Evaluating speech latencies during structured psychiatric interviews as an automated objective measure of psychomotor slowing.

We sought to derive an objective measure of psychomotor slowing from speech analytics during a psychiatric interview to avoid potential burden of dedicated neurophysiological testing. Speech latency, which reflects response time between speakers, shows promise from the literature. Speech data was obtained from 274 subjects with a diagnosis of bipolar I depression enrolled in a randomized, doubleblind, 6-week phase 2 clinical trial. Audio recordings of structured Montgomery-Åsberg Depression Rating Scale (MADRS) interviews at 6 time points were examined (k = 1,352). We evaluated speech latencies, and other aspects of speech, for temporal stability, convergent validity, sensitivity/responsivity to clinical change, and generalization across seven socio-linguistically diverse countries. Speech latency was minimally associated with demographic features, and explained nearly a third of the variance in depression (categorically defined). Speech latency significantly decreased as depression symptoms improved over time, explaining nearly 20 % of variance in depression remission. Classification for differentiating people with versus without concurrent depression was high (AUCs > 0.85) both cross-sectionally and longitudinally. Results replicated across countries. Other speech features offered modest incremental contribution. Neurophysiological speech parameters with face validity can be derived from psychiatric interviews without the added patient burden of additional testing.

Enrichment using speech latencies improves treatment effect size in a clinical trial of bipolar depression.

Clinical trials in depression lack objective measures. Speech latencies are an objective measure of psychomotor slowing with face validity and empirical support. 'Turn latency' is the response time between speakers. Retrospective analysis was carried-out on the utility of turn latencies as an enrichment tool in a clinical trial of bipolar I depression. Speech data was obtained from 274 participants during 1,352 Montgomery-Åsberg Depression Rating Scale (MADRS) recordings in a randomized, placebo controlled, 6-week clinical trial of SEP-4199 (200 mg or 400 mg). Post-randomization turn latencies were compared between patients with moderate to severe depression and patients whose depression had remitted. A cutoff was determined and applied to turn latencies pre-randomization to classify individuals into two groups: Speech Latencies Slow (SL-Slow) and Speech Latencies Normal (SL-Normal). At week 6, SL-Slow (N = 172) showed significant separation in MADRS scores between placebo and treatment arms. SL-Normal (N = 102) showed larger MADRS improvements and no significant separation between placebo and treatment arms. Excluding SL-Normal increased primary outcome effect size by 52 % and 100 % for the treatment arms. Turn latencies are an objective measure available from standard clinical assessments and may assess the severity of symptoms more accurately and screen out placebo responders.

A clinical trial inclusion criteria to enrich for patients presenting with canonical symptom structure in bipolar depression.

Clinical drug development in psychiatry is challenging due to heterogeneous patient populations and the uncertainty of measuring neuropsychiatric constructs with symptom rating scales. Here we describe the development and implementation of an enrichment algorithm that identifies canonical versus anomalous symptom presentations, at the individual subject level, based on MADRS ratings obtained at screening and baseline. Data from 5 randomized, placebo-controlled, phase 3 trials in bipolar I disorder was used (N = 2026 subjects and 15,239 MADRS assessments). A variance-covariance difference (VCD) vector was developed to encode individual symptom structure using the 10 items of MADRS from the two sequential assessments. An anomaly score, calculated from each subject's VCD vector was derived by isolation forest to quantify the degree of disparity from the hypothesized canonical item structure. A retrospective application of the algorithm reliably identified a threshold anomaly score above which the psychometric properties of MADRS deteriorate. Consistent with increasing the certainty of MADRS ratings, subjects with a canonical symptom structure at baseline demonstrated greater effect sizes post-baseline in a phase 2 placebo-controlled trial of non-racemic amisulpride (SEP-4199) for bipolar depression, analyzed retrospectively. Our analyses show that the developed algorithm can reduce the symptom structure heterogeneity at baseline and thus improve the measurement certainty of psychiatric symptoms in clinical trials. This novel enrichment method has been prospectively implemented in a Phase 3 clinical study of SEP-4199 and is consistent with regulatory guidelines aimed at increasing the statistical power and lowering patient-burden in clinical trials. Clinical Trials Registry: NCT00868452, NCT00868699, NCT01284517, NCT01986101, NCT03543410, NCT05169710.

