Evaluation of a new analyser for rapid measurement of blood propofol concentration during cardiac surgery


Correspondence to: Dr N. J. Cowley
Email: n.j.cowley@bham.ac.uk


We report laboratory and clinical evaluations of a blood propofol concentration analyser. Laboratory experiments used volunteer blood spiked with known propofol concentrations over the clinically relevant concentrations from 0.5 to 16 μg.ml−1 to assess linearity and the influence of haematocrit and concurrent drug administration. Analyser concentrations demonstrated excellent linearity (R2 = 0.999). Blood spiked with commonly used drugs showed no significant variation compared to unspiked controls. Propofol measurements were largely independent of haemoglobin concentration. A 6% decay in propofol concentration was observed at the highest prepared concentration. Clinical performance of the analyser was assessed using 80 arterial blood samples from 72 patients receiving propofol infusions during cardiac surgery. Samples were processed using the propofol analyser, and high performance liquid chromatography (HPLC) used as a gold-standard comparator. These data demonstrated excellent agreement between the propofol analyser and HPLC with a bias of 0.13 μg.ml−1 and precision of −0.16 to 0.42 μg.ml−1.

Target-controlled infusion (TCI) is a common method of propofol administration. The ability of TCI algorithms to predict blood propofol concentrations accurately is poor [1, 2], with measures of precision demonstrating errors of up to 60% [3]. Significant bias and large errors in precision are also identified in groups falling outside of limits used in the development of these algorithms [4–6].

Conventional methods of measuring propofol concentrations in blood include high performance liquid chromatography (HPLC), gas chromatography and liquid chromatography mass spectroscopy. These are labour intensive and time consuming methods and cannot be used to deliver clinically useful information. Estimation of blood propofol from expired gases has yet to demonstrate consistent and reliable results [7].

We assessed a device designed to measure the concentration of propofol in a blood sample in 4 min by extracting propofol from 0.5 ml of whole blood using solid phase extraction and quantifying the amount of propofol in the extract using colorimetric detection techniques. This technique for propofol extraction was first described in 2006 in a proof of concept study [8]. We conducted both laboratory and clinical evaluations of the device, the latter using samples from patients undergoing propofol-based general anaesthesia for cardiac surgery.


The analyser is a small tabletop device, somewhat smaller than a point-of-care blood gas analyser. It is capable of making a propofol concentration measurement in heparinised whole blood or plasma using automated solid phase extraction and detection. A minimum sample size of 0.5 ml heparinised whole blood is lysed with de-ionised water in the ratio of 1:2 (blood:water) and filtered through a 1-μm filter, followed by automated solid phase drug extraction [8]. Briefly, a sorbent cartridge (Phenomenex, Macclesfield, UK) is conditioned with methanol and equilibrated with water. The sample is flushed onto the sorbent cartridge and eluted with acetonitrile. A colour change is induced over 40 s by the addition of Gibbs reagent to the extract, and this is quantified using visible absorption spectroscopy. Drug concentration measurements are displayed on an accompanying monitor around 4 min following sample injection and stored. System setup takes less than one hour and is undertaken once a day using calibration solutions with three known propofol concentrations, using the gradient and offset of linear fit as the calibration coefficients for subsequent blood analysis. Data collection is undertaken using customised software developed with LabVIEW 2009 (National Instruments, Austin, TX, USA).

Measurement of whole blood propofol concentration was also made using HPLC [9]. Propofol concentrations were measured using a Shimadzu HPLC system (Shimadzu UK Ltd, Milton Keynes, UK) including a LC-10ATvp solvent delivery pump, a SIL-10A automatic sample injector driven via a SCL-10A system controller with a fluorescence detector set at 276 nm (excitation) and 310 nm (emission). A Phenomenex reverse phase column was used. The isocratic mobile phase consisted of acetonitrile-water mixture (75:25, v/v) at a flow rate of 0.6 ml.min−1. The system was calibrated using seven standard solutions of propofol in acetonitrile covering the range of 0.1–10 μg.ml−1. A quantity of 200 μl of the whole blood sample was lysed and mixed with 800 μl acetonitrile containing 1 μl.ml−1 thymol as an internal standard to compensate for variability of the extraction efficacy. Following mixing, these samples were centrifuged for 5 min at 16 110 g. The supernatant was filtered and injected into the HPLC for analysis, and the propofol concentration calculated from the ratio of the propofol and thymol peaks in the HPLC chromatograph [10]. Data collection was facilitated using Clarity Lite (DataApex, Prague, The Czech Republic). The HPLC system underwent evaluation with respect to its linearity, range and repeatability. The assay was highly linear over the clinical range of 0.5–10 μl.ml−1, with an R2 of 0.9999.

