Final Report Summary - EU-CLAMP (Clinical Application for Metabolic Profiling)
Executive Summary:
Europe faces a diabetes epidemic. More than 55 million people in Europe are currently diagnosed with diabetes and with an estimated 20% increase by 2030, the disease is certain to stay one of the most challenging health problems this century. Especially as diabetes no longer is a disease exclusively for adults, but affects children, young people and adults of all ages.
Despite the high prevalence of diabetes, the choice of anti-diabetic drugs is still limited and two thirds of patients with diabetes do not achieve the recommended glycaemic target levels. For each new anti-diabetic drug, it is essential to investigate the metabolic effect over time. The glucose clamp technique is regarded as the gold standard to evaluate the effectiveness of new anti-diabetic drugs. There are however only a few centres with limited research capacities that have experience in using the clamp technique, because automated clamp devices are no longer commercially available and the existing techniques are confounded by a number of limitations. Increased clamp capacities are urgently needed for the development of new, more efficacious anti-diabetic drugs.
Through the integration of newly emerging technologies proposed in part by the 4 participating SME organisations and with the outsourced research capacity of 3 of Europe’s leading RTD performers, the EU-CLAMP project aims to develop a new generation automated clamp device that will overcome the limitations of the existing devices by incorporating an intravenous microdialysis technique for reliable continuous glucose monitoring without blood loss. The EU-CLAMP project will facilitate clamp testing in a more efficient and cost effective manner. In addition to the significant contribution that could be made to development of improved treatment options for diabetes, the project will provide a platform from which the competitiveness of the participating SMEs can be improved offering alignment to the needs of their long term business strategies.
Project Context and Objectives:
The principle of the glucose clamp is based on the idea that a fall in blood glucose concentrations (e.g. induced by pharmaceutical agents) is prevented by a variable infusion of glucose. The amount of glucose needed to stabilise blood glucose concentrations (expressed as glucose infusion rates) is a direct measure of the metabolic effect of the compound. Glucose clamps can be performed either manually or automatically. During manual clamps, the investigator frequently samples the blood for glucose content and manually adjusts the rate of glucose infusion on the basis of the blood glucose measurements. As the automated clamp technique has obvious advantages over the manual technique (i.e. no user-bias, less labour intensive and higher clamp quality due to continuous glucose monitoring), only the automated clamp technique will be considered in the remainder of this report.
The glucose clamp technique is regarded as the gold standard for the assessment of the metabolic profiles of anti-diabetic compounds and insulin sensitivity. Currently, three devices for automated clamps have made it onto the market all with their own limitations:
1. The Biostator (Glucose controlled insulin infusion system, Miles Laboratories Inc., Elkhart, IN, USA). Developed in the 1970’s, the device was only sold for a short period of time, and consequently the availability of components is limited. Further limitations include glucose monitoring capabilities. The device uses an electrochemical membrane for continuous blood glucose monitoring which often shows drifts, even with frequent calibrations. Therefore, it is mandatory to check blood glucose concentrations with an external blood glucose analyser at least every 30 min. The blood loss for the continuous measurement and the external checks is about 2.5 ml/h. Because of a high incidence of technical errors, experienced staff are needed to perform the clamp experiments.
2. The Glucostator is a recently developed glucose clamp device, but does not overcome the limitations of the Biostator. The Glucostator uses similar principles for continuous blood glucose monitoring and the same algorithm for the calculation of glucose infusion rates as the Biostator. Therefore, the limitations with regard to blood loss, costs, labour intensity and quality of glucose monitoring (and consecutively variability of the glucose clamp) remain the same.
3. Nikkiso STG-22 (Nikkiso Co., Tokyo, Japan), a Japanese device that is not licensed for use in Europe or the USA.
The overarching aim of the EU-CLAMP project: to develop a prototype of a new generation clamp device. The new clamp device will overcome the current limitations of existing clamp techniques and is needed to evaluate the effectiveness of new anti-diabetic drugs in a clinical setting. To progress towards this aim, the following Research and Development objectives were defined:
1. To develop the individual components of the clamp device according to the user requirements. Four components have been identified: body-interface (iv microdialysis catheter), microfluidics, glucose sensor & electronics and control algorithm. The components will be individually tested in a laboratory setting at JR and MUG. The control algorithm will be developed or adjusted from an existing algorithm, and tested by UCAM (see also objective 3). On the basis of the outcome of the laboratory tests, the performance of the components may be optimised by the relevant consortium partners.
2. To combine the novel iv microdialysis technique with an online glucose sensor to develop a module for the continuous monitoring of blood glucose. PROBE’s iv microdialysis catheter will be linked to a BVT glucose sensor via the microfluidic technology developed by JR. Performance tests (accuracy, detection limit, linearity, variability) of the monitoring module will be performed by MUG in healthy human volunteers. The accuracy of the glucose measurements will be compared to a reference laboratory method.
3. To combine the control algorithm with commercially available high precision infusion pumps. In vivo measurements of this infusion module of the clamp device might be performed and used to optimise the control algorithm.
4. To construct a prototype new generation clamp device. Integration of the components and modules into a prototype clamp device will be lead by DATAMED.
5. To test the safety and performance of the prototype clamp device during a series of clinical tests under medical supervision at PROFIL. In these tests with healthy human volunteers and patients with diabetes safety parameters will be documented. The performance of the prototype new generation clamp device will be compared to the current ‘gold-standard’ clamp device, the Biostator, with respect to measurement accuracy, clamping precision (i.e. blood glucose control), blood loss and reliability.
The performance characteristics of the prototype device that EU-CLAMP project aims to achieve are:
• No blood loss
• Accurate and precise continuous glucose monitoring without signal drifts with modern laboratory standards
• Continuous glucose monitoring profile without time-delays
• Continuous glucose infusion profile
• Stabilisation of blood glucose concentrations at the target level with variability < 5%
• Automated electronic data capturing system for blood glucose concentrations and glucose/insulin infusion rates meeting regulatory requirements.
• Reliable technical performance using modern technology
• Reduced risk of infection (for nurses and physicians)
• Easy-to-use, less labour-intensive automated device
Project Results:
REFERENCE METHODS
-----------------------------
CMA 600: Maintenance of the CMA 600 did not improve its accuracy and precision. Its coefficient of variation (CV) was in some cases above 5% - up to 5 measurements per sample had to be performed in order to get reliable results; therefore the work with the CMA 600 was stopped and alternatives were investigated.
Super GL2: Our results for 0.9% saline solution as well as 5% Mannitol revealed that with the Super GL2 glucose levels down to 1.25mg/dl with a CV of less than 5% (in most cases approx. 1-2%) can be quantified. If concentrations below 11mg/dl have to be measured a modified procedure was used, namely the 20-fold volume (400µl) was injected. Thus the results have to be corrected for volume difference in the Glucocapil container (1400µl/1020µl) and have to be divided by the factor 20 (for 20-fold amount of requested glucose).
Conductivity measurement: The contactless conductivity measurement device TraceDec from I.S.T. was found to have a CVs smaller than 2% (Min: 0.61 Max: 1.89%).
Different Heparin concentrations of Heparin Immuno and Arixtra show very little (<4.8%) influence on the measured glucose concentrations, as no trend of the glucose concentration with increasing Heparin concentration was observed.
Different Heparin concentrations of Heparin Immuno and Arixtra show an influence on the measured conductivity signal. This fact has to be considered if the ionic reference technique is applied as values derived should be corrected for these values.
