the treatment of patients (N=1,710) with an acute exacerbation of schizophrenia. In this analysis, each of the seven transformed PANSS factors was found to have a moderate-to-high degree of correlation with its respective standard (Marder) factor: for the positive factor (0.79), for the disorganized factor (0.79), for negative-apathy/avolition with negative symptoms (0.75), for negative-deficit of expression with negative symptoms (0.65), for hostility/excitement (0.94), for anxiety with anxiety/depression (0.74), and for depression with anxiety/depression (0.76). These levels of correlation suggest that the transformed PANSS factors are measuring the same symptom domains (constructs) as the Marder factors. TRANSFORMED PANSS FAC TORS: LOW BET WEEN-FAC TOR CORRELATION In our original endpoint change score analysis of five pooled clinical trials, 14 transformed PANSS factors exhibited markedly reduced between-factor correlations when compared to the between-factor correlations observed for the standard (Marder) PANSS factors. Pearson's between-factor correlations ranged from 0.40 to 0.74 for endpoint change in the standard (Marder) PANSS factors, and from -0.01 to 0.27 in the transformed PANSS factors (with 0.27 being the correlation between the anxiety and depression subfactors). The low levels of between-factor correlation among the transformed PANSS factors suggest that these reconfigured factors are orthogonal to each other and are successfully measuring the effect of treatment with a high degree of specificity. TRANSFORMED PANSS FAC TORS: LURASIDONE VS. PLACEBO EFFEC T SIZES In an additional analysis of the five pooled clinical trials, 14 effect size estimates for endpoint change (lurasidone vs. placebo) were calculated using both standard (Marder) PANSS factors, and transformed PANSS factors. Effect size estimates exhibited a relatively consistent pattern across the standard (Marder) factors, ranging from 0.31 to 0.44. In contrast, greater between-factor heterogeneity was observed in effect size estimates using the transformed PANSS factors, ranging from 0.05 to 0.27, with greater lurasidone effects observed on positive and hostility symptoms, and smaller drug effects on disorganized, negative apathy/ avolition, deficit of expression, and anxiety/ depression symptoms. TRANSFORMED PANSS FAC TORS: CROSS- STUDY VALIDATION In a validation analysis using 12 separate clinical trials, 14 we confirmed that the

Articles in this issue

Archives of this issue

view archives of Innovations In Clinical Neuroscience - NOV-DEC 2017