This research was approved by the West Midlands Local Research and Ethics Committee. We conducted a laboratory evaluation of the propofol analyser using volunteer blood, anticoagulated with heparin, at a concentration of 10 IU.ml−1. Blood samples of varying concentration were prepared using the technique of serial dilution. For example, in the linearity experiment, 16 μg.ml−1 propofol blood stock was prepared by spiking 16 μl Propoven (1% propofol emulsion; Fresenius KABI, Runcorn, UK) into 10 ml whole blood. The propofol blood stock was then mixed for 15 min on a roller mixer. Subsequently, a serial dilution of propofol blood stock was carried out using unspiked whole blood to prepare concentrations for analysis of 16 μg.ml−1, 8 μg.ml−1, 4 μg.ml−1, 2 μg.ml−1, 1 μg.ml−1 and 0.5 μg.ml−1 propofol in whole blood. Aliquots of the same sample were tested on both the propofol analyser and using the HPLC standard contemporaneously. Repeatability and linearity was determined over the clinically relevant propofol concentration range. Testing for repeatability was performed on four separate occasions using separate samples of blood with the same sample concentration over a two-week period.

The machine underwent cross-interference studies with a range of drugs commonly used in the operating room and intensive care unit. Aliquots of 2.5 ml whole blood, prepared as described above to achieve a theoretical propofol concentration of 8 μg.ml−1, were spiked with each drug prepared to a total volume of 20 μl with water to achieve a sample drug concentration at twice the usual therapeutic maximum. Drugs used for testing were adrenaline at a concentration of 0.8 μg.ml−1, noradrenaline at a concentration of 0.8 μg.ml−1, cefuroxime at a concentration of 600 μg.ml−1, furosemide at a concentration of 60 μg.ml−1, amiodarone at a concentration of 240 μg.ml−1 and glyceryl trinitrate at a concentration of 0.8 μg.ml−1.

As propofol is strongly bound to cells and protein in blood, the effect of the haemoglobin concentration was also assessed. Heparinised porcine blood samples with haematocrit of 0% (plasma) and 40% (raw blood) were prepared, and spiked with equal amounts of propofol to achieve a known theoretical concentration. These samples were mixed to achieve blood of varying haematocrit, and serial dilution was performed with unspiked blood to achieve theoretical propofol concentrations of 8, 4 and 2 μg.ml−1 for analysis. Analysis of specimens was repeated on three separate occasions to ensure repeatability.

The West Midlands Local Research and Ethics Committee waived consent for samples in which discarded blood only was used for propofol analysis. A total of 80 arterial blood samples were collected from patients undergoing major cardiac surgery and receiving propofol by infusion as a routine part of their anaesthesia either as a sole anaesthetic agent or in combination with isoflurane. Residual blood from a blood gas syringe anticoagulated with non-injectable heparin (PICO sampler; Radiometer, Brønshøj, Denmark), usually amounting to 1 ml, was collected following use by the theatre team. As discarded samples were used, the time of sampling was not determined by the researcher, although data about the dose and duration of propofol infused were recorded. Samples were stored on a roller mixer at room temperature until analysis on the same day. Samples for both the new analyser and HPLC were prepared and analysed in the laboratory contemporaneously to avoid any error introduced by differences in duration of sample storage. A 0.5-ml aliquot of the blood sample was tested using the propofol analyser and a 0.2-ml aliquot of the same sample using the conventional laboratory based method of HPLC.