CHARACTERISATION OF PUMPS
-----------------------------
Both pumps (JR-pump and the BBRAUN Perfusor) show a pulsatile flow
When a microdialysis probe is connected to the pump the JR-pump shows a pulsatile flow due its push/pull operation mode in contrast to the BBRAUN Perfusor piston pump that shows a non-pulsatile flow.
Mean flow rates with or without body interface are similar for both pumps, thus no perfusate is lost over the membrane.
Pressure within the tubing system needs to equilibrate for more than 15min if the flow changes from 250 to 0µ/min and more than 5 min if the flow changes from 0 to 250µ/min, respectively.
The tubing system of the JR-pump showed wearing signs after a 24h experiment when being operated at a nominal flow of 10µl/min.
The Emmerich battery shows a life-time of around 16.5h whereas the EVE battery shows a life time of around 19.45h respectively. Thus at a maximal nominal flow rate of 10µl/min batteries have to be changed within one 24h experiment.
At a nominal flow rate of 10µl/min the pump tubing are extremely stressed but can be used at least for 24h. Occasionally we found lose bonding point of the tubing system.
BODY-INTERFACE
-----------------------------
Recovery of glucose and ions correlate very well which is a prerequisite for the ionic reference technique
For most cases the major decrease of recovery was between 1-50 µl/min (except for the CMA 64 in blood – there was a significant lower recovery for flow rates > 10µl/min.
Glucose recovery does not depend on the composition of the perfusate.
The major decrease of recovery was between 1-50 µl/min
A mathematical function can be found for the relationship of glucose and ion recovery (polynomial function of 3rd order).
A dis-configuration of the µEye PME 011 was found if the probe became wet
DEPENDENCY OF THE GLUCOSE SENSOR CURRENT ON IONIC STRENGTH
-----------------------------
The calibration curves of the glucose sensor are linear up to 20mg/dl.
The different ion concentrations influence the sensor signal. On average (mean over 8 sensors), and as already prior experiments have shown, a lower ion concentration leads to a higher signal current.
The influence of the ions gets less for higher ion concentrations: there is nearly no difference between the calibration curve between 15 and 20 % ions
The perfusate must contain a certain amount of ions (e.g. 20% NaCl: 8 parts 5% Mannitol and 2 parts 0.9% NaCl solution).
GLUCOSE SAMPLING UNIT – PRESSURE
-----------------------------
With increasing flow-rates the pressure within the system is increasing.
Flow through cell and sensor show nearly no back pressure.
The CMA 64 shows more back pressure than the µEye.
Pressure of up to 540mmH2O might significantly influence the recovery of ions and glucose.
At flow rates of 10µl/min the system can be operated with the BBRAUN Space or Perfusor fm pump.
GLUCOSE SAMPLING UNIT – FLOW
-----------------------------
A decreasing flow could be observed for peristaltic JR pump at a flow rate of 10µl/min, caused by the running-in characteristic of the tubing set.
Using a push-pull setup did not improve the flow stability and its variations if no sensor + sensor cell is attached to the bodyinterface. On the other hand the push setup of the BBRAUN Perfusor Space showed a constant flow and less variation.
BBRAUN Perfusor Space is the more convincing setup as the pumps are not attached to the subject`s arms and thus allows unrestricted movements of the subjects. Furthermore the system is well known to medical health personal and easier to handle.
GLUCOSE SAMPLING UNIT – SETUP
-----------------------------
A mean recovery lower than 5% is associated with a poor correlation (< 0.8) and thus results in a bad performance of the system.
37.5% of the µEye catheters showed a good correlation coefficient or more than 0.8.
93.3% of CMA-catheters showed a good correlation coefficient of more than 0.8.
All probes with systemic anticoagulation (Arixtra®) showed a good correlation of more than 0.8
The data indicate that adding Arixtra® only to the perfusate improves the recovery and thus might increase the correlation.
CMA body-interfaces gained generally higher recovery rates than µEyes.
The µEyes are more prone for thrombus formation around the membrane than the CMAs.
Fondaparinux (ARIXTRA®®) seems to be advantageous compared to Heparin due to its smaller molecular weight of 1.7 kDa instead of 4-6 kDa.
The data indicate that a drop of the recovery can be associated with a settlement of bodily components (e.g. plate lets, proteins, etc.) around the body-interface.
In order to prevent a thrombus formation an anticoagulation drug should be used either systemically, locally (addition to perfusate) or both.
GLUCOSE SAMPLING UNIT – ULTRASOUND INVESTIGATIONS
-----------------------------
In total 25 ultrasound investigations were performed: 9 CMA and 16 µEye, respectively.
In 6 CMA catheters no thrombus formation was found whereas in 3 thrombus formations were found (for more details see Appendix).
In 10 µEye catheters no thrombus formation was found whereas in 6 thrombus formations were found (for more details see Appendix).
GLUCOSE SAMPLING UNIT – IONIC REFERENCE TECHNIQUE
-----------------------------
A significant decrease of the recovery can be (partially) compensated with IRT.
Changes in the recovery (flow rate variations, movement, changing diffusion barrier, etc.) can be (partially) compensated with the IRT.
Overall performance (correlation) can be significantly increased with the IRT (33 out of 39 systems).
Recovery should be > 5% in order to achieve proper results to apply the IRT.
Higher recoveries result in better correlation between glucose concentrations found in blood and glucose values calculated from the glucose concentrations found in dialysate samples.
GLUCOSE SAMPLING UNIT – CALIBRATION: LIMIT OF THE SYSTEM ERROR < 5%
-----------------------------
In 34 out of 39 systems a decrease in the number of required calibration points can be observed by applying the IRT
In total all 15 CMA data sets could be improved using the IRT.
In total 1 data set of PME012 and 18 PME011 could be improved using the IRT whereas data sets of 1 PME 012 and 4 PME011 were worsened.
RISK MANAGEMENT / SAFETY CHECK
-----------------------------
The EU-CLAMP device cannot be operated without additional measures as the patient leakage currents exceed the limits of the IEC 60601-1 Therefore two possible solutions were investigated:
The system can be operated with two isolating transformer in series (not shown here and unhandy)
The system was operated with one single isolating transformer and an additional USB to USB isolator.
The results of the safety checks showed that both solutions are feasible and decrease the patient leakage currents below the IEC 60601-1 limits.
→ To keep the setup simple we chose option 2 using one insulated power supply and one USB to USB isolator.
As the system should also be applicable to diabetic subjects SYS1 and SYS2 were also tested with a glucose AND insulin infusion through the BBRAUN Space Tower
→ During the clinical trial only the setup for diabetic patients must be used!
The potentiostat does not cause any thermal hazard to the patient
Biological contamination must be prevented by preventing a back flow to the body-interface.
IN VITRO CHARACTERISATION OF IGMS
-----------------------------
When the perfusate is heated from room temperature to the temperature of the water bath (36 – 42°C) increasing temperature differences lead to increased out-gassing effects and thereby air bubble artefacts in the sensor signal.
A change of the perfusate syringe can disrupt the sensor signal for several hours. Although the changing procedure could be improved and optimized and signal disruptions were kept small during the later experiments, appropriate syringe volumes should be used during the clinical trials to avoid unnecessary syringe changes.
When the sensor signal decreases due to air bubbles covering the working electrode of the sensor the flow cell can be easily flushed with a 5ml syringe filled with perfusate. Thereby the flushing of the flow cell has to be done from the inlet port of the flow cell (including unplugging of the BI outlet tubing) into a waste bin. One must not flush the perfusate back into the body interface due to the risk of contamination and rupture of the BI membrane.
