Breathalyser-Based eHealth Data Suggest That Self-Reporting of Abstinence Is a Poor Outcome Measure for Alcohol Use Disorder Clinical Trials.

AIMS: To evaluate the efficacy and monitoring capabilities of a breathalyser-based eHealth system for patients with alcohol use disorder (AUD) and to investigate the quality and validity of timeline follow-back (TLFB) as outcome measure in clinical trials and treatment. METHODS: Patients (n = 115) were recruited to clinical trials from a 12-step aftercare programme (12S-ABS) and from hospital care with abstinence (HC-ABS) or controlled drinking (HC-CDR) as goal and randomly divided into an eHealth and a control group. The effect of the eHealth system was analysed with TLFB-derived primary outcomes-change in number of abstinent days (AbsDay) and heavy drinking days (HDDs) compared to baseline-and phosphatidyl ethanol (PEth) measurements. Validity and quality of TLFB were evaluated by comparison with breath alcohol content (BrAC) and eHealth digital biomarkers (DBs): Addiction Monitoring Index (AMI) and Maximum Time Between Tests (MTBT). TLFB reports were compared to eHealth data regarding reported abstinence. RESULTS: The primary outcome (TLFB) showed no significant difference between eHealth and control groups, but PEth did show a significant difference especially at months 2 and 3. Self-reported daily abstinence suffered from severe quality issues: of the 28-day TLFB reports showing full abstinence eHealth data falsified 34% (BrAC measurements), 39% (MTBT), 54% (AMI) and 68% (BrAC/MTBT/AMI). 12S-ABS and HC-ABS patients showed severe under-reporting. CONCLUSIONS: No effect of the eHealth system was measured with TLFB, but a small positive effect was measured with PEth. The eHealth system revealed severe quality problems with TLFB, especially regarding abstinence-should measurement-based eHealth data replace TLFB as outcome measure for AUD?

Maximum Time Between Tests: A Digital Biomarker to Detect Therapy Compliance and Assess Schedule Quality in Measurement-Based eHealth Systems for Alcohol Use Disorder.

AIM: To evaluate, in a breathalyzer-based eHealth system, whether the time-based digital biomarker 'maximum time between tests' (MTBT) brings valuable information on alcohol consumption patterns as confirmed by correlation with blood phosphatidyl ethanol (PEth), serum carbohydrate deficient transferrin (CDT) and timeline follow-back data. METHOD: Data on 54 patients in follow-up for treatment of alcohol use disorder were analysed. RESULTS: The model of weekly averages of 24-log transformed MTBT adequately described timeline follow-back data (P  <  0.0001, R =  0.27-0.38, n  =  650). Significant correlations were noted between MTBT and PEth (P  <  0.0001, R  =  0.41, n  =  148) and between MTBT and CDT (P  <  0.0079, R  =  0.22, n  =  120). CONCLUSIONS: The time-based digital biomarker 'maximum time between tests' described here has the potential to become a generally useful metric for all scheduled measurement-based eHealth systems to monitor test behaviour and compliance, factors important for 'dosing' of eHealth systems and for early prediction and interventions of lapse/relapse.

Real-time Monitoring using a breathalyzer-based eHealth system can identify lapse/relapse patterns in alcohol use disorder Patients.