Agreement between the propofol analyser and the HPLC reference method for the clinical samples was assessed using methods described by Bland and Altman for repeated measures. Performance of the novel device was analysed by calculating bias as the mean difference between the new analyser and the HPLC gold standard, and precision as SD from the mean. Limits of agreement were calculated as 1.96 SD from the mean. Percentage values for bias and precision were calculated from analysis of the percentage difference between the two methods of analysis. Data were analysed using SPSS for Windows (SPSS Inc., Rel. 19.0.0. 2010., Chicago, IL, USA)


The propofol analyser underwent linearity testing using healthy volunteer blood spiked with propofol over the clinically relevant propofol concentration range from 0.5 to 16 μg.ml−1 (Fig. 1). Results demonstrate excellent linearity with a R2 value of 0.999. Samples with propofol concentrations ranging 0.25–6 μg.ml−1 were tested on the new analyser on four occasions over a two-week period, demonstrating good repeatability (Fig. 2). Testing of blood spiked with drugs commonly used in operating theatres or critical care showed no significant variation of values when compared with unspiked control samples (Fig. 3). Figure 4 demonstrates that propofol concentration measurement is largely independent of the haemoglobin concentration. A slight decay in propofol concentration of 6% was observed at the highest prepared concentration. Repeatability of these measurements was good, with SDs of 0.02–0.08 μg.ml−1.

Figure 1.

 Assessment of the linearity of the new analyser in healthy volunteer blood spiked with propofol.

Figure 2.

 Propofol concentrations measured by the new analyser in propofol-spiked healthy volunteer blood samples; the measurements were repeated on four occasions over a two-week period to demonstrate repeatability. Concentrations used: 0.25, 0.5, 1, 2, 4, 6 μg.ml−1.

Figure 3.

 Assessment of the cross-interference of the new analyser showing propofol concentration measurements in samples spiked with commonly used drugs at twice their usual therapeutic concentration. Variance from control −0.47 to +0.14 μg.ml−1.

Figure 4.

 Assessment of the effect of haemoglobin concentration using the new analyser showing samples of varying haematocrit spiked with propofol to achieve approximate concentrations of 8 μg.ml−1, 4 μg.ml−1 and 2 μg.ml−1.

A total of 80 clinical samples were obtained over a four-week period from 72 patients undergoing major cardiac surgery while receiving propofol by infusion as part of their anaesthesia. The correlation between blood propofol measurements using the propofol analyser and HPLC gold standard is illustrated in Fig. 5. The Bland–Altman plot (Fig. 6) shows the deviation of the propofol concentration measured using the propofol analyser from HPLC derived measurements. This shows a small positive bias of 0.13 μg.ml−1 in this dataset. The precision of the propofol analyser (SD of mean) was −0.16 to 0.42 μg.ml−1, with limits of agreement (1.96 SD of mean) of −0.44 to 0.70 μg.ml−1. The bias (mean) and precision (SD of mean) of the propofol analyser when calculated as a percentage difference in propofol concentration between the two methods of analysis were 3.3% and ± 5.6%, respectively.

Figure 5.

 Correlation of the results obtained with new analyser with high performance liquid chromatography (HPLC) measurements using samples obtained during cardiac surgery.

Figure 6.

 Bland–Altman plot showing deviation of propofol concentrations measured using the new device from high performance liquid chromatography (HPLC). Mean bias (+0.13 μg.ml−1) illustrated with dotted line, limits of agreement (1.96 SD) shown with dashed lines.


Initial laboratory based work demonstrated excellent linearity and repeatability of results. We detected no significant impact on measurements when varying haematocrit, or during cross-interference studies with commonly used drugs.

We assessed the function of the propofol analyser using injection of heparinised whole blood into the device, which can output results in around 4 min. Our data demonstrate a good correlation between whole blood propofol concentrations measured using the new device and when using the HPLC standard, a labour intensive laboratory based method of propofol extraction and measurement. As the novel analyser was being directly compared with HPLC, the analysis was not performed at the point of care, to ensure that propofol levels were identical in samples analysed using each method. We did not set out in this study to demonstrate the clinical utility of a device capable of being used at the point of care.

Commercially available propofol TCI machines using pharmacokinetic modelling to estimate blood propofol concentration have drawbacks. Estimated propofol concentrations rely on the patient’s having similar characteristics to the patient cohort used during development of the algorithm, which is frequently not the case. A major concern is that estimated drug levels can differ significantly from actual drug levels, with clinician overconfidence in the estimated propofol concentration [1–6]. Intermittent blood propofol measurements of the type we have assessed could assist the anaesthetist by giving accurate drug concentrations and thus help in managing the patient’s depth of anaesthesia.

Competing interests

Dr. P. Laitenberger, Mr. B. Liu, and Dr. T. H. Clutton-Brock have been employed by, received consulting fees from, or own equity in, Sphere Medical Ltd.


This work was jointly funded by a Health Technologies Devices programme, project HTD 267; and by the SME company Sphere Medical Ltd., Cambridge, UK.