Anal. Chem. 2003, 75, 4397-4401
Minimally Invasive Technique Based on Ultraslow Ultrafiltration To Collect and Store Time Profiles of Analytes Bart Savenije,†,‡ Kor Venema,† Marien A. Gerritzen,‡ Elbert Lambooij,‡ and Jakob Korf*,†
Psychiatric University Clinic, AZG (P6.11), P.O. Box 30.001, NL 9700 RB Groningen, The Netherlands, and Institute for Animal Science and Health, ID-Lelystad, P.O. Box 65, NL 8200 AB Lelystad, The Netherlands
A device is described to collect and store continuously time profiles of analytes over periods of 24 h suitable to sample freely moving individuals (humans and animals). The device consists of a hollow fiber ultrafiltration probe, a long capillary and a nonmechanical unit (a disposable medical syringe) driven by vacuum to withdraw fluid. The principle is that at low rates (e100 nL/min), sample fluid is collected through the ultrafiltration probe into the capillary. A time resolution of less than 5 min over a 24-h collection and storage period was achieved for lactate and glucose. To illustrate an in vivo application, devices were fixed under the wing of freely moving broiler chickens, with subcutaneous or intravenous probe placements. The device can be produced as a disposable, and it may become applied for ex vivo and in vitro monitoring. A device allowing frequent collection of sterile samples facilitates long-term monitoring in biomedical research and routine settings. Monitoring of circadian variations of drugs, hormones, or metabolites is often required in ambulant patients or in freely active animals. For example, glucose and lactate monitoring is often mandatory in critically ill patients and in those persons with diabetes and related metabolic diseases to prevent or reduce organ damage.1 Other frequent determinations include those of toxic drugs, such as immune suppressives used in organ transplantation and antineoplastic drugs used in cancer therapy, endogenous or exogenous metabolites, electrolytes, and hormones that inform about therapeutic and toxic responses so that close monitoring may eventually lead to optimal therapeutic regimens.2,3 Currently used methods of frequent sampling of body fluid from the living individual require either inconvenient implantation of needles and catheters, thereby limiting ongoing behavior, or frequent puncturing of, for example, the tip of the finger or the earlobe.4 All of these sampling techniques are labor-intensive, stressful, and †
AZ-Groningen. ID-Lelystad. (1) Pickup, J. Lancet 2000, 355, 426-427. (2) Mueller, M. M. In Tietz Textbook of Clinical Chemistry; Burtis, C. A., Ashwood, E. R., III, Eds.; W.B. Saunders: Philadelphia, 1999, pp 1328-1358. (3) Moyer, Th. In Tietz Textbook of Clinical Chemistry; Burtis, C. A., Ashwood, E. R., III, Eds.; W.B. Saunders: Philadelphia, 1999, pp 862-905. (4) Young, D. S.; Berner, E. W. In Tietz Textbook of Clinical Chemistry; Burtis, C. A., Ashwood, E. R., III, Eds.; W.B. Saunders Philadelphia, 1999, pp 4272. ‡
10.1021/ac030039x CCC: $25.00 Published on Web 07/17/2003
© 2003 American Chemical Society
painful and not without a risk of infection and can therefore often not been used outside laboratory or medical facilities. We describe here a lightweight ultrafiltration collection device (UCD) to sample and store continuously blood filtrates or subcutaneous fluid; it was applied for the monitoring of lactate and glucose. The principle described previously by us5 and others6,7 is that an ultrafiltrate of a body fluid is collected in a long capillary by vacuum. It is essential to apply a stationary homogeneous laminar flow and to avoid turbulence. A slow, pulsefree and highly regular movement of the ultrafiltrate is achieved by vacuum. The rate of fluid collection (less than 100 nL/min) and the dimensions of and “smearing” by the probe and capillary are so small that the longitudinal (in the direction of the flow) diffusion of the analytes is minimal over a 24-h period. We illustrate the in vivo utility of the technique in freely moving chickens after subcutaneous or intravenous placement of ultrafiltration probes. The present advantages over earlier reports on ultrafiltration5,6 or microdialysis7 are the high time resolution, the duration of the collection period, and the biomedical applications. The present study may not only be considered as a challenge to prove the usefulness of our approach, but to illustrate its versatility for monitoring other species, including humans, and for other applications, as well. EXPERIMENTAL SECTION The UCD. The UCD consists of 3 parts: the ultrafiltration probe, the collection tubing, and the vacuum pump (Figure 1). The ultrafiltration probes were 4-cm hollow fiber membranes (acrylonitrilesodium methallyl sulfonate copolymer, 290 µm o.d., 240 µm i.d.; Filtral 16 AN69 HF, Hospital Industry Meyzieu, France). A spiral (stainless steel wire, 30 µm wire diameter, 200240 µm spiral diameter, 7-10 windings/cm; Vogelsang, Hagen, Germany) prevents collapsing and reduces the volume of the probe. A 15-cm restriction tube (fused silica, 90-µm o.d., 20-µm i.d.; Polymicro Technologies Inc., Phoenix, AZ) was inserted into the probe up to 2 mm from the tip. Membrane, spiral, and fused silica tubing were fixated with cyanoacrylic glue (Ruplo BV, Ten Boer, The Netherlands). The restriction tubing, needed for (5) Moscone, D.; Venema, K.; Korf, J. Med. Biol. Eng. Comput. 1996, 34, 290294. (6) Ash, S. R.; Janle-Swain, E. M. U.S. Patent 4,777,953. (7) Rossell, S; Gonzalez, L. E.; Hernandez, L. J. Chromatogr., B 2003, 784, 385-393.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003 4397