AIM: We introduce a new remote real-time breathalyzer-based method for monitoring and early identification of lapse/relapse patterns for alcohol use disorder (AUD) patients using a composite measure of sobriety, the Addiction Monitoring Index (AMI). METHODS: We constructed AMI from (a) obtained test results and (b) the pattern of ignored tests using data from the first 30 patients starting in the treatment arms of two on-going clinical trials. The patients performed 2-4 scheduled breath alcohol content (BrAC)-tests per day presented as blood alcohol content (BAC) data. In total, 10,973 tests were scheduled, 7743 were performed and 3230 were ignored during 3982 patient days. RESULTS: AMI-time profiles could be used to monitor the daily trends of alcohol consumption and detect early signs of lapse and relapses. The pattern of ignored tests correlates with the onset of drinking. AMI correlated with phosphatidyl ethanol (n = 61, F-ratio = 34.6, P < 0.0001, R = -0.61). The recognition of secret drinking could further be improved using a low alcohol detection threshold (BrAC = 0.025 mg/l, BACSwe = 0.05‰ or US = 0.0053g/dl), in addition to the legal Swedish traffic limit (BrAC = 0.1 mg/l, BACSwe = 0.2‰ or US = 0.021 g/dl). Nine out of 10 patients who dropped out from the study showed early risk signs as reflected in the level and pattern in AMI before the actual dropout. CONCLUSIONS: AMI-time profiles from an eHealth system are useful for monitoring the recovery process and for early identification of lapse/relapse patterns. High-resolution monitoring of sobriety enables new measurement-based treatment methods for proactive personalized long-term relapse prevention and treatment of AUD patients. CLINICAL TRIAL REGISTRATION: The data used for construction of AMI was from two clinical trials approved by the Regional Ethics Committee of Uppsala, Sweden and performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participating subjects. The study was registered at ClinicalTrials.gov (NCT03195894).

Evaluation of 6 years of eHealth data in the alcohol use disorder field indicates improved efficacy of care.

Background: Predictive eHealth tools will change the field of medicine, however long-term data is scarce. Here, we report findings on data collected over 6 years with an AI-based eHealth system for supporting the treatment of alcohol use disorder. Methods: Since the deployment of Previct Alcohol, structured data has been archived in a data warehouse, currently comprising 505,641 patient days. The frequencies of relapse and caregiver-patient messaging over time was studied. The effects of both introducing an AI-driven relapse prediction tool and the COVID-19 pandemic were analyzed. Results: The relapse frequency per patient day among Previct Alcohol users was 0.28 in 2016, 0.22 in 2020 and 0.25 in 2022 with no drastic change during COVID-19. When a relapse was predicted, the actual occurrence of relapse in the days immediately after was found to be above average. Additionally, there was a noticeable increase in caregiver interactions following these predictions. When caregivers were not informed of these predictions, the risk of relapse was found to be higher compared to when the prediction tool was actively being used. The prediction tool decreased the relapse risk by 9% for relapses that were of short duration and by 18% for relapses that lasted more than 3 days. Conclusions: The eHealth system Previct Alcohol allows for high resolution measurements, enabling precise identifications of relapse patterns and follow up on individual and population-based alcohol use disorder treatment. eHealth relapse prediction aids the caregiver to act timely, which reduces, delays, and shortens relapses.

Processing incomplete questionnaire data into continuous digital biomarkers for addiction monitoring.

PURPOSE: eHealth systems allow efficient daily smartphone-based collection of self-reported data on mood, wellbeing, routines, and motivation; however, missing data is frequent. Within addictive disorders, missing data may reflect lack of motivation to stay sober. We hypothesize that qualitative questionnaire data contains valuable information, which after proper handling of missing data becomes more useful for practitioners. METHODS: Anonymized data from daily questionnaires containing 11 questions was collected with an eHealth system for 751 patients with alcohol use disorder (AUD). Two digital continuous biomarkers were composed from 9 wellbeing questions (WeBe-i) and from two questions representing motivation/self-confidence to remain sober (MotSC-i). To investigate possible loss of information in the process of composing the digital biomarkers, performance of neural networks to predict exacerbation events (relapse) in alcohol use disorder was compared. RESULTS: Long short-term memory (LSTM) neural networks predicted a coming exacerbation event 1-3 days (AUC 0.68-0.70) and 5-7 days (AUC 0.65-0.68) in advance on unseen patients. The predictive capability of digital biomarkers and raw questionnaire data was equal, indicating no loss of information. The transformation into digital biomarkers enable a continuous graphical display of each patient's clinical course and a combined interpretation of qualitative and quantitative aspects of recovery on a time scale. CONCLUSION: By transforming questionnaire data with large proportion of missing data into continuous digital biomarkers, the information captured by questionnaires can be more easily used in clinical practice. Information, assessed by the capability to predict exacerbation events of AUD, is preserved when processing raw questionnaire data into digital biomarkers.

