AutoKinetics

Right Dose, Right Now

Thanks for your interest in the AutoKinetics approach to drug dosing. Our mission is to use readily available anonimised patient data from electronic patient records to personalize and optimize drug dosing using pharmacokinetic and pharmacodynamic principles and models. In particular, this website serves to support our research project 'Right dose, right now: using big data for bedside personalised drug dosing'. It focus is to optimize dosing of broad spectrum antibiotics in critically ill patients suffering from severe sepsis. For this project we will be cooperating with OLVG Amsterdam, AMC Amsterdam and research groups from Belgium, Canada and Australia.

The project was recently supported by a grant of more than EUR 500.000,= from the Rational Pharmacotherapy Program of ZonMw, The Netherlands Organisation for Health Research and Development. Two PhD students, one with a background in Pharmacy and one with a background in Medicine, will be funded through this contribution.

Further information including eductional background material may be found below.


Press release (Dutch only)

Wereldprimeur: optimaal antibioticagebruik door analyse big data aan bed
Betere behandeling bloedvergiftiging op de intensive care mogelijk

VUmc start een groot onderzoek om de behandeling van intensive care patiënten met bloedvergiftiging te verbeteren. Om voor hen de optimale dosering antibiotica te bepalen wordt ‘real time’ een grote hoeveelheid geanonimiseerde data uit het elektronisch patiëntendossier geanalyseerd. Deze vernieuwende vorm van ‘personalized medicine’ moet leiden tot een daling van het aantal sterfgevallen en sneller herstel. Deze inzet van ‘big data’ aan het bed van de patiënt is een wereldwijde primeur.

Wekelijks belanden driehonderd Nederlanders met een ernstige bloedvergiftiging op de intensive care (IC). Honderd opnames hebben een dodelijke afloop. De patiënt krijgt hoge koorts, gaat snel ademen en raakt uiteindelijk in shock. Antibiotica zijn hier levensreddend. Maar het bepalen van de juiste dosis is lastig omdat ernstig zieke patiënten anders op medicijnen reageren dan onder normale omstandigheden. Uit internationaal onderzoek blijkt dat slechts de helft van alle patiënten met bloedvergiftiging een optimale dosis antibiotica ontvangt.

Juiste dosis antibiotica
Voor het project ‘Right Dose, Right Now’ ontwikkelden VUmc-onderzoekers software die voor elke individuele patiënt op elk moment de optimale dosis antibiotica aangeeft. Het elektronisch patiëntendossier met de gegevens van alle IC-patiënten levert hiervoor de informatie. ‘Intensivisten, de medisch specialisten op de IC, vertrouwen nog altijd op standaarddoseringen voor antibiotica. Maar die zijn veelal gebaseerd op gegevens van vrijwilligers of niet ernstig zieke patiënten. Dat is eigenlijk heel verbazingwekkend’, zegt intensivist en onderzoeksleider Paul Elbers. “Ik pleit voor de inzet van informatie vanuit het elektronische informatiesysteem met miljoenen gegevens van honderden IC-patiënten. Wij kunnen met behulp van deze gegevens de optimale dosering bepalen.’

Hogere sterfte terugdringen
Door in een vroeg stadium de juiste dosis antibiotica toe te dienen, kan de hoge sterfte bij bloedvergiftiging worden teruggedrongen. Tevens verwachten de onderzoekers een kortere verblijfsduur van patiënten op de IC en een vermindering van het aantal complicaties. Zo wordt jaarlijks minimaal 15 miljoen euro bespaard. Ook kan het toedienen van de juiste dosis antibiotica resistentie van bacteriën tegengaan.

Schrikbeeld acute verwardheid
De onderzoekers hebben ook aan ex-patiënten en hun familieleden gevraagd wat zij aan de behandeling verbeterd willen zien. Naast een snellere genezing vroegen zij vooral om het voorkomen van ernstige acute verwardheid, een vaak optredende complicatie bij bloedvergiftiging. Deze gewenste uitkomsten worden meegenomen in het onderzoek.

Nota bene: het project ‘Right Dose Right Now’ werkt samen met het OLVG, AMC en onderzoeksgroepen uit België, Canada en Australië. Het onderzoek wordt mede mogelijk gemaakt door een bijdrage van meer dan 500.000 euro uit het programma ‘Goed Gebruik Geneesmiddelen’ van ZonMw. Deze bijdrage maakt de aanstelling van twee promovendi mogelijk.


Right Dose, Right Now: project summary

Sepsis remains a major healthcare problem. In the Netherlands, 15.000 patients develop severe sepsis each year and 30% of them will die, despite treatment in intensive care units (ICUs). Fast and adequate administration of antibiotics is crucial. However, antibiotic dosing is particularly difficult in the critically ill, as their pharmacokinetic profile is markedly disturbed. Still, doctors invariably use standard dosing schemes. This may cause drug-related toxicity because of overdosing, or treatment failure in case of underdosing.

Therefore, many hospitals rely on Therapeutic Drug Monitoring. However, availability is not universal, and especially lacking at the bedside. This is why we developed AutoKinetics (AutoK) software. AutoK has direct access to the large amount of patient data that is routinely collected in electronic patient records in ICUs, such as data on fluid balance and organ function. The software uses the data in pharmacokinetic models and thus allows for faster and more precise dosing advice. AutoK runs on the computer at the bedside. Thus, advice is readily available, even before treatment is started, and is continuously updated as disease and therapy evolve: true personalised dosing.