Figure 1. A. The ultrafiltration collection device. The device shown consists of the probe (tip), a restriction capillary, a collection coil and the vacuum pump unit; each element is described in detail in the text. B. The device is in enveloped in a plastic bag and placed under the wing of a broiler chicken. C. A schematic drawing of the tip of the ultrafiltration probe. Diameters are given in the figure; length of the probe is 4 cm.
determining the flow rate, is placed between the probe and the collection tubing (fused silica, 170-µm o.d., 100-µm i.d.; Polymicro Technologies Inc., Phoenix, AZ). The length of the collection tubing varies with the desired storage capacity and is 38.2 cm/h sampling time at a flow rate of 50 nL/min. Before use, the entire system is filled with double-distilled water. The pump unit is a disposable vacuum syringe (1.2 mL Monovette, Sarstedt, Nu¨mbrecht, Germany) producing a vacuum of 6250 Pa, when the barrel is fixed in the outward position, and so a constant flow of 50 nL/ min (as calculated with Poisseuille’s law), when connected to the restriction tubing. For in vivo applications, the vacuum syringe and the collection tubing were enveloped in a plastic bag. In Vitro Experiments. The properties of the UCD were assessed first in vitro. The effects of dead volume and diffusion of the probe and the longitudinal diffusion following storage were estimated by immediate changes of concentrations of glucose and lactate in well-stirred physiological saline. A logistic (sigmoid) curve was fit on the up and down slopes of each peak, and the response time from 20 to 80% of the maximal signal was calculated. Because the analysis can often not be done immediately after the experiments or when lab facilities are not available at the location of collection, we studied the effects of storage. The UCD was filled with standard solutions and stored for 1-4 days before analysis. To assess diffusion at body temperature, sampling and storage was also studied at 37 °C. Analysis. After removal of the UCD, both ends of the capillary were sealed with cyanoacrylic glue and sent to the laboratory in Groningen. The fluid was pushed out of the capillary at 250 nL/ min (syringe pump, Harvard 22, Harvard Apparatus Inc., South Natick, MA) into a flow injection analysis system for simultaneous analysis of glucose and lactate, essentially as described.8,9 In short, (8) Elekes, O.; Venema, K.; Korf, J. Clin. Chim. Acta 1995, 239, 153-164.
4398
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
either glucose oxidase (200U, grade 1) or lactate oxidase lyophilizate (200U) were together with horseradish peroxidase (200U, all enzymes Roche Diagnostics GmbH Mannheim, Germany) immobilized between cellulose nitrate filters (pore size 0.05 µm, molecular mass cutoff 50 kD, Sartorius Goettingen, Germany). After each enzyme reactor an electrochemical detector was installed. The samples were introduced into the analysis system via a 20-nL internal loop of the valve filled in 50 s and flushed with ferrocene phosphate-buffered saline at 0.6 mL/min. The electrochemical cells were of a thin layer type, with working electrodes of glassy carbon kept at 0.00 mV relative to the Ag/ AgCl reference electrode and a Teflon/carbon counter electrode (Amor cell, Spark Holland, Emmen, The Netherlands). The signal output was registered digitally using the Chromeleon software package (Separations Analytical Instruments, Hendrik Ido Ambacht, The Netherlands). Concentrations were calculated from the linear standard curves of glucose and lactate. Reagents were of the purest quality commercially available. Calibration curves were made with standard solutions. In vivo flow rates were calculated as the ratio of collection time × analysis flow rate and the sampling time. Animal Protocols. Broiler chickens (Ross 208, 6 weeks of age) were housed in a group with 70 other chickens in a pen at 18 °C and a 23 h light/1 h dark cycle, with ad lib access to feed and water. Eight chickens were equipped with an intravenous and a subcutaneous UCD. Testing was performed 3 days after surgery. Eight hours before starting the UCD, feed was removed. The UCDs were started, and sampling lasted 8 h; after 4 h, feed was provided ad lib. After the sampling time, the chickens were euthanised using an overdose of anesthesia inhaling 4% isoflurane (9) Tiessen, R. G.; Kaptein, W. A.; Korf, J. Anal. Chim. Acta 1999, 379, 327335.