During overnight periods some air bubbles vanish without any further measures. Still, the sensor signal is disrupted for several hours and therefore the flow cells should be manually flushed as soon as the sensor signal is influenced by air bubbles.
As the sensor signals still show small fluctuations, probably due to micro air bubbles, applying a sliding average correction may help to smooth the signal. Nevertheless, it should be kept in mind that this increases the response time of the system.
Temperature fluctuations do have a great impact on the sensor signal, but the influence can be mathematically compensated when an adequate model is applied. During clinical trials this may not be necessary as the sensor signal will be frequently calibrated to the blood reference level.
A syringe filter can hold back air bubbles that would be introduced into the system through the syringe (due to improper filling) or originate from out-gassing effects. But this filter might hamper the transport of Arixtra to the bodyinterface.
Degassing the perfusate avoids out-gassing effects due to heating and thereby air bubbles artefacts in the sensor signal.
The fluid warmer “Buddy IV” is not applicable for low flow rates as it does not allow the establishment of a stable flow. Therefore it had to be used differently from its intended use. But as the experiments revealed that a syringe filter and/or degassed perfusate is sufficient to avoid air bubble artefacts the fluid warmer “Buddy IV” should not be integrated in the combined system to keep the setup as simple as possible.
The response time of the system is < 3min when the connection tubing between BI and flow cell has a length of 5cm and the syringe pump is operated at a flow rate of 20µl/min.
The delay time can be reduced by 30s in a final setup, when the flow cell is placed directly behind the bodyinterfaces (no tubing of 5cm length and ID of 0.28mm) outlet and the system is operated at 10µl/min.
As the sensor signals still show small fluctuations, probably due to micro air bubbles, applying a low pass- filter may help to smooth the signal. Nevertheless, it should be kept in mind that this increases the response time of the system. Therefore more sophisticated filters could be used.
SETUP OF IN VITRO INVESTIGATION – MICROFLUIDICS
-----------------------------
Using the BBRAUN Perfusor fm in push mode together with flow cell, sensor and dialysate sampling unit showed a constant flow rate over 24 hours at 10 and 20µl/min. Thus it can be concluded that from a micro fluidic point of few the system was performing very well and the used commercially available pumps can be recommended for this application.
Larger flow rates cause lower glucose- and ion recoveries. It has been shown that low recoveries results in poor correlation for this particular set-up. The mean glucose recovery should be higher than 5% in order to achieve good results applying the IRT. This finding was also confirmed during the studies with the sensors.
From a sampling point of view it is mandatory to have low flows in order to gain high recovery for the used application. However this is in contrast to the linear range of the glucose sensor and the increased delay.
To avoid generation of air bubbles the perfusate has to be degassed before it is used to perfuse the probes. This has been done by applying under-pressure to the perfusate.
SETUP OF IN VITRO INVESTIGATION – ANTI-COAGULATION
-----------------------------
It can be concluded that anti-coagulation is best if Fondaparinux (ARIXTRA®®) is injected subcutaneously (2.5mg) before any trial related activities are started and if it is additionally added to the perfusate (additional 2.5mg).
The data indicate that the initial drop of the recovery can be improved by an anti-coagulation treatment, which makes it very likely that the drop of the recovery is caused by layers of components which can be prevented by the drug.
SETUP OF IN VITRO INVESTIGATION – PROBES
-----------------------------
No µEye had to be replaced during the clinical investigations whereas one reference blood sampling catheter (subject 026; t: 5:30) had to be replaced.
The data indicate that a drop of the recovery can be associated with a formation of a layer around the membrane.
In 3 µEye catheters no thrombus formation was found whereas in 2 (subject 021 – not systemically anti-coagulated and subject 023) thrombus formations protein/ deposition were found (for more details see Appendix).
48.3% of the µEye catheters showed a good correlation coefficient of more than 0.8.
SETUP OF IN VITRO INVESTIGATION – SENSOR AND FLOW CELL
-----------------------------
No flow cell or sensor had to be changed during the clinical investigations.
6 delivered flow cells showed dysfunctions: 2 times no and 1 time a loose contact, 1 blocked microfluidic channel, 1 flow cell broke during closure and 1 loose tubes to connect microfluidics, respectively
Mean value of the sensor current recorded throughout run in period of the sensors used for subject 021-026 + 5 backup sensor was found to be 4.39 +/-1.62nA after 10 hours and 2.23+/-0.49nA after 24 hours, respectively
All sensors could be operated without the occurrence of any air bubbles within the flow channel. Sometimes the narrow gap around the flow channel was withdrawing micro air bubbles from the dialysate which stayed there and may have caused a noisy sensor signal.
These micro air bubbles within the flow cell could not be removed by flushing the flow cell with saline solution.
Micro-bubble traps or a design of the flow cell which is not prone to air bubbles could improve the system
The influence of changing ion concentration above 15-20 % ions is minimal. Therefore it is recommended to add 20% ions to the perfusate.
Sensor currents have to be filtered to achieve appropriate data for point to point comparison.
The sensor filter improved the performance of the glucose monitoring system. The used filter improved the overall performance of the glucose monitoring unit and 39.1% less calibrations points were required to calibrate the senor signal to blood and meet the criteria that the system error was below +/- 10%. Further more sophisticated filter approaches could be used to further improve the performance.
For the proposed set-up the sensor and body-interface do not fit to each other due to the limited linear range of < 20mg/dl of the glucose sensor. If the body-interface would be operated at low flow rates to achieve higher recoveries (e.g. 50% or more) the sensor would need at least a linear range of 100mg/dl.
Ionic Reference Technique (IRT): Changes in the recovery (flow rate variations, movement, changing diffusion behaviour, etc.) can be (partially) compensated with the IRT.
The higher the recovery the higher the correlation between blood and calculated glucose for the particular set-up.
In order to avoid influence of ions on the glucose sensor 20 % ions should be added to the perfusate. This has to be taken into consideration when calculating the recovery
SETUP OF IN VITRO INVESTIGATION – GLUCOSE MONITORING SYSTEM
-----------------------------
To fulfil the EN/ ISO 15197 guideline a half-hourly calibration is mandatory to find 95% of all calibrated data points within the predefined region.
At flow rates of 5µl/min or less the correlation between reference blood samples and dialysate might be improved, what is in contradiction to the current linear range of <20mg/dl of the BVT glucose sensor. To apply a diffusion limiting coating onto the sensor’s active area might be a solution to enable the BVT sensor to measure even in a higher glucose concentration range.
SETUP OF IN SILICO INVESTIGATION - CONTROL ALGORITHM
-----------------------------
For the algorithm development, a virtual testing environment was created where a population of synthetic subjects with type 1 diabetes was tested in a virtual computer space. This type of in silico environment was developed by UCAM and it has been used extensively over the past decade to evaluate glucose control algorithms. The UCAM simulation environment was adapted for testing glucose clamp algorithms and was used to develop, tune and evaluate (in comparison to an existing control algorithm) the UCAM clamp control algorithm for use in the EU-Clamp closed loop system.
The in silico studies using the UCAM metabolic simulator have demonstrated safety and efficacy of the closed loop control algorithm gMPCeuClamp Version 1.0.3 and indicated that the UCAM algorithm results in a more accurate and less oscillatory dextrose infusion rate compared to the Biostator algorithm whilst retaining the ability to achieve the desired glucose level
-----------------------------
SUMMARY
-----------------------------
During the 2 years of the EU-Clamp project investigations and developments have been done for a new generation automated clamp device. The principle idea was to develop a system, which continuously monitors blood glucose without blood loss based on intravenous micro dialysis. It was decided by the consortium to do a stepwise approach starting with investigations related to micro dialysis, continue with the glucose sensor and finally integrate the subcomponents into the EU-Clamp’s monitoring system. The development was performed in iterative circles starting with in vitro investigations and continued with in vivo experiments.