The Clinical Course of Alcohol Use Disorder Depicted by Digital Biomarkers.

Aims: This study introduces new digital biomarkers to be used as precise, objective tools to measure and describe the clinical course of patients with alcohol use disorder (AUD). Methods: An algorithm is outlined for the calculation of a new digital biomarker, the recovery and exacerbation index (REI), which describes the current trend in a patient's clinical course of AUD. A threshold applied to the REI identifies the starting point and the length of an exacerbation event (EE). The disease patterns and periodicity are described by the number, length, and distance between EEs. The algorithms were tested on data from patients from previous clinical trials (n = 51) and clinical practice (n = 1,717). Results: Our study indicates that the digital biomarker-based description of the clinical course of AUD might be superior to the traditional self-reported relapse/remission concept and conventional biomarkers due to higher data quality (alcohol measured) and time resolution. We found that EEs and the REI introduce distinct tools to identify qualitative and quantitative differences in drinking patterns (drinks per drinking day, phosphatidyl ethanol levels, weekday and holiday patterns) and effect of treatment time. Conclusions: This study indicates that the disease state-level, trend and periodicity-can be mathematically described and visualized with digital biomarkers, thereby improving knowledge about the clinical course of AUD and enabling clinical decision-making and adaptive care. The algorithms provide a basis for machine-learning-driven research that might also be applied for other disorders where daily data are available from digital health systems.

Screening for NNRTIs with slow dissociation and high affinity for a panel of HIV-1 RT variants.

A lead optimization library consisting of 800 HIV-1 nonnucleoside reverse transcriptase inhibitors (NNRTIs) was screened in parallel against 4 clinically relevant variants of HIV-1 RT (Wt, L100I, Y181C, and K103N) using a surface plasmon resonance-based biosensor. The aim was to identify inhibitors suitable in specific topical microbicides efficient for preventing the transmission of a range of clinically significant strains of HIV-1. The authors hypothesized that such compounds should have high affinity and slow dissociation rates for multiple variants of the target. To efficiently analyze the large amount of real-time data (sensorgrams) that were generated in the screening, they initially used signals from 3 selected time points to identify compounds with high affinity and slow dissociation for the complete panel of enzyme variants. Hits were confirmed by visually inspecting the complete sensorgrams. Two structurally unrelated compounds fulfilled the hit criteria, but only 1 compound was found to (a) compete with a known NNRTI for binding to the NNRTI site, (b) inhibit HIV-1 RT activity, and (c) inhibit HIV-1 replication in cell culture, for all 4 enzyme variants. This novel screening methodology offers high-resolution real-time kinetic data for multiple targets in parallel. It is expected to have broad applicability for the discovery of compounds with defined kinetic profiles, crucial for optimal therapeutic effects.

Identification of MMP-12 inhibitors by using biosensor-based screening of a fragment library.

Small inhibitors of matrix metalloproteinase 12 (MMP-12) have been identified with a biosensor-based screening strategy and a specifically designed fragment library. The interaction between fragments and three variants of the target and a reference protein with an active-site zinc ion was measured continuously by surface plasmon resonance. The developed experimental design overcame the inherent instability of MMP-12 and allowed the identification of fragments that interacted specifically with the active-site of MMP-12 but not with the reference protein. The interaction with MMP-12 for selected compounds were analyzed for concentration dependence and saturability. Compounds interacting distinctly with the target were further evaluated by an activity-based assay, verifying MMP-12 inhibition. Two effective inhibitors were identified, and the compound with highest affinity was confirmed to be a competitive inhibitor with an IC50 of 290 nM and a ligand efficiency of 0.7 kcal/mol heavy atom. This procedure integrates selectivity and binding site identification into the screening procedure and does not require structure determination.

Label-free primary screening and affinity ranking of fragment libraries using parallel analysis of protein panels.

