A Phase I Study of Acapatamab, a Half-life Extended, PSMA-Targeting Bispecific T-cell Engager for Metastatic Castration-Resistant Prostate Cancer.

PURPOSE: Safety and efficacy of acapatamab, a prostate-specific membrane antigen (PSMA) x CD3 bispecific T-cell engager were evaluated in a first-in-human study in metastatic castration-resistant prostate cancer (mCRPC). PATIENTS AND METHODS: Patients with mCRPC refractory to androgen receptor pathway inhibitor therapy and taxane-based chemotherapy received target acapatamab doses ranging from 0.003 to 0.9 mg in dose exploration (seven dose levels) and 0.3 mg (recommended phase II dose) in dose expansion intravenously every 2 weeks. Safety (primary objective), pharmacokinetics, and antitumor activity (secondary objectives) were assessed. RESULTS: In all, 133 patients (dose exploration, n = 77; dose expansion, n = 56) received acapatamab. Cytokine release syndrome (CRS) was the most common treatment-emergent adverse event seen in 97.4% and 98.2% of patients in dose exploration and dose expansion, respectively; grade ≥ 3 was seen in 23.4% and 16.1%, respectively. Most CRS events were seen in treatment cycle 1; incidence and severity decreased at/beyond cycle 2. In dose expansion, confirmed prostate-specific antigen (PSA) responses (PSA50) were seen in 30.4% of patients and radiographic partial responses in 7.4% (Response Evaluation Criteria in Solid Tumors 1.1). Median PSA progression-free survival (PFS) was 3.3 months [95% confidence interval (CI): 3.0-4.9], radiographic PFS per Prostate Cancer Clinical Trials Working Group 3 was 3.7 months (95% CI: 2.0-5.4). Acapatamab induced T-cell activation and increased cytokine production several-fold within 24 hours of initiation. Treatment-emergent antidrug antibodies were detected in 55% and impacted serum exposures in 36% of patients in dose expansion. CONCLUSIONS: Acapatamab was safe and tolerated and had a manageable CRS profile. Preliminary signs of efficacy with limited durable antitumor activity were observed. Acapatamab demonstrated pharmacokinetic and pharmacodynamic activity.

Characterization and root cause analysis of immunogenicity to pasotuxizumab (AMG 212), a prostate-specific membrane antigen-targeting bispecific T-cell engager therapy.

Introduction: In oncology, anti-drug antibody (ADA) development that significantly curtails response durability has not historically risen to a level of concern. The relevance and attention ascribed to ADAs in oncology clinical studies have therefore been limited, and the extant literature on this subject scarce. In recent years, T cell engagers have gained preeminence within the prolific field of cancer immunotherapy. These drugs whose mode of action is expected to potently stimulate anti-tumor immunity, may potentially induce ADAs as an unintended corollary due to an overall augmentation of the immune response. ADA formation is therefore emerging as an important determinant in the successful clinical development of such biologics. Methods: Here we describe the immunogenicity and its impact observed to pasotuxizumab (AMG 212), a prostate-specific membrane antigen (PSMA)-targeting bispecific T cell engager (BiTE®) molecule in NCT01723475, a first-in-human (FIH), multicenter, dose-escalation study in patients with metastatic castration-resistant prostate cancer (mCRPC). To explain the disparity in ADA incidence observed between the SC and CIV arms of the study, we interrogated other patient and product-specific factors that may have explained the difference beyond the route of administration. Results: Treatment-emergent ADAs (TE-ADA) developed in all subjects treated with at least 1 cycle of AMG 212 in the subcutaneous (SC) arm. These ADAs were neutralizing and resulted in profound exposure loss that was associated with contemporaneous reversal of initial Prostate Surface Antigen (PSA) responses, curtailing durability of PSA response in patients. Pivoting from SC to a continuous intravenous (CIV) administration route remarkably yielded no subjects developing ADA to AMG 212. Through a series of stepwise functional assays, our investigation revealed that alongside a more historically immunogenic route of administration, non-tolerant T cell epitopes within the AMG 212 amino acid sequence were likely driving the high-titer, sustained ADA response observed in the SC arm. Discussion: These mechanistic insights into the AMG 212 ADA response underscore the importance of performing preclinical immunogenicity risk evaluation as well as advocate for continuous iteration to better our biologics.