Intravenous micro dialysis is not new but only recently CE-marked probes were available. These probes (µEye from partner Probe Scientific and IView from CMA) were investigated in vitro as well as in vivo and compared with each other. The first step in this process was to investigate existing reference devices/methods and adapt them to the specific needs of the project. Particular focus was put on the glucose measurement of small volumes and low concentrations. Another quantity to be measured was the electrical conductivity, which was used to calculate the partial equilibration (recovery) of the micro dialysis process using the ionic reference technique. An important part of the work was the investigation of the microfluidics in particular the investigation of pumps, which deliver perfusate into the micro dialysis probes. It is known that the flow-rate of the perfusate modulates directly the recovery hence it is important to have a constant and known flow. Two different pump concepts were investigated: the peristaltic and the syringe approach. With the former one the push-pull concept could be implemented with the syringe pump only the push mode was possible. Due to the fact that all approaches were tested in vivo only pumps with a CE mark were investigated (JR-pump and BBraun perfusor).
In the first set of in vivo investigations 39 micro dialysis probes applying the Ionic Reference Technique (IRT) were investigated in 20 subjects. Different parameters such as pumps, probes, flow-rates and anticoagulation strategies were investigated. The major findings were: both pumps can be used for the investigation. It was decided to prefer the push pump (BBraun) because of it more robust and stable operation. As expected with the flow rate the recovery rate could be modulated (high flow rate results in low recovery and the other way round). Since recoveries <5% showed a poor correlation between dialysate- and blood glucose flow rates of 10 µl/min or less turned out to be optimal for the particular application. The Ionic Reference Technique proved to be successful in improving the signal in 85% (33 out of 39 probes) of the investigations thus compensating changes of the recovery, which occurred during the in vivo investigations. Fondaparinux (Arixtra®) proved to be powerful in preventing thrombus formation around the microdialysis probe irrespectively of the used probes because of its small molecular weight.
Based on these findings the glucose sensors were investigated and integrated into a glucose monitoring system taking into account the risk identified by the risks assessment. During the in vitro tests of the integrated system it turned out that air bubble formation was a problem. Different strategies were applied and investigated and finally degassing of the perfusate turned out to be a viable solution to the problem. The integrated system was tested in 5 subjects with type 1 diabetes. Due to the fact that the sensors’ working range was limited to 20 mg/dl adaption of the flow rate during the investigation had to be made to limit the glucose concentration in the dialysate and hence the recovery of the micro dialysis process. There was a good correlation between the sensor signal and blood glucose. There is space for improvement for the quality of the electrical signal of the sensor, which could be achieved by further integrating/miniaturising the probe-sensor-electronic interface. Further improvements could be achieved by optimising the microfluidics (volume between probe and sensor) to decrease the delay, increasing the linear range of the sensor to be able to run the micro dialysis probe at higher recovery rates and to design the system that it is not prone to air bubbles (bubble trap, design of the flow cell that bubbles pass through).
Apart from the in silico data on the control algorithm performance, clinical data with the new algorithm will be needed before the algorithm can be integrated into a prototype clamp device which can be tested in further clinical trials
Overall, a glucose monitoring system based on intravenous micro dialysis and glucose sensors was developed and tested in vitro as well as in vivo. Although an integrated system needs further development it was possible to show that the approach is feasible and attractive for continuous glucose monitoring without blood loss during clamp investigations.
Potential Impact:
The prevention and treatment of diabetes is a high priority for The Health and Consumer Protection Directorate General (implemented through Diabetes Policy Frameworks in the EU Member States). Further developments of the EU-CLAMP project may complement these activities by:
• Strengthening diabetes research in Europe
• Improving inpatient glycaemic care while reducing associated healthcare costs
• Creating new jobs at each of the participating SMEs
After completion of the development of an improved glucose clamp device which will be used to evaluate the performance of new anti-diabetic drugs within the pharmaceutical industry, the EU-CLAMP device with only minor modifications could deliver a range of additional benefits in areas that current glucose clamping technologies are also applied. These additional applications include:
Glucose monitoring (for hospital care): High blood glucose concentrations (hyperglycaemia) in hospitalised patients are a common, serious and costly healthcare problem with profound medical consequences. Nearly 40% of patients admitted to hospital have hyperglycaemia, mostly patients with a known history of diabetes. Moreover, hyperglycaemia may develop acutely in hospitalised patients with and without diabetes during acute medical illness and/or post-surgical intervention, which greatly increases the risk of e.g. post-surgical infections. The market for glucose monitoring at the hospital setting is therefore substantial. Tight blood glucose control has proven to have a positive impact on mortality, co-morbidity in different patient populations at the intensive care unit and can substantially reduce the length of stay in hospital. Additionally, several investigations show findings consistent with the importance of inpatient glycaemic control in hospitalized medical and surgical patients suffering from diabetes or stress related hyperglycaemia. Either at the ICU or at the general ward, nurses and physicians would benefit from continuous glucose monitoring as an aid to achieve tight blood glucose control.
Fully automated closed loop system (for hospital care): As mentioned above, improvement in mortality and morbidity in patients at the ICU or at the general ward can be achieved if blood glucose levels are well controlled. However, tight blood glucose control also bears a substantial risk of hypoglycaemia, which again has been associated with increased mortality rates. A functioning closed loop device would be able to establish normal glucose levels in these patients without hypoglycaemia, without blood loss and, ideally, without attendance of nurses and physicians. Such a device would be highly cost-effective and therefore of significant interest to intensive care units and general hospitals throughout Europe. It has been estimated that closed-loop glycaemic control reduces healthcare costs by approximately €350 per patient per day. In addition, a device based on the microdialysis technique could also measure (and regulate) potassium levels in a critically ill patient, which would also be of high interest in particular for cardiology units. Thus, it seems conceivable that a closed-loop system could also be commercialised.
The research initiated during the EU-CLAMP Project may not only result in the development of more efficacious treatments that will improve the lives and well-being of EU citizens living with diabetes, but it may simultaneously help to contain the dramatic rise in the associated healthcare costs of diabetes.
The EU member states currently spend more than €60billion per annum on healthcare associated with diabetes and its complications. 75% of these costs are related to micro- and macrovascular complications, which can be reduced by better glycaemic control. If new compounds can reduce the incidence and thereby the costs for diabetic complications by only 5%, this would save the EU member states €3 billion per annum.
Apart from the obvious benefits of tight glycaemic control in the hospital to patients, the cost-saving potential of a closed loop system for glucose control in hospitals as spin-off technology is huge. Approximately 8.7 million patients with diabetes are hospitalised in the EU every year and another 29 million patients without a history of diabetes are admitted to hospital with hyperglycaemia. The length of stay of these patient groups is 4-5 days longer than patients without hyperglycaemia. Assuming a reduction in the length of stay of patients with hyperglycaemia by just one day and an estimated saving of €350 per patient per day, closed loop glucose control could potentially reduce inpatient healthcare costs in the EU by approximately €13.2 billion per annum. Moreover, indirect costs of diabetes e.g. due to disability for work and premature mortality, can even exceed the medical costs.