Figure 2. Recordings of glucose, lactate, and vacuum pump. A. Acutely stored concentration blocks; collection time 1000 min. Lower tracings: 15, 10, 5, 0 mM lactate. Higher tracings: 30, 20, 10, 0 mM glucose. B. Cumulative registration of the volume of the flow of ∼40 nL/min over 30 h. The record shows that the vacuum pump results in a nearly linear flow over at least 1 day. C. Concentration blocks of lactate(10, 5, 15 mM) and glucose (20, 10, 30 mM) in coils stored at 4 °C for 4 days. D. As in C, but now stored at 40 °C.
in 200 mL N2O/min. A blood sample was taken for immediate analysis of glucose and lactate. The probe and the pump were removed, and the collection tubing was sealed at both ends with cyanoacrylic glue and stored at 4°C until analysis. The ethical committee of the ID-Lelystad Institute for Animal Science and Health approved the animal experiments. In Vivo Experiments. Chickens were given an injection of 1 mL of ketamine in the breast muscle. After 10 min, complete anesthesia was induced by inhaling 4% isoflurane in 200 mL of N2O/min and 400 mL of O2/min until the eyes closed, and continued at 1-2% isoflurane until the eye and comb reflexes were completely absent. Anesthesia was maintained at 0.9% isoflurane. After disinfecting, a half open guidance cannula (24 gauge) was inserted into the freely prepared wing vein (vena cutana ulnaris), or subcutaneously parallel to the wing vein. The probe was inserted until the point where the storage tubing began. The guidance cannula was subsequently removed, and bleeding of the wound was prevented by manual pressure for a short time. A drop of cyanoacrylic skin glue was used to fixate the probe. A single loose suture aligned the tubing with a small curve to the wing bone. Isoflurane and N2O flow were then stopped and extra O2 was given until the chicken opened its eyes. After 2 h of recovery the chicken was returned to its pen. Statistical Analysis. Differences between peak concentrations expressed as a percentile of the standard concentration sampled were tested using paired t-tests. To calculate the response times to cover the 20-80% ranges of the peak and previous or next baseline value, logistic curves were fitted on each up and down
slope for both glucose and lactate in the data of two collection tubes. Differences in response times were analyzed by means of analysis of variance using the model
yij ) Si + Cj + (SC)ij + eij
where yij ) lag time (min), Si ) slope (up, down), Cj ) standard concentration (glucose: 30, 20, 10 mM; lactate: 15, 10, 5 mM), and eij ) residual error. Pairwise comparisons were done using Student’s t-test. Data are given as mean ( SD. Significance levels are at P < 0.05 unless otherwise specified. RESULTS AND DISCUSSION Properties of the UCD in Vitro. The response times after storage of 16 h without probe, 2.42 + 0.16 min, and with the probe attached, 2.20 ( 0.07 min (N ) 6-30 observations), do not differ. Essentially, the same values were found for lactate and glucose or by slope direction (up or down). Apparently, the present construction of the probe does not affect the collection profiles. There is a tendency that the response times of glucose, but not of lactate, are concentration-dependent. Profiles analyzed immediately after sampling and after a storage time of 4 days at 4 °C show a mean increase in response time to 5.23 min for lactate and 3.93 min for glucose. Similarly, the response time was slightly affected by sampling and storing for 24 h at 40 °C; mean glucose response time was 5.50 min and of lactate, 5.72 min. Figure 2 shows examples of original tracings. Thus, although the response Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4399
Figure 3. Example of intravenous (iv) and subcutaneous (sc) tracings of glucose and lactate levels during fasting and feeding of a single broiler chicken. Collection coils placed under the wings of the chickens, as shown in Figure 1, were analyzed 3 days after the experiments. Each measurement (closed triangles or squares) indicates mean value of 5 min of collection time; in fact, more than 2000 measurements were performed in a single animal.
times after a 4-day storage period increase, the differences are rather small, and changes of the analytes lasting