The authors present fragment screening data obtained using a label-free parallel analysis approach where the binding of fragment library compounds to 4 different target proteins can be screened simultaneously using surface plasmon resonance detection. They suggest this method as a first step in fragment screening to identify and select binders, reducing the demanding requirements on subsequent X-ray or nuclear magnetic resonance studies, and as a valuable "clean-up" tool to eliminate unwanted promiscuous binders from libraries. A small directed fragment library of known thrombin binders and a general 500-compound fragment library were used in this study. Thrombin, blocked thrombin, carbonic anhydrase, and glutathione-S-transferase were immobilized on the sensor chip surface, and the direct binding of the fragments was studied in real time. Only 12 microg of each protein is needed for screening of a 3000-compound fragment library. For screening, a binding site-blocked target as reference facilitates the identification of binding site-selective hits and the signals from other reference proteins for the elimination of false positives. The scope and limitations of this screening approach are discussed for both target-directed and general fragment libraries.

Biosensor analysis of the interaction between drug compounds and liposomes of different properties; a two-dimensional characterization tool for estimation of membrane absorption.

The interactions between 78 drug compounds and immobilised liposomes were investigated using an assay based on surface plasmon resonance technology. The drugs were screened at a single concentration and allowed to interact simultaneously with two different types of liposomes. When the drug-liposome responses are plotted against one another they generally fall into three distinct bands: low response-low percent fraction absorbed in humans (Fa), medium response-medium Fa, and high response-high Fa. For drugs with medium to high Fa values, basic compounds could be resolved from acidic and neutral compounds to a large extent. This technique has the potential to be utilized as a screening tool for binning novel compounds into low, medium, or high Fa based on a simple experimental measurement. The assay was applied to 11 kinase inhibitors, 9 thrombin inhibitors, and 11 carbonic anhydrase inhibitors highlighting a subset that may have incomplete intestinal absorption (low to medium Fa). Assay conditions were optimized making the assay suitable for routine analysis and for compound characterization early in drug discovery where solubility may be an issue.

Relationships between structure and interaction kinetics for HIV-1 protease inhibitors.

The interaction between HIV-1 protease and 58 structurally diverse transition-state analogue inhibitors has been analyzed by a surface plasmon resonance based biosensor. Association and dissociation rate constants and affinities were determined and displayed as k(on)-k(off)-K(D) maps. It was shown that different classes of inhibitors fall into distinct clusters in these maps. Significant changes in association and dissociation rates were found as a result of modifying the P1/P1' or P2/P2' side chains of a linear lead compound. Similarly, cyclic urea and cyclic sulfamide inhibitors displayed different kinetic features and the affinities of both classes of cyclic compounds were limited by fast dissociation rates. These results confirm that association and dissociation rates are important features of drug-target interactions and indicate that optimization of inhibitor efficacy may be guided by aiming for high association and low dissociation rates rather than high affinity alone. The present approach thus provides a new tool for structure-interaction kinetic analysis and drug discovery.

Elucidation of HIV-1 protease resistance by characterization of interaction kinetics between inhibitors and enzyme variants.

The kinetics of the interaction between drug-resistant variants of HIV-1 protease (G48V, V82A, L90M, I84V/L90M, and G48V/V82A/I84V/L90M) and clinically used inhibitors (amprenavir, indinavir, nelfinavir, ritonavir, and saquinavir) were determined using biosensor technology. The enzyme variants were immobilized on a biosensor chip and the association and dissociation rate constants (k(on) and k(off)) and affinities (K(D)) for interactions with inhibitors were determined. A unique interaction kinetic profile was observed for each variant/inhibitor combination. Substitution of single amino acids in the protease primarily resulted in reduced affinity through increased k(off) for the inhibitors. For inhibitors characterized by fast association rates to wild-type protease (ritonavir, amprenavir, and indinavir), additional substitutions resulted in a further reduction of affinity by a combination of decreased k(on) and increased k(off). For inhibitors characterized by slow dissociation rates to wild-type enzyme (saquinavir and nelfinavir), the decrease of affinity conferred by additional mutations was attributed to increased k(off) values. Development of resistance thus appears to be associated with a change of the distinctive kinetic parameter contributing to high affinity. Further inhibitor design should focus on improving the "weak point" of the lead compound, that being either k(on) or k(off).