List of Websites:
Website:
www.euclamp.eu
Contact information:
Eric Zijlstra
Administrative Manager
Tel: +49-2131-4018 219
Fax: +49-2131-4018 500
e-mail: eric.zijlstra@profil.com
Europe faces a diabetes epidemic. More than 55 million people in Europe are currently diagnosed with diabetes and with an estimated 20% increase by 2030, the disease is certain to stay one of the most challenging health problems this century. Especially as diabetes no longer is a disease exclusively for adults, but affects children, young people and adults of all ages.
Despite the high prevalence of diabetes, the choice of anti-diabetic drugs is still limited and two thirds of patients with diabetes do not achieve the recommended glycaemic target levels. For each new anti-diabetic drug, it is essential to investigate the metabolic effect over time. The glucose clamp technique is regarded as the gold standard to evaluate the effectiveness of new anti-diabetic drugs. There are however only a few centres with limited research capacities that have experience in using the clamp technique, because automated clamp devices are no longer commercially available and the existing techniques are confounded by a number of limitations. Increased clamp capacities are urgently needed for the development of new, more efficacious anti-diabetic drugs.
Through the integration of newly emerging technologies proposed in part by the 4 participating SME organisations and with the outsourced research capacity of 3 of Europe’s leading RTD performers, the EU-CLAMP project aims to develop a new generation automated clamp device that will overcome the limitations of the existing devices by incorporating an intravenous microdialysis technique for reliable continuous glucose monitoring without blood loss. The EU-CLAMP project will facilitate clamp testing in a more efficient and cost effective manner. In addition to the significant contribution that could be made to development of improved treatment options for diabetes, the project will provide a platform from which the competitiveness of the participating SMEs can be improved offering alignment to the needs of their long term business strategies.
Project Context and Objectives:
The principle of the glucose clamp is based on the idea that a fall in blood glucose concentrations (e.g. induced by pharmaceutical agents) is prevented by a variable infusion of glucose. The amount of glucose needed to stabilise blood glucose concentrations (expressed as glucose infusion rates) is a direct measure of the metabolic effect of the compound. Glucose clamps can be performed either manually or automatically. During manual clamps, the investigator frequently samples the blood for glucose content and manually adjusts the rate of glucose infusion on the basis of the blood glucose measurements. As the automated clamp technique has obvious advantages over the manual technique (i.e. no user-bias, less labour intensive and higher clamp quality due to continuous glucose monitoring), only the automated clamp technique will be considered in the remainder of this report.
The glucose clamp technique is regarded as the gold standard for the assessment of the metabolic profiles of anti-diabetic compounds and insulin sensitivity. Currently, three devices for automated clamps have made it onto the market all with their own limitations:
1. The Biostator (Glucose controlled insulin infusion system, Miles Laboratories Inc., Elkhart, IN, USA). Developed in the 1970’s, the device was only sold for a short period of time, and consequently the availability of components is limited. Further limitations include glucose monitoring capabilities. The device uses an electrochemical membrane for continuous blood glucose monitoring which often shows drifts, even with frequent calibrations. Therefore, it is mandatory to check blood glucose concentrations with an external blood glucose analyser at least every 30 min. The blood loss for the continuous measurement and the external checks is about 2.5 ml/h. Because of a high incidence of technical errors, experienced staff are needed to perform the clamp experiments.
2. The Glucostator is a recently developed glucose clamp device, but does not overcome the limitations of the Biostator. The Glucostator uses similar principles for continuous blood glucose monitoring and the same algorithm for the calculation of glucose infusion rates as the Biostator. Therefore, the limitations with regard to blood loss, costs, labour intensity and quality of glucose monitoring (and consecutively variability of the glucose clamp) remain the same.
3. Nikkiso STG-22 (Nikkiso Co., Tokyo, Japan), a Japanese device that is not licensed for use in Europe or the USA.
The overarching aim of the EU-CLAMP project: to develop a prototype of a new generation clamp device. The new clamp device will overcome the current limitations of existing clamp techniques and is needed to evaluate the effectiveness of new anti-diabetic drugs in a clinical setting. To progress towards this aim, the following Research and Development objectives were defined:
1. To develop the individual components of the clamp device according to the user requirements. Four components have been identified: body-interface (iv microdialysis catheter), microfluidics, glucose sensor & electronics and control algorithm. The components will be individually tested in a laboratory setting at JR and MUG. The control algorithm will be developed or adjusted from an existing algorithm, and tested by UCAM (see also objective 3). On the basis of the outcome of the laboratory tests, the performance of the components may be optimised by the relevant consortium partners.
2. To combine the novel iv microdialysis technique with an online glucose sensor to develop a module for the continuous monitoring of blood glucose. PROBE’s iv microdialysis catheter will be linked to a BVT glucose sensor via the microfluidic technology developed by JR. Performance tests (accuracy, detection limit, linearity, variability) of the monitoring module will be performed by MUG in healthy human volunteers. The accuracy of the glucose measurements will be compared to a reference laboratory method.
3. To combine the control algorithm with commercially available high precision infusion pumps. In vivo measurements of this infusion module of the clamp device might be performed and used to optimise the control algorithm.
4. To construct a prototype new generation clamp device. Integration of the components and modules into a prototype clamp device will be lead by DATAMED.
5. To test the safety and performance of the prototype clamp device during a series of clinical tests under medical supervision at PROFIL. In these tests with healthy human volunteers and patients with diabetes safety parameters will be documented. The performance of the prototype new generation clamp device will be compared to the current ‘gold-standard’ clamp device, the Biostator, with respect to measurement accuracy, clamping precision (i.e. blood glucose control), blood loss and reliability.
The performance characteristics of the prototype device that EU-CLAMP project aims to achieve are:
• No blood loss
• Accurate and precise continuous glucose monitoring without signal drifts with modern laboratory standards
• Continuous glucose monitoring profile without time-delays
• Continuous glucose infusion profile
• Stabilisation of blood glucose concentrations at the target level with variability < 5%
• Automated electronic data capturing system for blood glucose concentrations and glucose/insulin infusion rates meeting regulatory requirements.
• Reliable technical performance using modern technology
• Reduced risk of infection (for nurses and physicians)
• Easy-to-use, less labour-intensive automated device
Project Results:
REFERENCE METHODS
-----------------------------
CMA 600: Maintenance of the CMA 600 did not improve its accuracy and precision. Its coefficient of variation (CV) was in some cases above 5% - up to 5 measurements per sample had to be performed in order to get reliable results; therefore the work with the CMA 600 was stopped and alternatives were investigated.
Super GL2: Our results for 0.9% saline solution as well as 5% Mannitol revealed that with the Super GL2 glucose levels down to 1.25mg/dl with a CV of less than 5% (in most cases approx. 1-2%) can be quantified. If concentrations below 11mg/dl have to be measured a modified procedure was used, namely the 20-fold volume (400µl) was injected. Thus the results have to be corrected for volume difference in the Glucocapil container (1400µl/1020µl) and have to be divided by the factor 20 (for 20-fold amount of requested glucose).
Conductivity measurement: The contactless conductivity measurement device TraceDec from I.S.T. was found to have a CVs smaller than 2% (Min: 0.61 Max: 1.89%).
Different Heparin concentrations of Heparin Immuno and Arixtra show very little (<4.8%) influence on the measured glucose concentrations, as no trend of the glucose concentration with increasing Heparin concentration was observed.
Different Heparin concentrations of Heparin Immuno and Arixtra show an influence on the measured conductivity signal. This fact has to be considered if the ionic reference technique is applied as values derived should be corrected for these values.