We believe that AutoK can improve antibiotic dosing, morbidity and mortality for severe sepsis. Therefore, we intend to study which pharmacokinetic models best predict antibiotic concentrations for three antibiotics using AutoK (ceftriaxone, vancomycin and meropenem). We will then use these in a two-center clinical trial. We will randomise patients with severe sepsis (n=42 per group, per antibiotic), for antibiotic dosing through AutoK or standard therapy. The primary endpoint is attainment of relevant PD targets. Secondary endpoints include morbidity, mortality, quality of life and length of stay. By reducing length of stay, AutoK can lead to a cost saving of more than €50 million per year, in addition to expected improvements for mortality and quality of life.


Principles of Antibiotic Dosing

Depending on the type of antibiotic, different PK parameters have been shown to improve clinical outcome. For time dependent antibiotics such as beta lactams and carbapenems, T > MIC is important. This represents the time for which the concentration of a drug remains above the minimum inhibitory concentration (MIC) during a dosing interval. For concentration dependent antibiotics, such as aminoglycosides Cmax/MIC needs to be targeted. This is the ratio of the peak drug concentration to the MIC of the pathogen. AUC 0-24/MIC is the goals for antibioics with concentration-dependent with time dependence such as fluorquinolones and glycopeptides. This represents the ratio of the area under the concentration versus time curve (AUC) during a 24 h period to the MIC of the pathogen. Adapted from Roberts et al. 2008, CCM, with permission.


Preventing Antibiotic Resistance

Mutant selection window (MSW) and mutant prevention concentration (MPC). The figure illustrates three different situations where an antibiotic is administered. Curves represent the pharmacokinetics (concentration over time) of an antimicrobial agent and squared boxes represent the bacterial population. (A) The pharmacokinetic curve is below the MIC; thus, no selection of a resistant mutant subpopulation within the wild-type population is expected (B) The pharmacokinetic curve is mainly within the MSW; therefore, the resistant mutant subpopulation within the wild-type population can be selectable. (C) The pharmacokinetic curve surpasses the MPC; thus, the susceptible bacteria are inhibited and selection of a resistant mutant subpopulation is potentially avoided. Adapted from Canton et al. 2011, FEMS, with permission.


The mutant selection window

Mutant selection window and mutant prevention concentration. This graph represents reducing bacterial colonies with increased antibiotic exposure for a concentration-dependent antibiotic (e.g., fluoroquinolone). As exposure is increase, a greater reduction in colony forming units is observed. For bacterial colonies to survive the first drop-off at the bacteria's minimum inhibitory concentration (MIC), a first step mutation is required. For bacterial colonies to survive the second drop-off, a second step mutation is required. Selective antibiotic growth may occur when antibiotic concentrations are in the mutant selection window. The mutant prevention concentration requires at least a second-step mutation for bacterial survival. cfu, colony-forming units; Cmax, maximum serum antibiotic concentration. Adapted from Roberts et al. 2008, CCM, with permission.


Why dosing is difficult

Correct dosing requires ample PK knowledge which is often underdeveloped among physicians. In addition, it requires cumbersome calculations as shown in this example for amikacin dosing for empirical nosocomial sepsis for a 60 kg patient with residual glomerular filtration rate of 10 mL/min on continuous veno-venous haemodiafiltration (CVVHDF) using an AN69 filter and with targeted total effluent rate of 35 mL/kg/h. A minimum inhibitory concentration (MIC) breakpoint of 4 mg/L is used for illustrative purposes. CRRT, continuous renal replacement therapy; CLCVVHDF, drug clearance from CVVHDF; Qf , ultrafiltrate flow rate; Qd, dialysate flow rate; Cmax, maximum drug concentration. Adapted from Roberts et al. 2012, IJAA, with permission.


AutoKinetics as a solution using Big Data

As a solution to these obstacles, we designed AutoKinetics (AK) software. Combining many scientific domains, AK interacts with electronic patient records (EPRs). In the context of PK, EPRs are essentially data warehouses: many and precise patient data determining es their PK profile is instantly available. These include patient vital signs, demographics, diagnoses, and exact information on time and dose of drug administration.

These data are automatically fed in to AK and used to provide physicians with individualized, pre-emptive, real-time, graphical PK feedback and prediction modules to guide drug dosing. AK provides graphical predictions that are available even before starting drug therapy. Thus, AK helps to provide the right dose for the right patient right from the start of treatment. Interestingly, the vast availibilty of ICT infrastructure in intensive care, implies that the patient group most prone to AMR may actually be closest to a solution.

We have developed a beta version of AK for vancomycin which is currently functioning at OLVG hospital. It greatly improves succesfully targeting PK parameters and boasts a reduction to zero in prediction delay.


The pitfalls of Therapeutic Drug Monitoring

Therapeutic drug monitoring, if available at all, does not routinely focus on clinical efficacy or ARM prevention and is usually performed after a delay of days. It is typically performed far from the bedside by professionals unaware of the patient, indication or targeted pathogen. AutoKinetics solves these problems.


AutoKinetics can use more data

Therapeutic drug monitoring relies on manual data entry, which precludes complicated models and is prone to errors. ID and TDM are even more difficult in intensive care unit (ICU) patients who have markedly altered PK profiles because of multiple organ failure, organ support, augmented renal clearance, altered cardiac output and blood flow distribution as well as varying fluid balance. In addition, drug models for the ICU are even more sparse and inaccurate. This is problematic as ICU patients are prone to AMR development because of vast exposure to antibiotics and various extent of immune deficiencies. AutoKinetics solves these problems.


A bright future

The potential for exploitation is enormous. The widespread and increasing availability of ICT infrastructure in hospitals worldwide facilitate fast and easy dissemination. The predict concept is applicable both for existing and new antibiotics and can also be used by non-automated ICUs are hospital wards using the internet. Interaction with MIC and MPC databases such as EUCAST or CANWARD to finetune dosing advies is easy. The use of global feedback loops will continuously ameliorate the concept and benefit science and patients to conquer AMR.

info@autokinetics.eu     +31 (0)20 444 36 97