Increased sensitivity of SPR assays in plasma through efficient parallel assay optimization.

The sensitivity of biosensor assays in complex media such as plasma or serum is often limited by non-specific binding. The degree of binding often varies between individuals and therefore a large number of different plasma samples have to be used during assay development. Some surface plasmon resonance (SPR) biosensors allow for parallel screening of several running buffer compositions, with a number of different immobilization levels for each buffer. These technical possibilities combined with statistical design of experiments (DoE) enable efficient parallel optimization of multiple assay conditions. In this paper we outline how to increase the sensitivity in SPR-based assays by minimizing background binding and variability from negative control plasma while retaining high signals from positive samples. To mimic immunogenicity studies of biotherapeutics we have used a model assay with anti-rituximab as an anti-drug antibody to be detected in plasma by binding to immobilized rituximab. Immobilization level and sodium chloride concentration were found to be the most important factors to optimize. There were also a number of significant interaction effects and strong non-linearites between the buffer composition/immobilization level and the assay performance, which necessitated DoE based optimization strategies. The applicability of the optimized conditions was verified with several assays (anti-erythropoietin, omalizumab, anti-IgE and anti-myoglobin) in spiked plasma samples resulting in detection levels in the range of 80-170 ng ml(-1). The buffer conditions presented in this paper can be used for other immunogenicity assays on biosensor platforms or as a good starting point for further assay development for new immunogenicity assays.

Experimental validation of a fragment library for lead discovery using SPR biosensor technology.

A new fragment library for lead discovery has been designed and experimentally validated for use in surface plasmon resonance (SPR) biosensor-based screening. The 930 compounds in the library were selected from 4.6 million commercially available compounds using a series of physicochemical and medicinal chemistry filters. They were screened against 3 prototypical drug targets: HIV-1 protease, thrombin and carbonic anhydrase, and a nontarget: human serum albumin. Compound solubility was not a problem under the conditions used for screening. The high sensitivity of the sensor surfaces allowed the detection of interactions for 35% to 97% of the fragments, depending on the target protein. None of the fragments was promiscuous (i.e., interacted with a stoichiometry ≥5:1 with all 4 proteins), and only 2 compounds dissociated slowly from all 4 proteins. The use of several targets proved valuable since several compounds would have been disqualified from the library on the grounds of promiscuity if fewer target proteins had been used. The experimental procedure allowed an efficient evaluation and exploration of the new fragment library and confirmed that the new library is suitable for SPR biosensor-based screening.

Compounds binding to the S2-S3 pockets of thrombin.

A set of compounds designed to bind to the S2-S3 pockets of thrombin was prepared. These compounds included examples with no interactions in the S1 pocket. Proline, a common P2 in many thrombin inhibitors, was combined with known P3 residues and P1 substituents of varying size and lipophilicity. Binding constants were determined using surface plasmon resonance (SPR) biosensor technology and were found to be in good agreement with results from an enzyme assay. A dramatic increase in affinity (100-1000 times) was seen for compounds incorporating an amino group capable of forming a hydrogen bond with gly216 in the protein backbone. The ligand efficiency was increased by including substituents that form stronger hydrophobic interactions with the P1 pocket. The binding mode was confirmed by X-ray analysis, which revealed the anticipated binding motif that included hydrogen bonds as well as a tightly bound water molecule. A QSAR model indicated that hydrogen bonding and lipophilicity were important for the prediction of binding constants. The results described here may have implications for how directed compound libraries for shallow protein pockets, like S2 and S3 in serine proteases, can be designed.

Label-free kinetic binding data as a decisive element in drug discovery.