CHARACTERISATION OF PUMPS
-----------------------------
Both pumps (JR-pump and the BBRAUN Perfusor) show a pulsatile flow
When a microdialysis probe is connected to the pump the JR-pump shows a pulsatile flow due its push/pull operation mode in contrast to the BBRAUN Perfusor piston pump that shows a non-pulsatile flow.
Mean flow rates with or without body interface are similar for both pumps, thus no perfusate is lost over the membrane.
Pressure within the tubing system needs to equilibrate for more than 15min if the flow changes from 250 to 0µ/min and more than 5 min if the flow changes from 0 to 250µ/min, respectively.
The tubing system of the JR-pump showed wearing signs after a 24h experiment when being operated at a nominal flow of 10µl/min.
The Emmerich battery shows a life-time of around 16.5h whereas the EVE battery shows a life time of around 19.45h respectively. Thus at a maximal nominal flow rate of 10µl/min batteries have to be changed within one 24h experiment.
At a nominal flow rate of 10µl/min the pump tubing are extremely stressed but can be used at least for 24h. Occasionally we found lose bonding point of the tubing system.
BODY-INTERFACE
-----------------------------
Recovery of glucose and ions correlate very well which is a prerequisite for the ionic reference technique
For most cases the major decrease of recovery was between 1-50 µl/min (except for the CMA 64 in blood – there was a significant lower recovery for flow rates > 10µl/min.
Glucose recovery does not depend on the composition of the perfusate.
The major decrease of recovery was between 1-50 µl/min
A mathematical function can be found for the relationship of glucose and ion recovery (polynomial function of 3rd order).
A dis-configuration of the µEye PME 011 was found if the probe became wet
DEPENDENCY OF THE GLUCOSE SENSOR CURRENT ON IONIC STRENGTH
-----------------------------
The calibration curves of the glucose sensor are linear up to 20mg/dl.
The different ion concentrations influence the sensor signal. On average (mean over 8 sensors), and as already prior experiments have shown, a lower ion concentration leads to a higher signal current.
The influence of the ions gets less for higher ion concentrations: there is nearly no difference between the calibration curve between 15 and 20 % ions
The perfusate must contain a certain amount of ions (e.g. 20% NaCl: 8 parts 5% Mannitol and 2 parts 0.9% NaCl solution).
GLUCOSE SAMPLING UNIT – PRESSURE
-----------------------------
With increasing flow-rates the pressure within the system is increasing.
Flow through cell and sensor show nearly no back pressure.
The CMA 64 shows more back pressure than the µEye.
Pressure of up to 540mmH2O might significantly influence the recovery of ions and glucose.
At flow rates of 10µl/min the system can be operated with the BBRAUN Space or Perfusor fm pump.
GLUCOSE SAMPLING UNIT – FLOW
-----------------------------
A decreasing flow could be observed for peristaltic JR pump at a flow rate of 10µl/min, caused by the running-in characteristic of the tubing set.
Using a push-pull setup did not improve the flow stability and its variations if no sensor + sensor cell is attached to the bodyinterface. On the other hand the push setup of the BBRAUN Perfusor Space showed a constant flow and less variation.
BBRAUN Perfusor Space is the more convincing setup as the pumps are not attached to the subject`s arms and thus allows unrestricted movements of the subjects. Furthermore the system is well known to medical health personal and easier to handle.
GLUCOSE SAMPLING UNIT – SETUP
-----------------------------
A mean recovery lower than 5% is associated with a poor correlation (< 0.8) and thus results in a bad performance of the system.
37.5% of the µEye catheters showed a good correlation coefficient or more than 0.8.
93.3% of CMA-catheters showed a good correlation coefficient of more than 0.8.
All probes with systemic anticoagulation (Arixtra®) showed a good correlation of more than 0.8
The data indicate that adding Arixtra® only to the perfusate improves the recovery and thus might increase the correlation.
CMA body-interfaces gained generally higher recovery rates than µEyes.
The µEyes are more prone for thrombus formation around the membrane than the CMAs.
Fondaparinux (ARIXTRA®®) seems to be advantageous compared to Heparin due to its smaller molecular weight of 1.7 kDa instead of 4-6 kDa.
The data indicate that a drop of the recovery can be associated with a settlement of bodily components (e.g. plate lets, proteins, etc.) around the body-interface.
In order to prevent a thrombus formation an anticoagulation drug should be used either systemically, locally (addition to perfusate) or both.
GLUCOSE SAMPLING UNIT – ULTRASOUND INVESTIGATIONS
-----------------------------
In total 25 ultrasound investigations were performed: 9 CMA and 16 µEye, respectively.
In 6 CMA catheters no thrombus formation was found whereas in 3 thrombus formations were found (for more details see Appendix).
In 10 µEye catheters no thrombus formation was found whereas in 6 thrombus formations were found (for more details see Appendix).
GLUCOSE SAMPLING UNIT – IONIC REFERENCE TECHNIQUE
-----------------------------
A significant decrease of the recovery can be (partially) compensated with IRT.
Changes in the recovery (flow rate variations, movement, changing diffusion barrier, etc.) can be (partially) compensated with the IRT.
Overall performance (correlation) can be significantly increased with the IRT (33 out of 39 systems).
Recovery should be > 5% in order to achieve proper results to apply the IRT.
Higher recoveries result in better correlation between glucose concentrations found in blood and glucose values calculated from the glucose concentrations found in dialysate samples.
GLUCOSE SAMPLING UNIT – CALIBRATION: LIMIT OF THE SYSTEM ERROR < 5%
-----------------------------
In 34 out of 39 systems a decrease in the number of required calibration points can be observed by applying the IRT
In total all 15 CMA data sets could be improved using the IRT.
In total 1 data set of PME012 and 18 PME011 could be improved using the IRT whereas data sets of 1 PME 012 and 4 PME011 were worsened.
RISK MANAGEMENT / SAFETY CHECK
-----------------------------
The EU-CLAMP device cannot be operated without additional measures as the patient leakage currents exceed the limits of the IEC 60601-1 Therefore two possible solutions were investigated:
The system can be operated with two isolating transformer in series (not shown here and unhandy)
The system was operated with one single isolating transformer and an additional USB to USB isolator.
The results of the safety checks showed that both solutions are feasible and decrease the patient leakage currents below the IEC 60601-1 limits.
→ To keep the setup simple we chose option 2 using one insulated power supply and one USB to USB isolator.
As the system should also be applicable to diabetic subjects SYS1 and SYS2 were also tested with a glucose AND insulin infusion through the BBRAUN Space Tower
→ During the clinical trial only the setup for diabetic patients must be used!
The potentiostat does not cause any thermal hazard to the patient
Biological contamination must be prevented by preventing a back flow to the body-interface.
IN VITRO CHARACTERISATION OF IGMS
-----------------------------
When the perfusate is heated from room temperature to the temperature of the water bath (36 – 42°C) increasing temperature differences lead to increased out-gassing effects and thereby air bubble artefacts in the sensor signal.
A change of the perfusate syringe can disrupt the sensor signal for several hours. Although the changing procedure could be improved and optimized and signal disruptions were kept small during the later experiments, appropriate syringe volumes should be used during the clinical trials to avoid unnecessary syringe changes.
When the sensor signal decreases due to air bubbles covering the working electrode of the sensor the flow cell can be easily flushed with a 5ml syringe filled with perfusate. Thereby the flushing of the flow cell has to be done from the inlet port of the flow cell (including unplugging of the BI outlet tubing) into a waste bin. One must not flush the perfusate back into the body interface due to the risk of contamination and rupture of the BI membrane.