Machine learning-guided, big data-enabled, biomarker-based systems pharmacology: modeling the stochasticity of natural history and disease progression.

The incidence of systemic and metabolic co-morbidities increases with aging. The purpose was to investigate a novel paradigm for modeling the orchestrated changes in many disease-related biomarkers that occur during aging. A hybrid strategy that integrates machine learning and stochastic modeling was evaluated for modeling the long-term dynamics of biomarker systems. Bayesian networks (BN) were used to identify quantitative systems pharmacology (QSP)-like models for the inter-dependencies for three disease-related datasets of metabolic (MB), metabolic with leptin (MB-L), and cardiovascular (CVB) biomarkers from the NHANES database. Biomarker dynamics were modeled using discrete stochastic vector autoregression (VAR) equations. BN were used to derive the topological order and connectivity of a data driven QSP model structure for inter-dependence of biomarkers across the lifespan. The strength and directionality of the connections in the QSP models were evaluated using bootstrapping. VAR models based on QSP model structures from BN were assessed for modeling biomarker system dynamics. BN-restricted VAR models of order 1 were identified as parsimonious and effective for characterizing biomarker system dynamics in the MB, MB-L and CVB datasets. Simulation of annual and triennial data for each biomarker provided good fits and predictions of the training and test datasets, respectively. The novel strategy harnesses machine learning to construct QSP model structures for inter-dependence of biomarkers. Stochastic modeling with the QSP models was effective for predicting the age-varying dynamics of disease-relevant biomarkers over the lifespan.

Machine learning in pharmacometrics: Opportunities and challenges.

The explosive growth in medical devices, imaging and diagnostics, computing, and communication and information technologies in drug development and healthcare has created an ever-expanding data landscape that the pharmacometrics (PMX) research community must now traverse. The tools of machine learning (ML) have emerged as a powerful computational approach in other data-rich disciplines but its effective utilization in the pharmaceutical sciences and PMX modelling is in its infancy. ML-based methods can complement PMX modelling by enabling the information in diverse sources of big data, e.g. population-based public databases and disease-specific clinical registries, to be harnessed because they are capable of efficiently identifying salient variables associated with outcomes and delineating their interdependencies. ML algorithms are computationally efficient, have strong predictive capabilities and can enable learning in the big data setting. ML algorithms can be viewed as providing a computational bridge from big data to complement PMX modelling. This review provides an overview of the strengths and weaknesses of ML approaches vis-à-vis population methods, assesses current research into ML applications in the pharmaceutical sciences and provides perspective for potential opportunities and strategies for the successful integration and utilization of ML in PMX.

The cholesterol autoxidation products, 7-ketocholesterol and 7ß-hydroxycholesterol are associated with serum neurofilaments in multiple sclerosis.

BACKGROUND: Serum neurofilament light chain (sNfL) is an established marker of neuroaxonal injury in multiple sclerosis (MS). OBJECTIVES: To investigate if oxysterols produced from non-enzymatic and enzymatic cholesterol oxidation are differentially associated with sNfL measurements in MS. METHODS: This longitudinal study included 62 relapsing-remitting (RR-MS) and 36 progressive MS (PMS) patients with baseline and 5-year follow-up measures of serum levels of 6 oxysterols, sNfL and lipids. The oxysterols, 24-hydroxycholesterol (24HC), 25HC, 27HC, 7αHC, 7ßHC and 7-ketocholesterol (7KC), were measured using liquid chromatography-mass spectrometry. sNfL was measured using single molecular array assay. Serum high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) levels were obtained from a lipid profile. RESULTS: The enzymatically produced oxysterols 24HC, 25HC, 27HC and 7αHC were not associated with sNfL. However, baseline levels of reactive oxygen species (ROS) produced oxysterols, 7KC (p = 0.032) and 7ßHC (p = 0.0025), were positively associated with sNfL levels at follow-up. Follow-up 7KC (p = 0.038) levels were also associated with follow-up sNfL levels. The associations of 7KC or 7ßHC with sNfL remained significant after adjusting for LDL-C or HDL-C. CONCLUSIONS: 7KC and 7ßHC, produced by ROS-mediated cholesterol oxidation are associated with neuroaxonal injury as assessed by sNfL in MS.