The emerging possibilities to obtain label-free, kinetic-based binding data for drug-target and drug absorption, distribution, metabolism and excretion (ADME)-marker interactions have proven useful in many drug discovery related issues. Multiple reports have demonstrated that the common use of affinity as an early measure of drug potency may be directly misleading. This review summarises findings in the literature related to compound selection in the drug discovery process. It is important to understand the different properties of association and dissociation rates, the former being related to both structure and dosage and the latter depending solely on molecular structure. By performing parallel optimisations of association and dissociation rates, compounds with desirable kinetic profiles for target binding may be generated. In addition, compound selection may also consider the kinetic properties of the drug-ADME-marker binding profiles, further refining the quality of compounds early in the drug discovery process. The promising results found in the literature indicate that kinetic data on drug-protein interactions may soon become a crucial decisive element in modern drug discovery.

Kinetic and thermodynamic characterization of HIV-1 protease inhibitors.

Interaction kinetic and thermodynamic analyses provide information beyond that obtained in general inhibition studies, and may contribute to the design of improved inhibitors and increased understanding of molecular interactions. Thus, a biosensor-based method was used to characterize the interactions between HIV-1 protease and seven inhibitors, revealing distinguishing kinetic and thermodynamic characteristics for the inhibitors. Lopinavir had fast association and the highest affinity of the tested compounds, and the interaction kinetics were less temperature-dependent as compared with the other inhibitors. Amprenavir, indinavir and ritonavir showed non-linear temperature dependencies of the kinetics. The free energy, enthalpy and entropy (DeltaG, DeltaH, DeltaS) were determined, and the energetics of complex association (DeltaG(on), DeltaH(on), DeltaS(on)) and dissociation (DeltaG(off), DeltaH(off), DeltaS(off)) were resolved. In general, the energetics for the studied inhibitors was in the same range, with the negative free energy change (DeltaG < 0) due primarily to increased entropy (DeltaS > 0). Thus, the driving force of the interaction was increased degrees of freedom in the system (entropy) rather than the formation of bonds between the enzyme and inhibitor (enthalpy). Although the DeltaG(on) and DeltaG(off) were in the same range for all inhibitors, the enthalpy and entropy terms contributed differently to association and dissociation, distinguishing these phases energetically. Dissociation was accompanied by positive enthalpy (DeltaH(off) > 0) and negative entropy (DeltaS(off) < 0) changes, whereas association for all inhibitors except lopinavir had positive entropy changes (DeltaS(on) > 0), demonstrating unique energetic characteristics for lopinavir. This study indicates that this type of data will be useful for the characterization of target-ligand interactions and the development of new inhibitors of HIV-1 protease.

QSAR studies applied to the prediction of antigen-antibody interaction kinetics as measured by BIACORE.

The objective of this work was to investigate the potential of the quantitative structure-activity relationships (QSAR) approach for predictive modulation of molecular interaction kinetics. A multivariate QSAR approach involving modifications in peptide sequence and buffer composition was recently used in an attempt to predict the kinetics of peptide-antibody interactions as measured by BIACORE. Quantitative buffer-kinetics relationships (QBKR) and quantitative sequence-kinetics relationships (QSKR) models were developed. Their predictive capacity was investigated in this study by comparing predicted and observed kinetic dissociation parameters (k(d)) for new antigenic peptides, or in new buffers. The range of experimentally measured k(d) variations was small (300-fold), limiting the practical value of the approach for this particular interaction. However, the models were validated from a statistical point of view. In QSKR, the leave-one-out cross validation gave Q(2) = 0.71 for 24 peptides (all but one outlier), compared to 0.81 for 17 training peptides. A more precise model (Q(2) = 0.92) could be developed when removing sets of peptides sharing distinctive structural features, suggesting that different peptides use slightly different binding modes. All models share the most important factor and are informative for structure-kinetics relationships. In QBKR, the measured effect on k(d) of individual additives in the buffers was consistent with the effect predicted from multivariate buffers. Our results open new perspectives for the predictive optimization of interaction kinetics, with important implications in pharmacology and biotechnology.