During overnight periods some air bubbles vanish without any further measures. Still, the sensor signal is disrupted for several hours and therefore the flow cells should be manually flushed as soon as the sensor signal is influenced by air bubbles.
As the sensor signals still show small fluctuations, probably due to micro air bubbles, applying a sliding average correction may help to smooth the signal. Nevertheless, it should be kept in mind that this increases the response time of the system.
Temperature fluctuations do have a great impact on the sensor signal, but the influence can be mathematically compensated when an adequate model is applied. During clinical trials this may not be necessary as the sensor signal will be frequently calibrated to the blood reference level.
A syringe filter can hold back air bubbles that would be introduced into the system through the syringe (due to improper filling) or originate from out-gassing effects. But this filter might hamper the transport of Arixtra to the bodyinterface.
Degassing the perfusate avoids out-gassing effects due to heating and thereby air bubbles artefacts in the sensor signal.
The fluid warmer “Buddy IV” is not applicable for low flow rates as it does not allow the establishment of a stable flow. Therefore it had to be used differently from its intended use. But as the experiments revealed that a syringe filter and/or degassed perfusate is sufficient to avoid air bubble artefacts the fluid warmer “Buddy IV” should not be integrated in the combined system to keep the setup as simple as possible.
The response time of the system is < 3min when the connection tubing between BI and flow cell has a length of 5cm and the syringe pump is operated at a flow rate of 20µl/min.
The delay time can be reduced by 30s in a final setup, when the flow cell is placed directly behind the bodyinterfaces (no tubing of 5cm length and ID of 0.28mm) outlet and the system is operated at 10µl/min.
As the sensor signals still show small fluctuations, probably due to micro air bubbles, applying a low pass- filter may help to smooth the signal. Nevertheless, it should be kept in mind that this increases the response time of the system. Therefore more sophisticated filters could be used.
SETUP OF IN VITRO INVESTIGATION – MICROFLUIDICS
-----------------------------
Using the BBRAUN Perfusor fm in push mode together with flow cell, sensor and dialysate sampling unit showed a constant flow rate over 24 hours at 10 and 20µl/min. Thus it can be concluded that from a micro fluidic point of few the system was performing very well and the used commercially available pumps can be recommended for this application.
Larger flow rates cause lower glucose- and ion recoveries. It has been shown that low recoveries results in poor correlation for this particular set-up. The mean glucose recovery should be higher than 5% in order to achieve good results applying the IRT. This finding was also confirmed during the studies with the sensors.
From a sampling point of view it is mandatory to have low flows in order to gain high recovery for the used application. However this is in contrast to the linear range of the glucose sensor and the increased delay.
To avoid generation of air bubbles the perfusate has to be degassed before it is used to perfuse the probes. This has been done by applying under-pressure to the perfusate.
SETUP OF IN VITRO INVESTIGATION – ANTI-COAGULATION
-----------------------------
It can be concluded that anti-coagulation is best if Fondaparinux (ARIXTRA®®) is injected subcutaneously (2.5mg) before any trial related activities are started and if it is additionally added to the perfusate (additional 2.5mg).
The data indicate that the initial drop of the recovery can be improved by an anti-coagulation treatment, which makes it very likely that the drop of the recovery is caused by layers of components which can be prevented by the drug.
SETUP OF IN VITRO INVESTIGATION – PROBES
-----------------------------
No µEye had to be replaced during the clinical investigations whereas one reference blood sampling catheter (subject 026; t: 5:30) had to be replaced.
The data indicate that a drop of the recovery can be associated with a formation of a layer around the membrane.
In 3 µEye catheters no thrombus formation was found whereas in 2 (subject 021 – not systemically anti-coagulated and subject 023) thrombus formations protein/ deposition were found (for more details see Appendix).
48.3% of the µEye catheters showed a good correlation coefficient of more than 0.8.
SETUP OF IN VITRO INVESTIGATION – SENSOR AND FLOW CELL
-----------------------------
No flow cell or sensor had to be changed during the clinical investigations.
6 delivered flow cells showed dysfunctions: 2 times no and 1 time a loose contact, 1 blocked microfluidic channel, 1 flow cell broke during closure and 1 loose tubes to connect microfluidics, respectively
Mean value of the sensor current recorded throughout run in period of the sensors used for subject 021-026 + 5 backup sensor was found to be 4.39 +/-1.62nA after 10 hours and 2.23+/-0.49nA after 24 hours, respectively
All sensors could be operated without the occurrence of any air bubbles within the flow channel. Sometimes the narrow gap around the flow channel was withdrawing micro air bubbles from the dialysate which stayed there and may have caused a noisy sensor signal.
These micro air bubbles within the flow cell could not be removed by flushing the flow cell with saline solution.
Micro-bubble traps or a design of the flow cell which is not prone to air bubbles could improve the system
The influence of changing ion concentration above 15-20 % ions is minimal. Therefore it is recommended to add 20% ions to the perfusate.
Sensor currents have to be filtered to achieve appropriate data for point to point comparison.
The sensor filter improved the performance of the glucose monitoring system. The used filter improved the overall performance of the glucose monitoring unit and 39.1% less calibrations points were required to calibrate the senor signal to blood and meet the criteria that the system error was below +/- 10%. Further more sophisticated filter approaches could be used to further improve the performance.
For the proposed set-up the sensor and body-interface do not fit to each other due to the limited linear range of < 20mg/dl of the glucose sensor. If the body-interface would be operated at low flow rates to achieve higher recoveries (e.g. 50% or more) the sensor would need at least a linear range of 100mg/dl.
Ionic Reference Technique (IRT): Changes in the recovery (flow rate variations, movement, changing diffusion behaviour, etc.) can be (partially) compensated with the IRT.
The higher the recovery the higher the correlation between blood and calculated glucose for the particular set-up.
In order to avoid influence of ions on the glucose sensor 20 % ions should be added to the perfusate. This has to be taken into consideration when calculating the recovery
SETUP OF IN VITRO INVESTIGATION – GLUCOSE MONITORING SYSTEM
-----------------------------
To fulfil the EN/ ISO 15197 guideline a half-hourly calibration is mandatory to find 95% of all calibrated data points within the predefined region.
At flow rates of 5µl/min or less the correlation between reference blood samples and dialysate might be improved, what is in contradiction to the current linear range of <20mg/dl of the BVT glucose sensor. To apply a diffusion limiting coating onto the sensor’s active area might be a solution to enable the BVT sensor to measure even in a higher glucose concentration range.
SETUP OF IN SILICO INVESTIGATION - CONTROL ALGORITHM
-----------------------------
For the algorithm development, a virtual testing environment was created where a population of synthetic subjects with type 1 diabetes was tested in a virtual computer space. This type of in silico environment was developed by UCAM and it has been used extensively over the past decade to evaluate glucose control algorithms. The UCAM simulation environment was adapted for testing glucose clamp algorithms and was used to develop, tune and evaluate (in comparison to an existing control algorithm) the UCAM clamp control algorithm for use in the EU-Clamp closed loop system.
The in silico studies using the UCAM metabolic simulator have demonstrated safety and efficacy of the closed loop control algorithm gMPCeuClamp Version 1.0.3 and indicated that the UCAM algorithm results in a more accurate and less oscillatory dextrose infusion rate compared to the Biostator algorithm whilst retaining the ability to achieve the desired glucose level
-----------------------------
SUMMARY
-----------------------------
During the 2 years of the EU-Clamp project investigations and developments have been done for a new generation automated clamp device. The principle idea was to develop a system, which continuously monitors blood glucose without blood loss based on intravenous micro dialysis. It was decided by the consortium to do a stepwise approach starting with investigations related to micro dialysis, continue with the glucose sensor and finally integrate the subcomponents into the EU-Clamp’s monitoring system. The development was performed in iterative circles starting with in vitro investigations and continued with in vivo experiments.