Estimating heterogeneity of treatment effect in psychiatric clinical trials.

Currently, placebo-controlled clinical trials report mean change and effect sizes, which masks information about heterogeneity of treatment effects (HTE). Here, we present a method to estimate HTE and evaluate the null hypothesis (H0) that a drug has equal benefit for all participants (HTE=0). We developed measure termed 'estimated heterogeneity of treatment effect' or eHTE, which estimates variability in drug response by comparing distributions between study arms. This approach was tested across numerous large placebo-controlled clinical trials. In contrast with variance-based methods which have not identified heterogeneity in psychiatric trials, reproducible instances of treatment heterogeneity were found. For example, heterogeneous response was found in a trial of venlafaxine for depression (peHTE=0.034), and two trials of dasotraline for binge eating disorder (Phase 2, peHTE=0.002; Phase 3, 4mg peHTE=0.011; Phase 3, 6mg peHTE=0.003). Significant response heterogeneity was detected in other datasets as well, often despite no difference in variance between placebo and drug arms. The implications of eHTE as a clinical trial outcomes independent from central tendency of the group is considered and the important of the eHTE method and results for drug developers, providers, and patients is discussed.

Assessment of Negative Symptoms in Clinical Trials of Acute Schizophrenia: Test of a Novel Enrichment Strategy.

Drug trials for negative symptoms in schizophrenia select patients based on the severity and stability of negative symptoms, using criteria that are not suitable for trials of acute exacerbation of schizophrenia. Here we present a method to prognostically enrich subjects having a predefined factor structure in PANSS and apply it to the measurement of negative symptoms specifically in trials of acute schizophrenia. A vector of 1335 elements based on between- and within-item variances, covariances, and differences of PANSS items was created to calculate an index of heterogeneity and to enrich for a predetermined symptom construct in PANSS. Using prerandomization PANSS scores across N = 4876 subjects in 13 trials of acute schizophrenia, we demonstrate an ability to select for a subpopulation having the greatest amount of variance explained across the 7-items of the Marder PANSS negative symptom (MPNS) construct. Network analyses on subjects enriched for MPNS construct confirm that negative symptoms were most influential in overall psychopathology, distinct from subjects without the MPNS construct. As expected for D2 antagonists, drug-placebo differences on negative symptoms with lurasidone were not specific to the subpopulation having the MPNS construct. In contrast, the novel TAAR1 agonist ulotaront demonstrated specific improvements in negative symptoms which were greatest in the MPNS subpopulation. These results demonstrate the utility of a novel prognostic enrichment strategy that can address heterogeneity in clinical trials, where patients can be selected on the basis of a greater likelihood of having the measured symptom construct (negative symptoms) related to the disorder (schizophrenia). ClinicalTrials.gov Identifiers: NCT0296938, NCT00088634, NCT00549718, NCT00615433, NCT00790192.

Ranirestat for the management of diabetic sensorimotor polyneuropathy.

OBJECTIVE: Aldose reductase inhibitors (ARIs) are potential disease modifiers for diabetes complications. We aimed to determine whether ranirestat, an ARI, could slow or reverse the course of diabetic sensorimotor polyneuropathy (DSP). RESEARCH DESIGN AND METHODS: A total of 549 patients with DSP were randomly assigned to treatment with placebo or 10, 20, or 40 mg/day ranirestat for 52 weeks in this multicenter, double-blind study. Efficacy was evaluated by nerve conduction studies, the modified Toronto Clinical Neuropathy Score (mTCNS), and quantitative sensory tests (QSTs). RESULTS: At week 52, the summed sensory (bilateral sural plus proximal median sensory) nerve conduction velocity (NCV) did not show significant changes from baseline (2.0 m/s for placebo compared with 3.2-3.8 m/s for ranirestat). Significant improvement in the summed motor (peroneal, tibial, and median) NCV was observed with 20 and 40 mg/day ranirestat treatment at week 12 (P