Neurofilament levels are associated with blood-brain barrier integrity, lymphocyte extravasation, and risk factors following the first demyelinating event in multiple sclerosis.

BACKGROUND: Increased blood brain barrier (BBB) permeability, CNS inflammation and neuroaxonal damage are pathological hallmarks in early multiple sclerosis (MS). OBJECTIVE: To investigate the associations of neurofilament light chain (NfL) levels with measures of BBB integrity and central nervous system (CNS) inflammation in MS during the first demyelinating event. METHODS: Blood and cerebrospinal fluid (CSF) were obtained from 142 MS (McDonald 2017) treatment-naive patients from the SET study (63% female; age: 29.7 ± 7.9 years) following the disease onset. NfL, albumin, immunoglobulin G (IgG), and immunoglobulin M (IgM) levels were measured in CSF and blood samples. Albumin quotient was computed as a marker of BBB integrity. Immune cell subset counts in CSF were measured using flow cytometry. MS risk factors, such as Human leukocyte antigen DRB1 locus gene (HLA DRB1)*1501, anti-Epstein-Barr virus (EBV) antibodies, and 25-hydroxy vitamin D3, were also measured. RESULTS: Higher serum NfL (sNfL) levels were associated with higher albumin quotient (p < 0.001), CSF CD80+ (p = 0.012), and CD80+ CD19+ (p = 0.015) cell frequency. sNfL levels were also associated with contrast-enhancing and T2 lesions on brain magnetic resonance imaging (MRI; all p ⩽ 0.001). Albumin quotient was not associated with any of the MS risk factors assessed. sNfL levels were associated with anti-EBV viral capsid antigen (VCA) IgG levels (p = 0.0026). CONCLUSION: sNfL levels during the first demyelinating event of MS are associated with greater impairment of BBB integrity, immune cell extravasation, and brain lesion activity on MRI.

Neuroprotective associations of apolipoproteins A-I and A-II with neurofilament levels in early multiple sclerosis.

BACKGROUND: The role of cholesterol homeostasis in neuroaxonal injury in multiple sclerosis is not known. OBJECTIVE: The objective of the study is to investigate the associations of cerebrospinal fluid (CSF) and serum neurofilament light chain levels (CSF-NfL and sNfL, respectively), which are biomarkers of neuroaxonal injury, with cholesterol biomarkers at the clinical onset of multiple sclerosis. METHODS: sNfL, serum cholesterol profile (total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol), serum apolipoprotein (Apo) levels (ApoA-I, ApoA-II, ApoB, and ApoE), and albumin quotient were obtained for 133 patients (63% female, age: 29.9 ± 8.0 years) during the first demyelinating event. CSF-NfL was available for 103 (77%) patients. RESULTS: CSF-NfL and sNfL were negatively associated with serum ApoA-II (P = .005, P < .001) and positively associated with albumin quotient (P < .001, P < .0001). In addition, higher CSF-NfL was associated with lower serum ApoA-I (P = .009) levels and higher sNfL was associated with lower high-density lipoprotein cholesterol (P = .010). In stepwise regression, age (P = .045), serum ApoA-II (P = .022), and albumin quotient (P < .001) were associated with CSF-NfL; albumin quotient (P = .002) and ApoA-II (P = .001) were associated with sNfL. Path analysis identified parallel pathways from ApoA-II (P = .009) and albumin quotient (P < .001) to the sNfL outcome that were mediated by CSF-NfL (P < .001). The associations of CSF-NfL with ApoA-I (P = .014) and ApoA-II (P = .015) and sNfL with ApoA-II (P < .001) remained significant after adjusting for number of contrast-enhancing lesions and T2 lesion volume. CONCLUSION: Lower serum ApoA-II and ApoA-I levels are associated with greater neuroaxonal injury as measured by CSF-NfL.