Intravenous micro dialysis is not new but only recently CE-marked probes were available. These probes (µEye from partner Probe Scientific and IView from CMA) were investigated in vitro as well as in vivo and compared with each other. The first step in this process was to investigate existing reference devices/methods and adapt them to the specific needs of the project. Particular focus was put on the glucose measurement of small volumes and low concentrations. Another quantity to be measured was the electrical conductivity, which was used to calculate the partial equilibration (recovery) of the micro dialysis process using the ionic reference technique. An important part of the work was the investigation of the microfluidics in particular the investigation of pumps, which deliver perfusate into the micro dialysis probes. It is known that the flow-rate of the perfusate modulates directly the recovery hence it is important to have a constant and known flow. Two different pump concepts were investigated: the peristaltic and the syringe approach. With the former one the push-pull concept could be implemented with the syringe pump only the push mode was possible. Due to the fact that all approaches were tested in vivo only pumps with a CE mark were investigated (JR-pump and BBraun perfusor).
In the first set of in vivo investigations 39 micro dialysis probes applying the Ionic Reference Technique (IRT) were investigated in 20 subjects. Different parameters such as pumps, probes, flow-rates and anticoagulation strategies were investigated. The major findings were: both pumps can be used for the investigation. It was decided to prefer the push pump (BBraun) because of it more robust and stable operation. As expected with the flow rate the recovery rate could be modulated (high flow rate results in low recovery and the other way round). Since recoveries <5% showed a poor correlation between dialysate- and blood glucose flow rates of 10 µl/min or less turned out to be optimal for the particular application. The Ionic Reference Technique proved to be successful in improving the signal in 85% (33 out of 39 probes) of the investigations thus compensating changes of the recovery, which occurred during the in vivo investigations. Fondaparinux (Arixtra®) proved to be powerful in preventing thrombus formation around the microdialysis probe irrespectively of the used probes because of its small molecular weight.
Based on these findings the glucose sensors were investigated and integrated into a glucose monitoring system taking into account the risk identified by the risks assessment. During the in vitro tests of the integrated system it turned out that air bubble formation was a problem. Different strategies were applied and investigated and finally degassing of the perfusate turned out to be a viable solution to the problem. The integrated system was tested in 5 subjects with type 1 diabetes. Due to the fact that the sensors’ working range was limited to 20 mg/dl adaption of the flow rate during the investigation had to be made to limit the glucose concentration in the dialysate and hence the recovery of the micro dialysis process. There was a good correlation between the sensor signal and blood glucose. There is space for improvement for the quality of the electrical signal of the sensor, which could be achieved by further integrating/miniaturising the probe-sensor-electronic interface. Further improvements could be achieved by optimising the microfluidics (volume between probe and sensor) to decrease the delay, increasing the linear range of the sensor to be able to run the micro dialysis probe at higher recovery rates and to design the system that it is not prone to air bubbles (bubble trap, design of the flow cell that bubbles pass through).
Apart from the in silico data on the control algorithm performance, clinical data with the new algorithm will be needed before the algorithm can be integrated into a prototype clamp device which can be tested in further clinical trials
Overall, a glucose monitoring system based on intravenous micro dialysis and glucose sensors was developed and tested in vitro as well as in vivo. Although an integrated system needs further development it was possible to show that the approach is feasible and attractive for continuous glucose monitoring without blood loss during clamp investigations.
Potential Impact:
The prevention and treatment of diabetes is a high priority for The Health and Consumer Protection Directorate General (implemented through Diabetes Policy Frameworks in the EU Member States). Further developments of the EU-CLAMP project may complement these activities by:
• Strengthening diabetes research in Europe
• Improving inpatient glycaemic care while reducing associated healthcare costs
• Creating new jobs at each of the participating SMEs
After completion of the development of an improved glucose clamp device which will be used to evaluate the performance of new anti-diabetic drugs within the pharmaceutical industry, the EU-CLAMP device with only minor modifications could deliver a range of additional benefits in areas that current glucose clamping technologies are also applied. These additional applications include:
Glucose monitoring (for hospital care): High blood glucose concentrations (hyperglycaemia) in hospitalised patients are a common, serious and costly healthcare problem with profound medical consequences. Nearly 40% of patients admitted to hospital have hyperglycaemia, mostly patients with a known history of diabetes. Moreover, hyperglycaemia may develop acutely in hospitalised patients with and without diabetes during acute medical illness and/or post-surgical intervention, which greatly increases the risk of e.g. post-surgical infections. The market for glucose monitoring at the hospital setting is therefore substantial. Tight blood glucose control has proven to have a positive impact on mortality, co-morbidity in different patient populations at the intensive care unit and can substantially reduce the length of stay in hospital. Additionally, several investigations show findings consistent with the importance of inpatient glycaemic control in hospitalized medical and surgical patients suffering from diabetes or stress related hyperglycaemia. Either at the ICU or at the general ward, nurses and physicians would benefit from continuous glucose monitoring as an aid to achieve tight blood glucose control.
Fully automated closed loop system (for hospital care): As mentioned above, improvement in mortality and morbidity in patients at the ICU or at the general ward can be achieved if blood glucose levels are well controlled. However, tight blood glucose control also bears a substantial risk of hypoglycaemia, which again has been associated with increased mortality rates. A functioning closed loop device would be able to establish normal glucose levels in these patients without hypoglycaemia, without blood loss and, ideally, without attendance of nurses and physicians. Such a device would be highly cost-effective and therefore of significant interest to intensive care units and general hospitals throughout Europe. It has been estimated that closed-loop glycaemic control reduces healthcare costs by approximately €350 per patient per day. In addition, a device based on the microdialysis technique could also measure (and regulate) potassium levels in a critically ill patient, which would also be of high interest in particular for cardiology units. Thus, it seems conceivable that a closed-loop system could also be commercialised.
The research initiated during the EU-CLAMP Project may not only result in the development of more efficacious treatments that will improve the lives and well-being of EU citizens living with diabetes, but it may simultaneously help to contain the dramatic rise in the associated healthcare costs of diabetes.
The EU member states currently spend more than €60billion per annum on healthcare associated with diabetes and its complications. 75% of these costs are related to micro- and macrovascular complications, which can be reduced by better glycaemic control. If new compounds can reduce the incidence and thereby the costs for diabetic complications by only 5%, this would save the EU member states €3 billion per annum.
Apart from the obvious benefits of tight glycaemic control in the hospital to patients, the cost-saving potential of a closed loop system for glucose control in hospitals as spin-off technology is huge. Approximately 8.7 million patients with diabetes are hospitalised in the EU every year and another 29 million patients without a history of diabetes are admitted to hospital with hyperglycaemia. The length of stay of these patient groups is 4-5 days longer than patients without hyperglycaemia. Assuming a reduction in the length of stay of patients with hyperglycaemia by just one day and an estimated saving of €350 per patient per day, closed loop glucose control could potentially reduce inpatient healthcare costs in the EU by approximately €13.2 billion per annum. Moreover, indirect costs of diabetes e.g. due to disability for work and premature mortality, can even exceed the medical costs.
List of Websites:
Website:
www.euclamp.eu
Contact information:
Eric Zijlstra
Administrative Manager
Tel: +49-2131-4018 219
Fax: +49-2131-4018 500
e-mail: eric.zijlstra@profil.com