Apolipoproteins AI and E are associated with neuroaxonal injury to gray matter in multiple sclerosis.

Purpose To investigate the associations between longitudinal changes in lipid biomarkers and serum neurofilament (sNfL) levels in multiple sclerosis (MS) neurodegeneration and disease progression. Methods 5-year prospective, longitudinal study included 75 relapsing-remitting MS (RR-MS) and 37 progressive-MS (P-MS) patients. sNfL, plasma total cholesterol (TC), high-density (HDL-C) and low-density (LDL-C) lipoprotein cholesterol, apolipoproteins (Apo), ApoA-I, Apo-II, ApoB, ApoC-II and ApoE were measured at baseline and 5-years. Annual percent changes in whole brain volume (PBVC), gray matter volume (PGMVC) and cortical volume (PCVC) were obtained from MRI at baseline and 5-years. Results sNfL levels at 5-year follow-up were associated with ApoE at follow-up (p = 0.014), age at follow-up, body mass index (p < 0.001) and RR vs. P-MS status at follow-up. APOE4 allele was associated with greater sNfL levels at 5-years (p = 0.022) and pronounced in the P-MS group. PGMVC and PCVC were associated with percent changes in HDL-C (p = 0018 and p < 0.001, respectively) and ApoA-I (p = 0.0073 and p = 0.006). PGMVC and PCVC remained associated with percent change in HDL-C (p = 0.0024 and p < 0.001, respectively) after sNfL was included as a predictor. Conclusions HDL-C percent change is associated with decreased gray matter atrophy after adjusting for baseline sNfL.

Generalized Pharmacometric Modeling, a Novel Paradigm for Integrating Machine Learning Algorithms: A Case Study of Metabolomic Biomarkers.

There is an unmet need for identifying innovative machine learning (ML) strategies to improve drug treatment regimens and therapeutic outcomes. We investigate Generalized Pharmacometric Modeling (GPM), a novel paradigm that integrates ML algorithms with pharmacokinetic and pharmacodynamic structural models, population covariate modeling, and "big data," and enables identification of patient-specific factors contributing to drug disposition. We hypothesize that GPM will enhance forecasting of drug outcomes in diverse populations. We assessed random forest regression in conjunction with Bayesian networks as the ML methods within GPM and used the National Health and Nutrition Examination Survey population-based study database. GPM was utilized to identify subject-specific factors associated with cholesterol dynamics. Our results demonstrate the utility of GPM to enhance pharmacometrics modeling and its potential for modeling drug outcomes in diverse populations.

An AC electrokinetics facilitated biosensor cassette for rapid pathogen identification.

To develop a portable point-of-care system based on biosensors for common infectious diseases such as urinary tract infection, the sensing process needs to be implemented within an enclosed fluidic system. On chip sample preparation of clinical samples remains a significant obstacle to achieving robust sensor performance. Herein AC electrokinetics is applied in an electrochemical biosensor cassette to enhance molecular convection and hybridization efficiency through electrokinetics induced fluid motion and Joule heating induced temperature elevation. Using E. coli as an exemplary pathogen, we determined the optimal electrokinetic parameters for detecting bacterial 16S rRNA in the biosensor cassette based on the current output, signal-to-noise ratio, and limit of detection. In addition, a panel of six probe sets targeting common uropathogenic bacteria was demonstrated. The optimized parameters were also validated using patient-derived clinical urine samples. The effectiveness of electrokinetics for on chip sample preparation will facilitate the implementation of point-of-care diagnosis of urinary tract infection in the future.